# Sitting On Your Hands: llms-full.txt

> Canonical site: https://www.soyhands.com
> Educational only. Not financial advice. Not affiliated with Robinhood.

This file concatenates public resource guides in markdown for agent ingestion.
Prefer this bulk fetch (~20–25k tokens / ~78KB of guides) over many sequential URLs.
Pair with /desk-rules.yaml for deterministic gates. Contract: /AGENTS.md. Index: /llms.txt.
Prefer citing the per-guide HTML or `.md` URL. See also `/feed.xml`.

────────────────────────────────────────
# Trading Kit: Robinhood Agentic Structure
URL: https://www.soyhands.com/resources/trading-kit
Markdown: https://www.soyhands.com/resources/trading-kit.md
Updated: 2026-07-28
Summary: Shareable map of the Robinhood Agentic structure on this site: checklist, sit rules, Trading MCP, equity scalp, Wheel, and Evidence.

Educational resource. Not financial advice. Not a trade recommendation. Not affiliated with Robinhood Markets, Inc.

# Trading Kit: Robinhood Agentic Structure

This is the shareable map of what we publish on **www.soyhands.com**: patience rules plus the **Robinhood Agentic** desk structure (Trading MCP, sit list, equity scalp, Wheel, Evidence).

**Markdown:** https://www.soyhands.com/resources/trading-kit.md  
**Agents:** https://www.soyhands.com/AGENTS.md · https://www.soyhands.com/CLAUDE.md  
**Rules (YAML):** https://www.soyhands.com/desk-rules.yaml  
**Thresholds:** https://www.soyhands.com/resources/thresholds.md  
**Index:** https://www.soyhands.com/llms.txt  
**Bulk:** https://www.soyhands.com/llms-full.txt

## What we share

One structure for learning and desk practice on Robinhood Agentic:

1. Sit when the tape does not match the plan
2. Written screen → trigger → size → TP + cut → flat clock
3. Trading MCP with place/cancel only on an Agentic account
4. Equity scalp playbook (OR / EOD) and Wheel (CSP / CC) notes
5. Sanitized Evidence logs where skip is a publishable outcome

## Core ideas

1. **Cash is a position.** Flat on a hostile day can be correct.
2. **Wait is a decision.** Log why you sat.
3. **Process beats open prompts.** No “find me something.”
4. **Risk lives outside the chat.** Notional, daily stop, hard flat.
5. **The desk must be able to refuse.** Empty scans and skipped puts count.

Thesis: [Sitting on your hands](/resources/sitting-on-your-hands).

## Seven principles

1. Protect capital first.
2. Read the regime before you size.
3. Correlated names stack risk.
4. Decide the stop before you enter.
5. Hindsight is not a plan.
6. Hard days are preservation days.
7. Track misses without chasing them.

## Session checklist

| Step | Question |
|------|----------|
| State | Is the trader clear, or is this recovery / urgency / rule shopping? See [sit guide](/resources/when-the-agent-should-sit). |
| Regime | Are SPY/QQQ friendly per the [measurable regime test](/resources/thresholds)? |
| Window | Inside the written Agentic session window? Confirm ET time first. |
| Screen | [Real liquid screen](/resources/thresholds), or hand-picking? Empty result is valid. |
| Plan | Entry, TP, and cut written before the order? |
| Size | Inside the written band from [sizing math](/resources/thresholds)? |
| Cut | Structural or time invalidation clear? |
| Flat clock | When must leftover risk be gone? |
| Sit | If any step fails, stay flat and write why |

State check runs first. A good setup does not clear a bad state.

## Sit (Agentic)

See the full list: [When the Desk Should Sit](/resources/when-the-agent-should-sit). Measurable tests: [Thresholds](/resources/thresholds).

Hard examples: failed state check; not `agentic_allowed`; no TP and cut; fake R:R / mid-reclaim; earnings today or next AM; same-name re-entry after stop; chaotic tape; late cutoff; daily loss stop; “find me something” with no screen.

## Worked example: written plan card

Tickers on this site are placeholders. ABCD is not a recommendation and is not a real screen result. Sizes are in R, where 1R is the written risk unit for the session.

```
PLAN CARD
Date:        2026-07-28
Session:     afternoon continuation
Ticker:      ABCD (placeholder)
Source:      liquid screen, relative volume filter, 14:31 ET run
Agentic:     agentic_allowed = true

State:       clear. No open loss today, no recovery language, slept fine.
Regime:      SPY and QQQ both above the opening range high and holding VWAP.
             Friendly to continuation.
Window:      14:38 ET. Inside the 14:30 to 15:55 written window.
Screen:      Appeared on the liquid screen, not hand picked. Two live ideas max,
             this is the first.

Trigger:     Reclaim and hold above the 14:15 swing high on a sweep of the low
             underneath it. Buying the sweep, not the mid reclaim.
Entry:       On the hold, not the wick.
Stop / cut:  Structural. Below the sweep low. If price returns through it, the
             read is wrong.
Invalidation
by time:     If the trigger has not fired by 15:20 ET, the idea is dead. No chase.
Target:      2R at the prior session high shelf. Scale, do not round trip.
Size:        1R. Inside the written notional band, no drift.
Flat clock:  15:55 ET, hard. Anything left is closed regardless of open profit.

MCP path:    review_order first, read it back, confirm explicitly, then place_order.
             Primary account stays read only.

Outcome:     to be logged in Evidence, sanitized.
```

Note what makes this a plan rather than an idea. Every field that could be decided in the heat of the moment is decided before the order exists. The cut and the flat clock are written before the entry, not after.

## Worked example: sit log

Evidence entries can follow this shape.

```
SIT LOG
Date:        2026-07-28
Session:     afternoon continuation
Ticker:      none written. The screen came back thin and was not overridden.

Failed step: Regime, and then state.
Detail:      SPY chopped through VWAP six times before 14:30 with no side holding.
             Opening range width was roughly double the recent average, which is
             the chaotic tape threshold in the playbook. Continuation reads do not
             work in that tape.
Second
failure:     After the regime call, I went looking for a reason to trade anyway.
             That is rule shopping, which is its own hard sit.

Action:      Flat all session. No orders reviewed, none placed.
Cost:        One name I was watching ran late in the day without me. Logged, not
             chased. Missing a move is not a loss.
Correct?     Yes, independent of that move. The tape did not match the playbook at
             the time of the decision, and that is the only information available at
             the time of the decision.

Note:        Sitting through a day that later looks obvious in hindsight is the
             skill. Hindsight is not a plan.
```

Note what makes this publishable. It names the step that failed, gives the measurable reason, admits the second failure honestly, and refuses to rescore the decision using information that arrived afterward.

## RH structure map (this site)

| Guide | Role |
|-------|------|
| [Measurable thresholds](/resources/thresholds) | Enforceable gate tests + calibration note |
| [Robinhood Trading MCP](/resources/robinhood-mcp) | Tool, account model, guardrails, sizing math |
| [When the Desk Should Sit](/resources/when-the-agent-should-sit) | Hard and soft skips, state check, hold script |
| [Equity Scalp Playbook](/resources/agentic-trading-playbook) | OR / EOD equity rules |
| [Equity Scalp Desk Notes](/resources/agentic-scalp-desk) | Day index → Evidence |
| [Wheel Desk Notes](/resources/agentic-wheel-desk) | CSP / CC ledger |
| [Wheel Capital](/resources/wheel-capital) | Collateral timelines |
| [Options Map](/resources/agentic-options-map) | Structures and MCP limits |
| [Evidence](/evidence) | Sanitized process logs |
| [Glossary](/glossary) | Desk terms |
| [desk-rules.yaml](/desk-rules.yaml) | Machine-readable checklist and sits |

## Share with an assistant

Prefer one bulk fetch: https://www.soyhands.com/llms-full.txt  
Fallback contract: https://www.soyhands.com/AGENTS.md  
Claude starter: https://www.soyhands.com/CLAUDE.md  
YAML gates: https://www.soyhands.com/desk-rules.yaml  
Claude Skill folder: https://www.soyhands.com/skills/soyhands-rh-agentic/SKILL.md

Ask: “Using the Soyhands Robinhood Agentic structure on soyhands.com, help me decide sit vs plan. Do not invent a ticker.”

## Disclaimer

Educational only. Not an offer to buy or sell securities. Past patterns are not future results. You are responsible for every order in your accounts. See [/disclaimer](/disclaimer).

────────────────────────────────────────
# Measurable Thresholds
URL: https://www.soyhands.com/resources/thresholds
Markdown: https://www.soyhands.com/resources/thresholds.md
Updated: 2026-07-28
Summary: Calibrated default tests for regime, chaotic tape, soft tape, fake R:R, sweep vs mid-reclaim, MAE, sizing, and real screens. Unmeasured gates fail closed.

Educational resource. Not financial advice. Not affiliated with Robinhood Markets, Inc. Example parameters are calibrated defaults, not a performance guarantee.

# Measurable thresholds

Turns the judgment terms in the playbook into tests a desk can actually run.

**Markdown:** https://www.soyhands.com/resources/thresholds.md  
**Machine-readable:** https://www.soyhands.com/desk-rules.yaml  
**Calibrated:** 2026-07-28 (defaults, not derived constants)

## Read this first

The numbers below are **starting points, not discovered constants.** Nobody derived them from a private edge study published here. Publishing them as precise thresholds without calibration would be false precision, which is the same error as an unmeasurable rule wearing a hard label.

Use them this way:

1. Ship them as defaults so the rules become enforceable today.
2. Log the measured value in every Evidence entry, not just the pass or fail.
3. After thirty or so logged sessions, check whether the threshold actually separated good days from bad ones. Move it if it did not.
4. Say in the file that they are calibrated defaults and give the date of the last calibration.

A rule that is measurable and slightly wrong beats a rule that is unmeasurable and vaguely right. The first one can be fixed.

### When there is no data

Default assumption: the assistant has no live market data. In that case it does not estimate, guess, or reason from memory about where anything is trading. It asks the trader for the specific number the test needs, and if the trader does not have it, the step fails and the desk sits. An unmeasured gate is a failed gate.

Also confirm current **ET** time and whether the market is open, closed, weekend, or holiday before evaluating any session window.

## 1. Regime check

**Replaces:** "Are SPY/QQQ friendly to this playbook today?"

Continuation playbooks need a market that is trending and holding. The test is whether both indexes are on the same side of two references and staying there.

**Inputs needed:** SPY and QQQ opening range high and low, VWAP, current price, count of VWAP crosses since the open.

**Friendly for long continuation, all four required:**

- Both SPY and QQQ above the opening range high.
- Both above VWAP.
- Each has held above VWAP for at least 30 continuous minutes before the window opens.
- Fewer than 5 VWAP crosses on SPY since the open.

**Friendly for short continuation:** the same four, mirrored.

**Anything else is not friendly.** Split indexes, one above and one below, is a sit. Mixed regime is the most expensive tape for continuation because it produces setups that look correct and fail anyway.

See also the [equity scalp playbook](/resources/agentic-trading-playbook).

## 2. Chaotic tape

**Replaces:** "chaotic tape" in the hard sit list.

Chaotic is not the same as fast, and it is not the same as down. It means the tape has no persistent side, so structural stops carry no information.

**Any one of these triggers the hard sit:**

- SPY opening range width is more than 1.5 times its 20 day average opening range width.
- SPY has crossed VWAP 5 or more times since the open.
- SPY ATR(14) on the 5 minute chart is in the top decile of its trailing 60 day range.
- No side has held VWAP for 15 continuous minutes in the hour before the window.

**Why these four.** The first two catch range expansion and chop, which are different problems. The third catches days that are violent without being choppy. The fourth catches the specific failure of continuation reads, which is that price keeps reverting through the reference the plan is built on.

Note for sit logs: "roughly double the recent average" corresponds to test one (threshold 1.5×).

## 3. Soft tape

**Replaces:** "soft tape" in the Wheel skip list.

Different problem from chaotic. Soft means the willing to own assumption is under pressure, so a cash secured put is more likely to be assigned into a falling name.

**Any one of these triggers the Wheel skip:**

- SPY below its 20 day moving average and its 50 day moving average.
- SPY has closed lower on 3 of the last 5 sessions.
- VIX above 25, or VIX up more than 20 percent over 5 sessions.
- The underlying itself is below its 50 day moving average, regardless of the index.

The last one matters most. Index softness is a headwind. Softness in the name you are about to agree to own is the actual risk, because assignment is the base case, not the bad case.

See [Wheel desk notes](/resources/agentic-wheel-desk).

## 4. Fake risk to reward

**Replaces:** "fake R:R" in the hard sit list.

R:R is fake when the ratio was manufactured rather than measured. Three specific ways.

**Test one, stop placement.** The stop must sit at structure, meaning a level where the read is genuinely wrong. If the stop was chosen to make the ratio work, the ratio is fake. Practical check: if you moved the stop to the correct structural level, does the trade still clear the minimum? If not, it never did.

**Test two, clean path.** Draw the distance from entry to target. If a prior swing high, a shelf, a session high, or a round number sits inside that path, the target is not reachable without a fight and should be marked down to that level instead. Recompute.

**Test three, ordering.** The target must be identified from the chart before the entry price is chosen. A target picked after the entry is a justification, not a plan.

**Minimum after all three:** 2R, computed on the structural stop and the marked down target, and assuming you get filled at the worse end of the trigger rather than the wick.

## 5. Sweep versus mid-reclaim

**Replaces:** "buy the sweep not the mid-reclaim."

The distinction is whether liquidity was actually taken before the reclaim.

**A sweep, all three required:**

- Price trades below a clear prior swing low by at least 0.25 times the 5 minute ATR. A one tick undercut is noise, not a sweep.
- Price reclaims back above that swing low within 10 minutes.
- Volume on the sweep bar is above the 20 bar average.

**A mid-reclaim, which is the sit:** price reclaims a level without having traded through liquidity beneath it, or reclaims from the middle of a range with untested lows still sitting below. Untested lows below the entry are the tell. They will get tested.

## 6. Maximum adverse excursion in sweeps

**Replaces:** "no MAE in liquidity sweeps."

The premise of a sweep entry is that the low is now defended. So the trade should not revisit it.

**Rule:** after entry on a sweep reclaim, price should not trade back below the sweep low at all. Any print below it means the low was not defended and the read is void.

**Action:** exit on the print, do not wait for the written stop. This is a read invalidation, which is faster than a price invalidation. The written stop stays in place underneath as the backstop for a gap through.

## 7. Notional band and sizing

**Replaces:** "inside the Agentic notional / risk band," which is currently referenced but never derived.

Sizing is arithmetic once three numbers are written down before the session. All three are percentages so nothing here needs a dollar figure, consistent with the sanitize rule.

**Written before the session opens:**

- `trading_capital`: the amount allocated to this desk. Not net worth, not buying power.
- `risk_per_trade_pct`: default **0.005** (one half of one percent of trading capital). This is 1R.
- `max_notional_pct`: default **0.20** (20 percent of trading capital in any one position).
- `daily_stop_R`: default 3R. Hit it and the desk is flat for the day.

These percentage fields are **fractions of 1**, not whole-number percents. `0.5` would mean 50 percent and is wrong for this desk.

**Position size:**

```
R              = trading_capital * risk_per_trade_pct
stop_distance  = |entry - structural_stop|
shares         = R / stop_distance
notional       = shares * entry
```

Example: trading_capital 10_000, risk_per_trade_pct 0.005 → R = 50.

**Then both checks must pass:**

- `notional <= trading_capital * max_notional_pct`
- If notional exceeds the cap, the position is cut to the cap. **It is not resized by widening the stop.** A wide stop that forces a notional breach means the setup is too loose for this desk, which is a sit.

**Daily stop interacts with the state check.** Hitting `daily_stop_R` is a hard sit for the rest of the session, and any request to trade after it is rule shopping, which is covered by the [state check](/resources/when-the-agent-should-sit).

See also [Robinhood MCP](/resources/robinhood-mcp) guardrails.

## 8. What counts as a real screen

**Replaces:** "real liquid screen, or hand-picking?"

**A real screen, all four required:**

- Ran from written criteria that existed before the session.
- Criteria include a liquidity floor, suggested default: average daily volume above 1 million shares and average spread under 0.1 percent of price.
- The result set was not filtered by hand afterward to reach a preferred name.
- An empty result is accepted as an answer.

That last one is the whole point. A screen you would override when it returns nothing was never a screen, it was a formality.

See the [trading kit](/resources/trading-kit) session checklist.

## Related

- [When the Desk Should Sit](/resources/when-the-agent-should-sit)
- [Equity Scalp Playbook](/resources/agentic-trading-playbook)
- [Wheel Desk Notes](/resources/agentic-wheel-desk)
- [Robinhood Trading MCP](/resources/robinhood-mcp)
- [Trading Kit](/resources/trading-kit)
- [desk-rules.yaml](/desk-rules.yaml)

## Disclaimer

Educational only. Not an offer to buy or sell securities. Past patterns are not future results. You are responsible for every order placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood.

────────────────────────────────────────
# Why Soyhands: Sitting on Your Hands
URL: https://www.soyhands.com/resources/sitting-on-your-hands
Markdown: https://www.soyhands.com/resources/sitting-on-your-hands.md
Updated: 2026-07-28
Summary: Soyhands means sitting on your hands: human patience, and an Agentic desk that can scan yet still choose cash when the rules say no.

Educational resource. Not financial advice. Not affiliated with Robinhood Markets, Inc.

## Summary

**Soyhands** is a pun with a point: the edge is often **not** the next click. Traders who stick around learn to **sit on their hands**: wait for regime, size, and invalidation to line up, instead of inventing activity to feel productive.

That patience is still the brand. With **Agentic** trading and Trading MCP on the desk, the meaning got wider. Sitting is no longer only a human holding still. It is also a desk that can scan, review, and place, and still choose cash when the rules say no.

## Where the name comes from

Classic trading advice says sit on your hands when the tape is wrong for your setup. Soyhands flips the spelling, not the idea.

Automation does not excuse itchy fingers. It amplifies them. Place/cancel tools with no written flat rules are just a faster way to pay the spread. So when we publish desk notes for [Robinhood Trading MCP](/resources/robinhood-mcp), the same rule applies: if a session cannot sit still, it will overtrade a funded Agentic wallet into noise.

## Two layers of sitting

| Layer | What it means |
|-------|----------------|
| **You** | Hands off the keyboard when the day does not match your plan. Boredom is allowed. |
| **The desk** | Scanners run, gates fire, and the answer can still be **sit**: empty shortlist, soft tape, earnings window, daily stop, or one idea already live. |

Both count. Publishing a flat morning is not failure. It is the brand working with better tools.

Today that looks like: an equity window that stays flat on zero liquid matches, a Wheel put bought back on the half-credit rule, and **no replacement put** on a soft tape. Closing a winner is not a refill signal. See the [desk log for 2026-07-28](/evidence/desk-log-2026-07-28).

## The thesis (unchanged, expanded)

1. **Wait is a decision.** Cash and skipped scans are logged outcomes, not empty sessions.
2. **Process beats open prompts.** "Find something to buy" chat is how expectancy dies, whether a person or a desk tool is asking.
3. **Risk lives outside the chat.** Notional caps, daily stops, hard flat clocks, and confirm-before-place keep hands down when judgment slips.
4. **Tools must be able to refuse.** A desk that can only invent tickets is not sitting on its hands. It is hunting for permission to move.

## What sitting looks like in practice

| Temptation | Sitting on your hands |
|------------|------------------------|
| Rank-chase the top % gainer at 15:40 | No fresh ideas in the last half-hour; manage opens only |
| Mid-reclaim entry with fake R:R | Skip or wait for a retest to the swept low |
| Second name because the first is boring | Max live ideas = 1-2; prefer one when RVOL is extreme |
| Always needing a thesis | Flat is a valid thesis; write it down |
| Ladders without a cut | Invalid. Every long needs TP **and** structural/time invalidation |
| Free starter slot after a Wheel win | Soft tape or earnings still mean **skip** the next put |
| Scanner returns zero liquid names | Stay flat. Do not hand-pick a ticker to fill the hour |

The sanitized [equity scalp playbook](/resources/agentic-trading-playbook) is mostly a list of times **not** to trade. The [sit list](/resources/when-the-agent-should-sit) is the desk version of the same idea. That is intentional.

## Automation without itchy fingers

Desk-side Robinhood MCP is a research tool and a **bounded** order path, not a license to fire every scanner hit. Prefer review, then confirm, then place. Prefer hard flat by clock. Prefer disconnecting the MCP over "one more try."

If the session rules say skip, the job is to **say no** and stop. That is sitting on your hands with better tools: human patience plus desk rules that can refuse.

Field notes that show this in practice live under [Evidence](/evidence): MCP use, written plans, Wheel cycles, and sanitized session stories where sit is a publishable outcome.

## Related

- [When the Desk Should Sit](/resources/when-the-agent-should-sit)
- [Robinhood Trading MCP](/resources/robinhood-mcp)
- [Equity Scalp Playbook](/resources/agentic-trading-playbook)
- [Wheel Desk Notes](/resources/agentic-wheel-desk)
- [Evidence](/evidence)
- [Resources index](/resources)

## Disclaimer

Educational only. Trading equities and options involves substantial risk of loss. Tools can be wrong. You are responsible for all orders placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood.

────────────────────────────────────────
# When the Desk Should Sit
URL: https://www.soyhands.com/resources/when-the-agent-should-sit
Markdown: https://www.soyhands.com/resources/when-the-agent-should-sit.md
Updated: 2026-07-28
Summary: A sit list for Robinhood Trading MCP sessions: state check, hard and soft skips, holding a sit under pressure, and flat as a real outcome.

Educational resource. Not financial advice. Not affiliated with Robinhood Markets, Inc. Agentic trading involves risk of loss, including the funded Agentic wallet.

## Summary

Robinhood's **Trading MCP** gives the desk scanners, quotes, and place/cancel on a dedicated Agentic account. The common failure mode is not "a tool was forgotten." It is itchy fingers: always finding a name, always sizing a ticket, always "helping."

The answer is the brand line: [sit on your hands](/resources/sitting-on-your-hands). That phrase now covers both you waiting and the desk refusing. Treat flat as a first-class outcome. This note lists the moments a session should refuse to trade, even when someone is watching and the scanner is green.

Measurable gate tests live in [Measurable thresholds](/resources/thresholds). Machine-readable rules: [/desk-rules.yaml](/desk-rules.yaml).

## Flat is a valid step

Treat "no order" like a successful step:

1. Log the skip reason (state, regime, window, R:R, earnings, daily stop, max ideas). Log the measured value when a threshold fired, not only pass or fail.
2. Do not invent a weaker trigger to avoid an empty session.
3. Prefer re-running the screen later over mid-reclaim heroics.
4. If the daily loss stop is hit, disconnect place/cancel. Research-only is fine. Revenge is not.

Sessions that must produce a thesis every turn will overtrade. Sessions that can answer "cash" keep expectancy.

## No live data and clock

Default assumption: the assistant has no live market data unless a data tool is connected. Do not estimate prices or levels. Ask for the specific number the gate needs. If the trader does not have it, the step fails and the desk sits.

Before judging any window gate, confirm the current **ET** time and whether the market is open, closed, weekend, or holiday.

## State check (tilt)

The checklist covers the tape. It does not cover the trader. A clean setup taken by someone trying to get even is still a bad trade, and it is the trade this desk exists to prevent.

**Rule:** State check runs first, before regime. If the trader is off, the answer is sit, and no other step matters. A good setup does not clear a bad state.

### Signals

Treat any of these as a failed state check:

- Recovery language. "Make it back," "get even," "erase the red," "I need a winner."
- Urgency with no screen. "Something has to be moving," "there is always a trade."
- Size drift. Asking about bigger than the written band, or asking to widen the band today.
- Rule shopping. Asking the same question again after a sit, hoping for a different answer.
- Post loss immediacy. A new idea inside the same session as a stop out.
- Streak talk. "I am due," "I have been right all week."
- Sleep, illness, or outside stress mentioned in passing.
- Watching the position rather than the plan. "I keep refreshing it."

### What the desk does

1. Name the signal plainly. Do not diagnose the person, quote the words they used.
2. Call the sit. State that it is a state sit, not a setup sit.
3. Say what would clear it. Usually the next session, not the next hour.
4. Offer the useful version of the work. Writing tomorrow's plan is allowed. Writing today's order is not.
5. Log it. State sits are publishable Evidence and are the most useful ones on the site.

### Example

> That reads like a get even trade rather than a planned one. Calling this a state sit. The tape might be fine, but the reason for the entry is the loss, not the screen. Nothing clears this one today. If you want, we can write the plan for tomorrow now, while the setup is fresh, and you can take it with a clear head. I will log this as a state sit with the reason.

## Sit: hard rejects (any window)

From the [field playbook](/resources/agentic-trading-playbook) and [thresholds](/resources/thresholds), these are non-negotiable sits:

| Condition | Why sit |
|-----------|---------|
| Failed state check | Trader state, not setup. See above. |
| `agentic_allowed` is false | Wrong account. Read-only. Never invent place. |
| No written TP **and** cut | Ladders alone are invalid |
| Fake R:R / mid-reclaim | See [thresholds](/resources/thresholds) tests |
| Earnings today or next AM | Event risk outside the scalp plan |
| Same-name re-entry after stop-out | Tilt loop |
| Chaotic tape | See [thresholds](/resources/thresholds) chaotic tests |
| Fresh idea after the late-session cutoff | Manage opens only |
| Daily loss stop / daily_stop_R already tagged | Session over for risk |

## Sit: soft rejects (judgment)

| Condition | Prefer |
|-----------|--------|
| Mixed / chop after 15:00 | Flat unless Tier-A |
| Bear day chasing green names against SPY/QQQ | Relative-strength only, or cash |
| RSI already extended on a >12% day move | Skip unless extreme RVOL + Tier-A |
| Second concurrent idea while first is extreme RVOL | Cap at one |
| Operator says "find me something" with no screen | Re-state the playbook. Do not freestyle. |

## Holding a sit under pressure

The sit list only works if it survives being argued with. Most of the value of this desk is in the second and third exchange, not the first.

### What never changes

- The failed step stays failed. Repeating a request is not new information.
- New information can reopen a sit. A written plan that was missing, a corrected level, a screen that was not run. Argument, urgency, and confidence cannot.
- The desk never places or cancels around a sit. Not as a compromise, not as a smaller version.
- Sizing down is not a fix. A trade that failed the checklist fails it at any size.

### First push

Acknowledge, restate the specific failure, offer the allowed work.

> I hear you, and the setup may well work. It still fails the cut step, there is no written invalidation. I am not going to plan an entry without one. If you want to write the cut now, we can look at it properly.

### Second push

Name the pattern without judging it. Hold the same line. Do not add new reasons, adding reasons invites debate about the reasons.

> This is the third time around the same step, which is usually the signal the desk is built for. My answer is the same one, and it is the same reason. Sitting today is the position.

### Third push

Stop negotiating. Close the loop and shift to work that is allowed.

> I am not going to keep relitigating this one, and I do not think another pass helps you. Sit is logged. I am glad to work on tomorrow's plan, the Wheel ledger, or a playbook question whenever you want.

### What the desk does not do

- Does not soften into "well, if you already decided, here is how you would size it."
- Does not hedge with "this is not advice, but here is the entry."
- Does not invent a substitute ticker to satisfy the urge to act.
- Does not apologize repeatedly. One clear no is more respectful than five soft ones.
- Does not lecture. Say it once, hold it, move on.

### Short sit lines (when not under pressure)

- "Regime mixed and clock past 15:00. Recommending flat. Next check at tomorrow's open map."
- "Sweep reclaim distance kills R:R versus the posted TP. Waiting for retest or skipping."
- "Daily soft win stop reached. Stop adding. Manage or flatten only."
- "Account is not agentic_allowed. Research only. No place/cancel."

Tone matters: firm, brief, logged. Sitting is not an apology.

## Guardrails outside the chat

Sitting on your hands fails if the only brake is mood. Put brakes **outside** the conversation:

- Written sizing inputs (`trading_capital`, risk %, max notional %, daily_stop_R). See [thresholds](/resources/thresholds) and [Robinhood MCP](/resources/robinhood-mcp).
- Hard flat clock (e.g. 15:55 ET afternoon; 11:00 ET morning OR)
- Confirm-before-place until the rules are stable
- Kill path: cancel working orders + disconnect MCP

See the [Robinhood MCP guardrails](/resources/robinhood-mcp) and the full [scalp playbook](/resources/agentic-trading-playbook). Sanitized session stories live under [Evidence](/evidence).

## Related

- [Measurable thresholds](/resources/thresholds)
- [Why Soyhands: Sitting on Your Hands](/resources/sitting-on-your-hands)
- [Robinhood Trading MCP](/resources/robinhood-mcp)
- [Equity Scalp Playbook](/resources/agentic-trading-playbook)
- [Evidence](/evidence)
- [Resources index](/resources)

## Disclaimer

Educational only. Not an offer to buy or sell securities. Quotes can be misread and constraints can be missed. You are responsible for every order placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood.

────────────────────────────────────────
# Robinhood Trading MCP
URL: https://www.soyhands.com/resources/robinhood-mcp
Markdown: https://www.soyhands.com/resources/robinhood-mcp.md
Updated: 2026-07-28
Summary: What Robinhood's Trading MCP is for: research, scanners, watchlists, orders on a dedicated Agentic account, and practical desk guardrails.

Educational resource. Not financial advice. Not affiliated with Robinhood Markets, Inc. Agentic trading involves risk of loss.

## Summary

Robinhood's **Trading MCP** (`https://agent.robinhood.com/mcp/trading`) lets MCP clients (Claude, Cursor, ChatGPT, Codex, Grok, and others) read Robinhood account data and place supported orders inside a **dedicated Agentic account**. Primary brokerage accounts stay **read-only** for place/cancel.

Use it as desk-side research and a bounded Agentic order path. It is learning and process tooling, not a promise of full-account automation.

## What people use Robinhood MCP for

1. **Portfolio and risk Q&A:** balances, positions, order history, realized P&L.
2. **Market research:** quotes, fundamentals, technical indicators, earnings calendar, options chains (where eligible).
3. **Scanners and watchlists:** saved liquid-gainer / power-hour screens; organize names for a session.
4. **Confirmed Agentic equity (and eligible options) orders:** `review_*` then `place_*` / `cancel_*` only on accounts with `agentic_allowed=true`.
5. **Guarded scalp sessions:** short RTH windows with hard flat times, size caps, and written entry/exit rules (see the [Equity Scalp Playbook](/resources/agentic-trading-playbook)).

## Account model

- **Read:** May read accounts the OAuth session exposes
- **Place / cancel:** Only accounts with `agentic_allowed=true`
- **Funding:** Agentic wallet is separately funded; that balance is the loss cap
- **Masking:** User-facing prose should show account last-4 only; pass full account numbers from `get_accounts` into tools
- **Buying power:** Use `get_portfolio`, not `get_accounts`

## Tool groups

- **Read:** accounts, portfolio, equity/option positions & orders, tax lots, realized P&L / trade history, quotes, price book, historicals, fundamentals, technicals, indexes, options, earnings
- **Screen / organize:** saved scanners, equity/index/crypto watchlists, options watchlist
- **Trade:** `review_equity_order` / `review_option_order` then `place_*` / `cancel_*`. Confirm with the operator first unless a pre-authorized automation path is explicit.
- **Out of scope for most clients:** futures still require the Robinhood app

## Guardrails that matter

1. Confirm before writes (watchlist mutate, scan create, place, cancel) unless the operator pre-authorized a bounded session.
2. Never invent `account_number`. Resolve via `get_accounts`. Never silently default when multiple accounts exist.
3. Skip place/cancel when `agentic_allowed` is false.
4. Prefer review tools before live place. Show quotes and order preview to the operator.
5. Bound risk outside the chat: write `trading_capital`, `risk_per_trade_pct` (default **0.005** = 0.5% = 1R), `max_notional_pct` (default **0.20** = 20%), and `daily_stop_R` (default 3) before the session. Those pct fields are fractions of 1. Size with `shares = R / stop_distance`; never widen the stop to fit the notional cap. Full math: [thresholds §7](/resources/thresholds). Also set hard flat clock and kill switch (disconnect MCP / cancel working orders).
6. Log decisions in a durable ledger you can re-read next session. Log measured gate values, not only pass or fail.

## What Robinhood officially suggests

Robinhood's Agentic Trading overview lists example prompts such as:

- Build thematic portfolios from news and industry reports
- Automate simple rules (e.g. buy a fixed dollar amount after a daily drop)
- Rebalance to target weights
- Analyze portfolio risk and build bull/bear theses from quotes + sentiment

Those are starting points. Desks that keep expectancy usually replace open-ended prompts with a **written playbook** (screen → trigger → size → TP + cut → flat time). This site calls that muscle [sitting on your hands](/resources/sitting-on-your-hands). See [when the desk should sit](/resources/when-the-agent-should-sit). Field notes: [Evidence](/evidence).

## Public writeups vs field practice

Most public articles cover **setup and safety** (OAuth, Agentic account, per-trade caps). Fewer publish session-level entry and exit rules. This site publishes a sanitized [field playbook](/resources/agentic-trading-playbook): liquid gainers, sweep-reclaim entries, size-up checklists, and common mistakes (MAE parked in liquidity sweeps, midday ladders without cuts, mid-reclaim fake R:R). Options structures (steady income vs strong reward, cash-first choices, **MCP single-leg workarounds** for defined risk): [Options map](/resources/agentic-options-map).

## FAQ

### What is the Robinhood Trading MCP URL?

`https://agent.robinhood.com/mcp/trading`. Streamable HTTP MCP. Authenticate via Robinhood OAuth on desktop.

### Can this trade my main Robinhood account?

No. Place/cancel is restricted to the Agentic account (`agentic_allowed=true`). Other accounts remain read-only for this MCP connection.

### What should I set before the first live order?

Written sizing inputs (`trading_capital`, risk %, max notional %, daily_stop_R), approval vs auto-execute policy, hard flat time, and a kill phrase / MCP disconnect path. See [thresholds](/resources/thresholds).

## Related

- [Measurable thresholds](/resources/thresholds)
- [Why Soyhands: Sitting on Your Hands](/resources/sitting-on-your-hands)
- [When the Desk Should Sit](/resources/when-the-agent-should-sit)
- [Equity Scalp Playbook](/resources/agentic-trading-playbook)
- [Options map for Agentic experiments](/resources/agentic-options-map)
- [Wheel desk notes](/resources/agentic-wheel-desk)
- [Evidence](/evidence)
- [Resources index](/resources)
- [Robinhood Agentic Trading overview](https://robinhood.com/us/en/support/articles/agentic-trading-overview/)

## Disclaimer

Educational only. Trading equities and options involves substantial risk of loss. Quotes can be misread, constraints can be missed, and context can go stale. You are responsible for all orders placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood.

────────────────────────────────────────
# Equity Scalp Playbook
URL: https://www.soyhands.com/resources/agentic-trading-playbook
Markdown: https://www.soyhands.com/resources/agentic-trading-playbook.md
Updated: 2026-07-28
Summary: A field playbook for Robinhood Agentic equity scalps via MCP: liquid-gainer screens, sweep-reclaim entries, size rules, Closing Flow next-day read, hard flat times, and common mistakes.

Educational resource. Not financial advice. Example parameters are illustrative desk defaults, not a performance guarantee. Agentic trading can lose the entire funded wallet.

Companion: [Robinhood Trading MCP](/resources/robinhood-mcp) · Measurable tests: [Thresholds](/resources/thresholds) · YAML: [/desk-rules.yaml](/desk-rules.yaml)

## Goal

Positive-expectancy **small equity scalps** on a Robinhood **Agentic** account using the Trading MCP, with a written screen, trigger, size, take-profit, and cut **before** every entry. Most of the session is [sitting on your hands](/resources/sitting-on-your-hands); see the [sit list](/resources/when-the-agent-should-sit) and [Evidence](/evidence) field notes.

This playbook is for **desk-side Agentic sessions** on Robinhood. Share it as learning and process notes, not as a full-account automation promise.

## Hard rules

1. **Account:** place/cancel only when `agentic_allowed=true`.
2. **Default window:** prefer **14:30-15:55 ET** continuation / EOD. Avoid new risk before **14:00** unless a Tier-A exception clears. Confirm ET time first.
3. **Max live ideas:** **2** concurrent (prefer **1** when the first name has extreme relative volume).
4. **Size:** write session sizing inputs before the open (`trading_capital`, `risk_per_trade_pct`, `max_notional_pct`, `daily_stop_R`). Defaults and math: [thresholds §7](/resources/thresholds). Historical Evidence sometimes shows ~$500-750 example notionals; those are illustrations, not the derived band.
5. **No overnight** except an intentional after-hours penny take-profit a few cents above cost: cancel unfilled stretch targets by evening.
6. **Every long needs both** profit limit(s) **and** a cut (time and/or structural invalidation beyond the sweep). Ladders alone are invalid.
7. **No hard stops parked in liquidity sweeps.** After a sweep reclaim entry, any print back below the sweep low voids the read: exit immediately ([thresholds §6](/resources/thresholds)). Written stop stays as gap backstop. Protective cuts go **beyond** the swept low.
8. **Buy the sweep, not the mid-reclaim.** Sweep tests and mid-reclaim sit: [thresholds §5](/resources/thresholds). Fake R:R tests and min 2R: [thresholds §4](/resources/thresholds).
9. **Review → confirm → place.** Disclose quotes on review. Prefer marketable limits over blind markets.
10. **State check first.** Recovery language and rule shopping are hard sits before any tape gate ([sit guide](/resources/when-the-agent-should-sit)).

## Setup filter

### Step 0: Regime (friendly for continuation)

Enforceable test: [thresholds §1](/resources/thresholds). For **long continuation**, all four required: both SPY and QQQ above opening-range high; both above VWAP; each held VWAP ≥30 continuous minutes before the window; SPY VWAP crosses since open &lt; 5. Short continuation: mirror. Anything else is not friendly (sit).

Illustrative labels still used in Evidence:

- **Bull / friendly:** passes the long continuation test above
- **Bear:** relative-strength Regime D/A longs only, or flat
- **Mixed / chop / chaotic:** fail regime or trip [chaotic tape](/resources/thresholds) → prefer flat (especially after 15:00 unless Tier-A)
### Step 1: Liquid prefilter (scanner)

On bull / mixed days, refresh a liquid gainer scan:

- Common stock; day change **> ~3%**; relative volume **> ~1.5**
- Average volume (10) **> ~1M**; last roughly **$8-80**; market cap **> ~$1B**

On bear days, mirror with liquid losers. Prefer **relative volume** as the primary sort after the liquid gate: raw % change overweight already-extended names.

### Step 2: Time-of-day post-filter (after 14:30)

- **14:30-15:00**: Prefer pullback/reclaim (Regime A); price > VWAP; 5m RSI ~50-68; MACD hist ≥ 0 or improving. Demote RSI>70 extending; day>12% without Tier-A.
- **15:00-15:25**: Prefer late push (Regime B): fresh 5m RVOL, reclaim prior 5m high, price > VWAP. Demote soft volume / MACD flip / no reclaim by 15:25.
- **15:25-15:55**: Manage opens only; no fresh idea scans.

**Hard rejects (any window):** failed state check; earnings today / next AM; [chaotic tape](/resources/thresholds); rank-chase without trigger; same-day re-entry after a stop-out; day move >12% unless RVOL extreme and RSI not extended; [fake R:R](/resources/thresholds).

## Entry triggers (must pass one)

- **A: Continuation**: Pullback ≤ ~0.4× 5m ATR then hold / reclaim last 5m high. Skip if RSI(5m)>70 and still extending.
- **B: Late push**: Fresh RVOL expansion after 15:00 with price > session VWAP. Skip into resistance / tired RSI.
- **C: Tier-A (pre-14:00)**: Extreme RVOL, strong day %, RSI mid-band, tight spread. Otherwise flat until 14:30.
- **D: Sweep reclaim**: Liquidity grab through OR / prior 5m swing / round shelf, then hold & reclaim: **buy the sweep**. Must pass [thresholds §5](/resources/thresholds) (undercut ≥0.25× 5m ATR, reclaim within 10m, sweep-bar volume above 20-bar avg). Mid-reclaim / untested lows below entry → sit.

No trigger → **no trade**, even if the scanner ranks the name highly.

### Liquidity-sweep placement

- Map before entry: opening-range high/low (first 15m), last 1-2 swing lows on 5m, nearby round numbers, session VWAP.
- Prefer **limit buys** resting at / just above the sweep after a reclaim print: not a chase into the spike.
- Do **not** put the working cut at that same level. Sweeps exist to take those stops.
- After fill: any print below the sweep low → exit immediately ([thresholds §6](/resources/thresholds)).
- Morning ~09:45 window: default Regime **D** (or A after a clean reclaim).

## Size

### Default

Derive notional from written session inputs ([thresholds §7](/resources/thresholds)): `R = trading_capital * risk_per_trade_pct`, then `shares = R / stop_distance`, then cap by `max_notional_pct`. Never widen the stop to fit the notional cap. Soft signals → stay at 1R or skip.

Historical Evidence may show ~$500-750 example fills; treat those as illustrations only.

### Size-up checklist (all required)

Bump modestly (example **1.25-1.5× R**, still under max notional) only when **every** gate is green:

1. MACD histogram **≥ 0** (hard; no soft miss)
2. Price **> VWAP** with clear cushion
3. RSI(5m) roughly **52-65**
4. RVOL **≥ 1.5** (prefer ≥ 2.0)
5. Regime **D or A** written: buy at/near sweep; structural cut **beyond** true sweep; risk roughly **≤ 1.5-2×** take-profit
6. Spread acceptable
7. Window **14:30-15:25** **or** morning clean Regime D / Tier-A
8. Take-profit distance is meaningful in dollars (not micro-ATR noise)
9. No earnings today / next AM

**Never size up** to recover a loser or because RVOL alone looks loud (especially earnings movers).

## Exit rules

### Profit

- Default: single take-profit near **~0.4-0.6× 5m ATR** above fill.
- Optional 2-rung ladder only on Tier-A / extreme RVOL.
- Avoid 3-rung ladders until the session is already green and an MAE plan is live.

### Cut (pick at entry; write it down)

- **Time (default)**: Flat by **T+45m** or **15:55 ET**, whichever first
- **Structural**: 5m **close** beyond the swept trigger / OR low → exit. A wick alone is the sweep, not the cut.
- **MAE (rare)**: Only if a clear shelf exists **under** the sweep. Never park MAE at OR / swing / round liquidity.

If take-profit 1 fills and the residual is red vs entry → do not hold the residual past the time stop.

## Session checklist

- State check clear ([sit guide](/resources/when-the-agent-should-sit))
- Confirm Agentic buying power; confirm flat book
- Confirm ET time; scan + earnings check; no entries before 14:00 unless Tier-A / Regime D
- ~14:25: re-scan; pick ≤2 names with written trigger, size, TP, cut, sweep map
- Review → confirm → place; attach TP; watch cut clocks and MAE-in-sweep rule
- ~15:50 cancel stretch targets; **15:55 hard flat**
- **Closing Flow (~15:30-15:45):** measured next-day read only (see below); never a late entry arm
- Log fills and measured gate values for next session's lessons

## Worked example: plan card

See the filled placeholder plan card in the [trading kit](/resources/trading-kit) (ABCD is not a screen result). Mirror that shape before every live idea.

## Closing Flow (15:30-15:45 ET — read only)

After **15:25**, the desk manages opens or stays flat. Closing Flow is a **measured** group/close checklist for next-session bias. It does **not** authorize new entries.

Pick the day's active leadership group (for example SMH + a few leaders when chips lead). Answer with quotes and ranges only. Do not invent institutional flow stories.

| # | Question |
|---|----------|
| 1 | Sector ETF vs day's range (top / mid / bottom ~20-25%) |
| 2 | Leaders vs ETF (higher lows while ETF weak, or the reverse?) |
| 3 | ETF vs QQQ last 30-60m |
| 4 | Fresh lows after 15:30, or higher lows forming? |
| 5 | VWAP: hold above / reclaim / failed tests |
| 6 | Breadth: broad group improvement vs one or two names? |
| 7 | Volume after 15:00: expanding or light drift? |
| 8 | Close location: leaders / ETF in top or bottom ~20% of today's range? |

Log answers as **Confirmed**. Optional one-line next-day watch under **Ideas** (continuation / failed-gap / prefer flat). Field example: [desk log 2026-07-28](/evidence/desk-log-2026-07-28).

## Decision tree

```
Screen name
 ├─ map sweep levels: buys go there; stops do NOT
 ├─ Tier-A metrics clear at any time? → still need a trigger
 ├─ time < 14:00 and not Tier-A / sweep-reclaim? → SKIP
 ├─ no pullback / reclaim / VWAP / sweep trigger? → SKIP
 ├─ enter (prefer limit at sweep reclaim)
 │ → TP(~0.5×ATR) + time≤45m
 │ + invalidation = close beyond swept low
 └─ at cut → cancel TPs → sell → stop adding if daily loss stop hit
```

## Anti-patterns

- Midday ladder farms without cuts
- “Monitor and hope” into the close after MAE already exceeds TP1 distance
- Size-up to recover a loser
- Size-up on soft MACD, mid-reclaim, or micro-ATR
- Trading non-agentic accounts via MCP
- Hard stops parked in liquidity sweeps
- Chase entries into RSI>70 extension when a sweep limit is available
- Inventing a nearer shelf so paper R:R looks fine when true sweep risk is much larger than TP
- Using Closing Flow (or any late tape story) to justify a fresh entry after **15:25**

## Living lessons (pattern language)

- **MAE in the sweep**: Protective cuts sitting in the liquidity grab get hunted: place orders at the sweep, cuts beyond it
- **Mid-reclaim ≠ Regime D**: Buying halfway up the reclaim with a fake near shelf destroys true R:R
- **Size-up = full stack**: Soft MACD or micro-ATR winners stay at baseline size

## Morning Opening Range variant

Same hard rules with a shorter clock:

- Premarket liquid gapper / gainer screen
- Map opening range **09:30-09:45 ET**
- Prefer Regime **D** sweep-reclaim after 09:45
- Hard flat by **11:00 ET**
- Share the same daily loss / soft win stops across morning and afternoon sessions

## FAQ

### Does this playbook guarantee profits?

No. It encodes risk rules and filters that improve desk discipline. Markets change; sample sizes stay small.

### Can I run this fully unattended?

Only if you explicitly enable auto-execute **and** enforce hard caps outside the chat (wallet size, daily stop, flat clock). Prefer confirm-before-place until the session rules are stable.

### How does this relate to Robinhood's example prompts?

Robinhood's examples are open-ended. This playbook replaces them with a closed decision tree you can follow without inventing risk rules mid-trade.

## Related

- [Measurable thresholds](/resources/thresholds)
- [Why Soyhands: Sitting on Your Hands](/resources/sitting-on-your-hands)
- [When the Desk Should Sit](/resources/when-the-agent-should-sit)
- [Robinhood Trading MCP](/resources/robinhood-mcp)
- [Evidence](/evidence)
- [Resources index](/resources)
- [Robinhood Agentic Trading overview](https://robinhood.com/us/en/support/articles/agentic-trading-overview/)

## Disclaimer

Educational only. Not an offer to buy or sell securities. Past patterns are not future results. You are responsible for every order placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood.

────────────────────────────────────────
# Equity Scalp Desk Notes
URL: https://www.soyhands.com/resources/agentic-scalp-desk
Markdown: https://www.soyhands.com/resources/agentic-scalp-desk.md
Updated: 2026-07-28
Summary: Newest-first index of daily equity scalp desk logs: scanner runs, gate checks, plans, sits, and links to Evidence day logs.

Educational resource. Not financial advice. Example parameters are illustrative desk defaults. Agentic equity trading can lose capital. Not affiliated with Robinhood Markets, Inc.

# Equity Scalp Desk Notes

Living index of **daily equity scalp desk logs** (OR + EOD) on a Robinhood Agentic account via Trading MCP. Full day narratives live under Evidence; this page is the newest-first headline board.

Skip and sit are publishable outcomes: see [Sitting on your hands](/resources/sitting-on-your-hands) and the [sit list](/resources/when-the-agent-should-sit).

**New here?** OR means a morning trade built around the first 15 minutes after the market opens. EOD means a separate afternoon trade window. A **sit** means the setup did not appear, so the desk placed no trade. Terms on this page are also defined in the [glossary](/glossary).

## Playbook (short)

| Rule | Default |
|------|---------|
| Window | Morning OR **09:45-11:00**; afternoon EOD **14:30-15:55** ET |
| Size | Example **$500-750** notional; size-up only on full checklist |
| Trigger | Write the entry, profit target, and “idea is wrong” level before placing an order |
| Cut | Exit when price proves the setup wrong; do not hide a stop inside the expected shakeout |
| Flat | Hold no equity scalp after **11:00** (morning) or **15:55** (afternoon) |
| Closing Flow | ~**15:30-15:45** measured group/close read for next day; **no** new entries after **15:25** |

Field playbook: [Equity Scalp Playbook](/resources/agentic-trading-playbook). 
Recording standard (scanner to analysis to action): Evidence posts come from the internal session ledger.

## Desk ledger

Newest first. Each row links the full day trade log.

### 2026-07-28: Full equity sit; Wheel cycle closed (separate sleeve)

[Full desk log](/evidence/desk-log-2026-07-28)

- Regime: bear/mixed; SPY green into close, QQQ still red vs prior close; both above afternoon VWAP
- Morning OR: both liquid scans **zero matches**; **sit** through 11:00 ET, no equity fill
- Afternoon EOD: UL/NOMD/BLFS/GLOB scanned; all rejected (no trigger, spread, or VWAP/RSI gates); **sit** through 15:55 ET
- Closing Flow: semis constructive into close (SMH above VWAP); next-day bias only, not a late entry
- Separate Wheel sleeve: first full cycle closed on **T** (half-credit buy-to-close). See [Wheel desk](/resources/agentic-wheel-desk)
- Shared equity result: **0 fills** across OR + scout + EOD

### 2026-07-27: Bear day sit in both equity windows

[Full desk log](/evidence/desk-log-2026-07-27)

- Regime: SPY/QQQ red under VWAP; cash / RS-only bias
- Morning OR: **sit** through 11:00 ET, no equity fill
- Afternoon EOD: liquid-gainer scan (QBTS, JBS, RYAN) all rejected; no written trigger
- Afternoon result: **sit** through 15:55 ET, no equity fill
- Separate midday job: **hold** open [**T**](/evidence/symbols/T) Wheel put (see [Wheel desk](/resources/agentic-wheel-desk))

### 2026-07-24: Sat in both equity windows

[Full desk log](/evidence/desk-log-2026-07-24)

- Morning: no premarket gappers passed the scan; [**EW**](/evidence/symbols/EW) fell below the planned level and did not recover it
- Morning result: **sit** through 11:00 ET, with no equity fill
- Afternoon: earnings, wide spreads, or failure to recover VWAP rejected the shortlist
- Afternoon result: **sit** through 15:55 ET, with no equity fill
- Separate midday job: first Wheel cash-secured put on [**T**](/evidence/symbols/T) (see [Wheel desk](/resources/agentic-wheel-desk))

### 2026-07-23: Bear RS day: 4 plans, late sit

[Full desk log](/evidence/desk-log-2026-07-23)

- Regime: SPY/QQQ red under VWAP: RS Regime D/A only
- Live: [IREN](/evidence/symbols/IREN) (TP); [CLSK](/evidence/symbols/CLSK) (structural cut: mid-reclaim mistake); [JHX](/evidence/symbols/JHX) (TP, soft MACD); [HUT](/evidence/symbols/HUT) (TP)
- Late power hour: empty RSI/MACD scan, so **sit** through 15:55
- Lesson: mid-reclaim ≠ Regime D; size-up only on full stack

### 2026-07-21: Bull day: fail-closed arms, late SBRA TP

[Full desk log](/evidence/desk-log-2026-07-21)

- Midday [BANC](/evidence/symbols/BANC) MAE, then same-day re-entry ban held
- Power-hour scanner upgrade (RVOL + RSI/MACD gates, not %Δ sort)
- EGO / HBM / [SBRA](/evidence/symbols/SBRA) 15:20 / TSM: **fail closed** sits
- Late SBRA TP after false-start; [BA](/evidence/symbols/BA) hard-flat cleanup
- Lesson: review ≠ place; lost holds do not get chased
- Also: [HPE](/evidence/symbols/HPE) lesson page covers the Jul 22 MAE-in-sweep rule that followed

## FAQ

**Is this a trade signal?** 
No. These are educational process logs. They are not recommendations to buy or sell any security.

**Where is the detail?** 
Open the linked Evidence **Desk log** for that date: scanners, gate tables, plans, exits, and lessons.

**What about wallet P&L?** 
We publish exit class and process outcomes, not personal dollar track records. See [Session stories without the ledger](/evidence/session-stories-without-the-ledger).

## Related

- [Equity scalp playbook](/resources/agentic-trading-playbook)
- [Wheel desk notes](/resources/agentic-wheel-desk)
- [Options map for Agentic experiments](/resources/agentic-options-map)
- [Sitting on your hands](/resources/sitting-on-your-hands)
- [When the desk should sit](/resources/when-the-agent-should-sit)
- [Robinhood MCP](/resources/robinhood-mcp)
- [Evidence](/evidence)

## Disclaimer

Educational only. Not an offer to buy or sell securities. Past patterns are not future results. You are responsible for every order placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood. See [/disclaimer](/disclaimer).

────────────────────────────────────────
# Wheel Desk Notes
URL: https://www.soyhands.com/resources/agentic-wheel-desk
Markdown: https://www.soyhands.com/resources/agentic-wheel-desk.md
Updated: 2026-07-28
Summary: Sanitized cash-secured put / covered call desk ledger: completed cycles, published skips, earnings gates, collateral fits, and Option Alpha-style Wheel defaults via MCP.

Educational resource. Not financial advice. Example parameters are illustrative. Agentic options trading can lose capital, including assignment into 100-share stock risk. Not affiliated with Robinhood Markets, Inc.

# Wheel Desk Notes (Sanitized Skips & Setups)

Living ledger of **cash-secured put / covered call** desk decisions on a Robinhood Agentic account via Trading MCP. **Skip is a publishable outcome**: see [Sitting on your hands](/resources/sitting-on-your-hands).

**New here?** A cash-secured put pays a credit now while reserving enough cash to buy 100 shares at a chosen price. If assigned, you buy the shares. The Wheel may then sell a covered call against those shares. Read [Wheel capital](/resources/wheel-capital) before treating the credit as income.

## Playbook (short)

| Rule | Default |
|------|---------|
| Structure | Sell a cash-secured put, then possibly receive 100 shares, then possibly sell a covered call |
| Time to expiration | Usually 30-45 days when opened |
| Strike selection | Usually about 0.20-0.30 delta, with a price we are willing to pay for the shares |
| Exit | Buy back near 50% of the original credit, or review by 21 days before expiration |
| Timing | Scout after the open settles; live mid-morning |
| Hard skip | Earnings before expiration; too much cash required; [soft tape](/resources/thresholds) (index or underlying); collateral misfit |

Trading MCP places one option leg at a time. Multi-part spreads and iron condors are not placed as one order through this tool. Capital: [Wheel capital](/resources/wheel-capital). Other structures and tool limits: [Options map](/resources/agentic-options-map). Soft tape tests: [thresholds §3](/resources/thresholds).

## Desk ledger

Newest first.

### 2026-07-28: First Wheel cycle closed (T half-credit BTC)

[Full desk log](/evidence/desk-log-2026-07-28).

- **Completed cycle:** the July 24 **T** September **$22** cash-secured put (about **$0.27** credit) was **bought back near $0.14** at 10:19 ET when the mark hit the half-credit rule.
- Collateral freed. Starter slot open again. No assignment into shares (Path A in [Wheel capital](/resources/wheel-capital)).
- **SKIP** a replacement put at the 11:00 mid-morning window. Tape recovered enough to pass the bear gate, but several willing-to-own names sat in an earnings print window (**F**, **SOFI**, **HOOD**, **RIVN**). The best collateral-fit chain (**RIVN**) also had a wide bid-ask spread.
- **14:15 close:** broker re-check showed zero open option legs and zero equity. No healthy short premium to hold or buy back early. Wheel desk closed for the day; afternoon equity scalp owns the book after 14:25 ET.
- Action: **close the winner, stay flat on new short premium.** Freeing the slot is not a refill signal when earnings still block the menu.

### 2026-07-27: Manage day, no new CSP

[Full desk log](/evidence/desk-log-2026-07-27).

- **SKIP** a second cash-secured put. The open **T** September **$22** put from July 24 still occupies the starter slot (one short put at a time).
- Midday manage window (through 14:15 ET): the put stayed open; the live mid was about **$0.175**, still above the rough **$0.14** half-credit buyback level, with about **53 days** left (not yet in the 21-day review band).
- Action: **hold**. No buy-to-close. No new short premium. The open put carries into the next session.
- Scheduled automation had an auth hiccup before the live window; later ticks re-checked the broker and kept the same hold rule.

### 2026-07-24: First cash-secured put on T

[Full desk log](/evidence/desk-log-2026-07-24). [Symbol notes](/evidence/symbols/T)

- The first candidate list failed because earnings were too close or one contract required too much cash
- Later fill: one **T** cash-secured put, expiring **September 18, 2026**, with a **$22** strike and about **$0.27 per share** of credit
- One contract received about **$27 before fees** and reserved **$2,200** to cover possible assignment
- If assigned, the effective share cost is about **$21.73 before fees**
- Plan: buy the put back near **$0.14**, or review by 21 days before expiration; no second short put while this one is open
- **Outcome (2026-07-28):** bought back near half credit. Cycle closed. See [today's Wheel ledger](/resources/agentic-wheel-desk) and [desk log](/evidence/desk-log-2026-07-28)
- The [full desk log](/evidence/desk-log-2026-07-24) explains the maximum reward, stock downside risk, locked cash, and likely days-versus-weeks paths

The $27 credit is the maximum reward, not the maximum loss. Assignment creates the risk of owning 100 falling shares.

## FAQ

**What is the Wheel setup?** 
A cash-secured put may turn reserved cash into 100 shares. If assigned, a covered call may be sold against those shares. This desk usually opens with 30-45 days remaining and buys the put back near half the original credit or by the 21-day review point.

**When does the agent sit?** 
When every willing-to-own name fails a hard gate: especially earnings inside the short option life, collateral that does not fit the starter band, or soft tape per [thresholds §3](/resources/thresholds) (SPY below MA20 and MA50, weak closes, elevated VIX, or the underlying below its MA50). Those skips are published here.

**Is this a trade signal?** 
No. Entries are educational process logs. They are not recommendations to buy or sell any security.

## Related

- [Measurable thresholds](/resources/thresholds)
- [Wheel capital & timelines](/resources/wheel-capital)
- [Options map for Agentic experiments](/resources/agentic-options-map)
- [Sitting on your hands](/resources/sitting-on-your-hands)
- [When the agent should sit](/resources/when-the-agent-should-sit)
- [Robinhood MCP](/resources/robinhood-mcp)
- [Equity scalp playbook](/resources/agentic-trading-playbook)
- [Evidence field notes](/evidence)

## Disclaimer

Educational only. Not an offer to buy or sell securities. Past patterns are not future results. You are responsible for every order placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood. See [/disclaimer](/disclaimer).

────────────────────────────────────────
# Options Map for Agentic Experiments
URL: https://www.soyhands.com/resources/agentic-options-map
Markdown: https://www.soyhands.com/resources/agentic-options-map.md
Updated: 2026-07-28
Summary: Steady income versus strong reward on a Robinhood Agentic book: CSP, long options including planned 0DTE, covered calls, MCP single-leg workarounds for defined risk, and how to stay in cash if you do not want stock.

Educational resource. Not financial advice. Example parameters are illustrative. Options can lose money quickly, including total loss of premium paid or assignment into stock. Not affiliated with Robinhood Markets, Inc.

# Options Map for Agentic Experiments

A plain map of **what this desk can try** on a Robinhood **Agentic** cash book via Trading MCP, what belongs in the **Robinhood app** instead, and how to choose between **steady income** and **strong reward** without pretending they are the same product.

This page is also where we document **MCP learning**: what the tool can place today, what we approximate with extra single-leg orders, and what we still leave to the app.

Desk notes: [Wheel desk](/resources/agentic-wheel-desk), [Scalp desk](/resources/agentic-scalp-desk), [Evidence desk logs](/evidence). Capital locks: [Wheel capital](/resources/wheel-capital).

## Desk constraints (start here)

| Constraint | What it means |
|------------|----------------|
| **Agentic cash book** | No naked margin shorts; cash-secured puts need cash equal to strike × 100 |
| **Options Level 2** | Long calls/puts, cash-secured puts, covered calls are in scope on the account |
| **Trading MCP** | **Single-leg** place/cancel only. True multi-leg tickets (spreads, condors) are **not** placeable on this MCP path yet |
| **Starter rule** | Prefer **one** open short put at a time; keep an equity scalp sleeve free |

Multi-leg “defined risk” income with efficient collateral usually needs the **RH app**. Below we also cover a **two-order MCP workaround** and its costs.

## Two jobs: steady earnings vs strong reward

| Job | What you optimize for | Typical tools | Honest trade-off |
|-----|------------------------|---------------|------------------|
| **Steady earnings** | Small, repeatable credits; rules over drama | CSP, covered call, (app) credit spreads / condors | Capped win. Cash can lock. Stock ownership is possible |
| **Strong reward** | Asymmetric upside if the thesis is right | Long call / long put, debit spreads (app), equity | Many ideas expire worthless or stop out. Not a paycheck |

Mixing the jobs in one ticket is how people get confused. A CSP is **rent for downside risk**, not a savings account and not a call on a moonshot.

## Strategy map (experiments)

### A. Live on this MCP desk today

| Strategy | Job | Own stock? | Liquidity while open | Pros | Cons |
|----------|-----|------------|----------------------|------|------|
| **Cash-secured put (CSP)** | Steady | Only if assigned | Locks strike × 100 | Clear rules; theta can help; assignment is planned | Capped credit; crash loss ≫ premium; cash locked |
| **CSP, manage early** | Steady, cash-first | Avoid if you BTC before assign | Same lock until BTC | Keeps the Wheel optional; frees cash on ~50% / ≤21 DTE | Pin / ITM risk still exists near expiry |
| **CSP + long lower put** (two tickets) | Steady, capped crash | Less likely as the default | Often still locks short-strike cash **plus** long-put debit | Caps economic downside vs naked CSP | Leg risk; usually worse capital than an app spread |
| **Covered call (CC)** | Steady | **Yes** (need 100 shares) | Capital sits in stock | Income on shares you already hold | Upside capped; overnight stock risk |
| **Long call** | Strong reward | No | Debit only (premium paid) | Upside without buying 100 shares | Can lose 100% of debit; timing matters |
| **Long put** | Strong reward / hedge | No | Debit only | Pays if the stock falls hard | Often decays; not “income” |
| **0DTE long call / put** (SPY / QQQ / IWM) | Strong reward | No | Debit only; same-day expiry | Capped max loss = premium; liquid underlyings | Gamma / time decay extreme; hard flat by mid-afternoon; **not** Wheel income |
| **Equity scalp** | Tactical | Briefly | Flexible | Full stock P&L; no CSP lock | Different playbook; hard flat clocks |

Field notes: [Equity scalp playbook](/resources/agentic-trading-playbook), [Wheel desk](/resources/agentic-wheel-desk).

### B. App / higher-level track (not MCP place yet)

| Strategy | Job | Own stock? | Why people use it | Watch-outs |
|----------|-----|------------|-------------------|------------|
| **Put credit spread** | Steady | Usually no | Defined risk; collateral ≈ width, not full strike | Smaller credit than naked CSP; still can lose the width |
| **Call credit spread** | Steady / mild bear | No | Collect premium with a ceiling | Needs a real thesis; not free money |
| **Iron condor** | Steady “range” | No | Profit if price stays quiet | Loses on big moves either way |
| **Debit call / put spread** | Strong reward, cheaper | No | Caps loss vs naked long option | Caps win too |

Document experiments here when the desk actually runs them. Until then, this table is a **menu**, not a live session log.

## MCP learning: fake multi-leg with separate orders

Trading MCP places **one leg per order**. Robinhood’s own MCP notes say multi-leg structures belong in the **apps**, and stock+option “combo” orders are not supported on this path.

That still leaves a useful experiment: **place two (or more) single-leg orders** that together look like a spread.

### Synthetic put credit spread (short put + long put)

| Step | Order | Role |
|------|-------|------|
| 1 | **Sell** cash-secured put at strike A | Collect credit; willing to buy shares at A |
| 2 | **Buy** put at lower strike B (same expiry, same size) | Cap how bad a crash can get |

Illustrative shape: short **$22** put, long **$20** put. Width is **$2**. Rough economic max loss on the package is on the order of **width minus net credit** (about **$200** per package before fees), not “stock to zero on full short-strike cash.”

| | True app put credit spread | Two MCP single-leg orders |
|--|----------------------------|---------------------------|
| Ticket | One multi-leg order | Two separate orders |
| Fill risk | Legs usually fill together | **Leg risk:** one fills, the other misses or slips |
| Crash cap | Yes (defined width) | Yes **if** both legs are on |
| Cash on Agentic | Often collateral ≈ width | Cash CSP may still lock **strike A × 100**, and you also **pay** for the long put |
| Desk fit | Best when you want defined risk **and** free capital | Experiment when you accept worse liquidity for a hedge |

**Do not confuse:** selling a put and separately selling a call is not this hedge. It is a different risk book.

### Downside controls that do not need a second leg

These are the default Agentic Wheel controls today. They shrink the largest loss by **exiting**, not by hedging:

1. **Profit BTC** near ~50% of credit (free collateral early)
2. **Time BTC** by ≤21 DTE (avoid expiry-week pin theater)
3. **Defense BTC** if the put mark blows out as the stock cracks toward the strike (take a small loss vs sit into a hole)
4. **Size and strike** (one contract; OTM delta band; skip earnings)
5. **Resting buy-to-close** after fill (limit near half credit) so manage is mechanical

For crash risk **and** efficient capital, prefer a **real put credit spread in the RH app**. For crash risk **on MCP today**, the two-order synthetic is the honest workaround, with leg risk and possible full CSP lock documented up front.

### Experiment status (desk)

| Idea | Status on this desk |
|------|---------------------|
| Naked CSP + 50% / 21 DTE manage | **Live** (see [Wheel desk](/resources/agentic-wheel-desk), [2026-07-24 log](/evidence/desk-log-2026-07-24); first cycle BTC [2026-07-28](/evidence/desk-log-2026-07-28)) |
| Defense BTC / resting BTC as rules | Playbook-compatible; publish when a session uses them |
| Synthetic short+long put via two MCP orders | **Documented menu**; not yet a published live experiment |
| **0DTE long call / put** (SPY→QQQ→IWM) | **Playbook drafted** (chat / plan-only). Debit caps, 15:15 hard flat, sit on Wheel conflict. No launchd auto-trade until several logged sessions. Not a paycheck job. |
| App multi-leg put credit spread | **Future track** (better capital for defined risk) |

When we run a synthetic or app-spread session, it lands in Evidence that day and gets a headline on the Wheel desk index.

## If you want cash liquidity (and do not want stock)

Assignment is the middle of a classic Wheel. It is **not** failure in that playbook. It **is** a problem if your real goal is “keep buying power in cash.”

Practical ladder:

1. **Do not sell puts you refuse to own.** Willing-to-own is the first gate.
2. **CSP with hard manage:** buy to close near **~50% of credit** or by **≤21 DTE**. Do not milk into expiry week if shares are unacceptable.
3. **Defense BTC** if the short put mark blows out before assignment week.
4. **Prefer defined-risk put credit spreads in the app** when you want income **without** full strike collateral and without assignment as the default path.
5. **Optional MCP experiment:** add a long lower put (two tickets) knowing cash may still stay locked on the short strike.
6. **Use long options (or debit spreads)** when the idea is directional: you risk premium, not “become a shareholder by force.”
7. **Skip covered calls** if you are trying to stay in cash. They start from stock.

Pin risk remains near expiry if a short put finishes ITM. Cash-first still means **watching the clock**, not ignoring it.

## Failure mode vs favorable path (CSP example)

**Costs money:** stock gaps or trends well below the short strike. You get assigned into a drawdown that can dwarf the credit you collected. Premium was rent for that risk, not a free upgrade to idle cash.

**Turns favorable:** stock holds or rises, volatility calms, put mark falls toward ~half credit. You buy to close, keep most of the credit, and unlock collateral. A rip far above the strike is great for the **put seller**, but you still only keep the credit. You do not own the moonshot.

**With a long lower put on:** the crash path is capped near the width, but you paid for that insurance and may still have short-strike cash locked until the short put is closed.

More capital timing detail: [Wheel capital & timelines](/resources/wheel-capital). First live CSP framing: [Desk log 2026-07-24](/evidence/desk-log-2026-07-24).

## How we publish experiments

| Surface | What goes there |
|---------|-----------------|
| [Evidence desk logs](/evidence) | Day narrative: scanners, gates, sits, fills, lessons (sanitized) |
| [Wheel desk](/resources/agentic-wheel-desk), [Scalp desk](/resources/agentic-scalp-desk) | Newest-first headlines |
| This page | Strategy map, MCP limits, synthetic multi-leg learning, cash-first choices |
| [Wheel capital](/resources/wheel-capital) | What locks, for how long, assignment path |
| [Robinhood Trading MCP](/resources/robinhood-mcp) | What the tool is for, review → place guardrails |

Skip and sit remain publishable outcomes. See [Sitting on your hands](/resources/sitting-on-your-hands) and [When the desk should sit](/resources/when-the-agent-should-sit).

## FAQ

**Is this a signal list?**  
No. It is an educational map of structures the Agentic experiments may use. Nothing here is a recommendation to buy or sell any security.

**Why not automate spreads tomorrow?**  
Trading MCP on this desk is single-leg today. True multi-leg tickets belong in the app track until that changes.

**Can we fake a multi-leg on MCP anyway?**  
Yes: two single-leg orders (for example short put + long lower put). That can cap crash risk. It does not remove leg risk, and on a cash CSP book it often does **not** free the full short-strike collateral the way an app spread can.

**What is the simplest control without a second leg?**  
Manage and defense buy-to-close rules, plus size and earnings gates. Exit early beats hoping the hole fills.

**Can I do “strong reward” and “steady earnings” on the same small book?**  
Yes as **separate sleeves** with separate rules. No as one confused ticket that wants paycheck income and lottery upside at once.

## Related

- [Wheel capital & timelines](/resources/wheel-capital)
- [Wheel desk notes](/resources/agentic-wheel-desk)
- [Equity scalp playbook](/resources/agentic-trading-playbook)
- [Equity scalp desk notes](/resources/agentic-scalp-desk)
- [Robinhood Trading MCP](/resources/robinhood-mcp)
- [When the desk should sit](/resources/when-the-agent-should-sit)
- [Evidence](/evidence)
- [Glossary](/glossary)
- [Disclaimer](/disclaimer)

## Disclaimer

Educational only. Not an offer to buy or sell securities. Past patterns are not future results. You are responsible for every order placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood. See [/disclaimer](/disclaimer).

────────────────────────────────────────
# Wheel Capital & Timelines
URL: https://www.soyhands.com/resources/wheel-capital
Markdown: https://www.soyhands.com/resources/wheel-capital.md
Updated: 2026-07-24
Summary: What short-premium locks: collateral sleeves, how long cash is tied up, assignment into stock, and why frequent Wheel work needs more capital than a busy scan.

Educational resource. Not financial advice. Illustrative bands only. Not a recommendation of account size or strategy. Agentic options trading involves substantial risk of loss, including assignment into stock. Not affiliated with Robinhood Markets, Inc.

# Wheel Capital & Timelines

What selling cash-secured premium actually commits: **how much cash locks**, **for how long**, **what can happen while it is locked**, and why **frequent** short-premium work needs more capital than a busy scanner.

Companion desk ledger: [Wheel desk](/resources/agentic-wheel-desk). Equity day-scalps are a separate sleeve: see the [scalp playbook](/resources/agentic-trading-playbook).

## Investment thesis (one paragraph)

Sell **cash-secured** short premium on names you are willing to own, targeting **30-45 DTE** puts (~0.20-0.30Δ), taking profits at **~50% of credit** or closing by **≤21 DTE**. Capital is the product: every open CSP locks **strike × 100** of cash until buy-to-close, expire, or assignment. Frequency is limited by that lock: not by how many names you can scan.

## Book split (sleeves)

Treat sleeves as hard partitions on a small cash Agentic book that also runs equity scalps:

| Sleeve | Role | Starter rule | While open |
|--------|------|--------------|------------|
| **Wheel collateral** | Secure 1 CSP (or hold 100 shares + CC) | ≤ ~**40%** of buying power | Locked until BTC / assignment / call-away |
| **Equity scalp** | Opening-range / end-of-day small equity risk | Keep a **fixed residual** free (illustrative floor on the order of hundreds of dollars) | Day-trade; flat by session hard times |
| **Buffer** | Fees, rejects, assign drag | Remainder: do not CSP to the wall | Soft reserve |

**Sizing formula (illustrative):**

```text
max_strike ≈ floor( (0.40 × buying_power) / 100 )
residual after CSP must leave the scalp sleeve intact
collateral = strike × 100 × contracts # starter: 1 contract
```

If the best willing-to-own name needs more collateral than the band allows → **skip** and publish why. Do not “almost” fit.

On a **small** agentic wallet, a ~40% starter band often means max strikes in the **tens of dollars** for one contract: not mega-cap full CSPs.

## What gets tied up, and for how long

### Cash-secured put

| Item | Tied | Typical duration under manage rules | Ends when |
|------|------|-------------------------------------|-----------|
| Cash collateral | strike × 100 | Days to ~3 weeks | BTC @ ~50% credit, BTC @ ≤21 DTE, expire OTM, or **assignment** |
| Opportunity | Cannot reuse that cash for another CSP | Same | Same |

| Path | Rough timeline | Capital outcome |
|------|----------------|-----------------|
| **A: Early winner** | Often 3-14 calendar days if ~50% hits | Collateral freed; keep part of credit |
| **B: Time stop** | Enter 30-45 DTE → manage by ≤21 DTE left → often ~1-3 weeks | BTC; free cash |
| **C: Assign** | Through expiry (or earlier ITM) | Cash → **100 shares** |

Starter desk: **max one** open CSP: one Wheel income unit at a time.

### Assigned shares

Assignment is a **state change**, not automatically a “failed” trade: income book → **stock book** with optional covered calls. Expect weeks to months if you keep selling calls; full stock drawdown risk; overnight/gap risk; no same-day flat clock like equity scalps.

### Covered call

Shares stay tied; upside capped above the short call; called away returns cash (+ keeps call premium).

## What can happen while capital is locked

| Event | While CSP open | While holding shares (± CC) |
|-------|----------------|-----------------------------|
| Underlying drifts down | Put ITM; assign risk rises | Real stock drawdown |
| Underlying rips up | Often path A (50% BTC) | Short call may be tested; upside capped |
| IV crush / spike | Helps or hurts short premium marks | Same for short call; stock can gap |
| Earnings / binary | Hard reject into window | Do not hold short premium into print |
| Assignment / early exercise | Cash becomes shares | Short call can be assigned early |
| Second CSP | Blocked by design (starter) | Usually blocked until shares called/sold |

**Honest failure mode:** assignment turns reserved cash into a full 100-share position you must manage or exit: a drawdown can exceed many months of collected premium.

## When the desk acts (illustrative ET)

| Window | Action |
|--------|--------|
| 09:30-10:00 | No new short premium (open auction) |
| Mid-morning settle | Scout / optional live CSP or CC |
| Through early afternoon | Manage (50% / 21 DTE) |
| Late day | Equity EOD scalp owns the desk: no new Wheel |

**Cadence on a small starter book:** frequent **scouts and skips**, not frequent fills. Rough band of closed CSP cycles per month is low if you hit early winners; lower still if assigned into stock.

## Minimum capital for “frequent” short-premium

Define **frequent** as running **≥2 independent CSP sleeves** most weeks (or replacing capital quickly after 50% BTC) without starving the equity scalp sleeve.

| Goal | Rough scale (illustrative) |
|------|----------------------------|
| One liquid CSP in mid-teens-low-$30 names + scalp sleeve | Small five-figure buying power |
| Comfortable one CSP with fewer “strike too high” skips | Higher five figures |
| Two concurrent CSPs (or CSP + share lot) without >60% BP in Wheel | Low-mid five figures toward **~$20k-$30k+** |
| Multi-name rotation (3 sleeves) | Substantially more |

Premium income scales with **contracts × credit**, not scanner busyness. Do not loosen collateral % to fake frequency: add capital or accept selective skips.

## Lifecycle storyboard

```text
[Idle cash]
 │ scout skip → still idle (publish why)
 ▼
[CSP open] collateral locked
 ├─ BTC @ ~50% or ≤21 DTE → cash free
 └─ Assigned → 100 shares ± CC → called away or sell stock → idle
```

## Before locking capital

- Willing to **own 100 shares** at strike for weeks
- Earnings clean for the short option life
- Collateral fits the starter band; scalp residual intact
- Accept max one Wheel unit (starter)
- Accept that “frequent” may mean frequent **sits**, not frequent fills
- If assignment is intolerable at this size → do not sell the put. Crash-cap options (defense BTC, two-order long put, app put credit spread): [Options map](/resources/agentic-options-map).

## Related

- [Options map for Agentic experiments](/resources/agentic-options-map)
- [Wheel desk](/resources/agentic-wheel-desk)
- [When the agent should sit](/resources/when-the-agent-should-sit)
- [Robinhood MCP](/resources/robinhood-mcp)
- [Equity scalp playbook](/resources/agentic-trading-playbook)
- [Disclaimer](/disclaimer)

## Disclaimer

Educational only. Not an offer to buy or sell securities. Illustrative parameters are not a performance guarantee or account-size advice. Past patterns are not future results. You are responsible for every order placed in your accounts. This site is not a broker-dealer and is not affiliated with Robinhood. See [/disclaimer](/disclaimer).