stockbee-setup-fluency-trainer

---
name: stockbee-setup-fluency-trainer
description: Use when building a Stockbee-style setup model book from momentum-burst screener candidates, tracking 3-day/5-day outcomes, or reviewing setup quality before increasing position size.
category: Performance
version: 1.0.0
tools: []
---

# Stockbee Setup Fluency Trainer

Build and maintain a model book for Stockbee-style Momentum Burst setups, turning daily screener candidates into structured study records, then updating them once the 3-day and 5-day windows mature and summarizing which setup features are working or failing.

## When to Use

Use when the user wants to study Stockbee Momentum Burst setups systematically; asks to build a model book from `stockbee-momentum-burst-screener` output; wants to review failed candidates, missed trades, or setup quality; wants 3-day/5-day forward returns, MFE, MAE, and stop-hit outcomes; wants to improve setup recognition before increasing position size; or asks which Stockbee tags should be promoted, downgraded, or filtered.

## Workflow

**Ingest candidates.** Take the output of a `stockbee-momentum-burst-screener` run (or a compatible candidate list) and record each as a study entry: ticker, setup date, primary trigger, rating, setup score, setup tags, entry/stop reference, risk percent to stop, and any human label, decision, or notes. Include rejected candidates only when intentionally building a negative-example set.

**Update outcomes.** Once 3 and 5 trading days have passed, record for each candidate: the forward close return at each horizon, MFE and MAE over each horizon, whether the stop was hit and when, and an outcome tag:

- `STRONG_WINNER` — 5-day close return ≥ 8% or MFE ≥ 12%, with no stop hit
- `WORKED` — 5-day close return ≥ 4% or MFE ≥ 6%, with no stop hit
- `FAILED_STOP` — stop touched within the horizon
- `FAILED_FADE` — forward return ≤ -2% without a recorded stop hit
- `CHOPPY_FAILURE` — large adverse excursion with poor forward progress
- `NEUTRAL` — no decisive follow-through or failure
- `PENDING` — not enough future bars yet

**Summarize cohorts.** Group matured records by rating, primary trigger, and setup tags, with a minimum sample size (5 is a reasonable floor) before drawing conclusions.

**Convert evidence into practice.** For cohorts with enough examples: promote tags showing a high win rate, positive 5-day expectancy, and acceptable average MAE; downgrade or filter tags showing weak 5-day expectancy, frequent stop hits, or repeated fade failures. Always inspect representative charts manually before changing a trade rule, and log accepted lessons in `trader-memory-core` or the user's monthly review process. Treat cohort output as evidence prompts, not automatic rule changes.

## Output

Return, in chat: the ingested candidate list; the updated outcome table per horizon with outcome tags; and the cohort summary with win rates, expectancy, and any promote/downgrade recommendations, each tied to its sample size.

## Key Principles

- A model book compounds in value over time — consistency in what gets recorded matters more than any single outcome.
- Treat small-sample cohorts as directional only; require the stated minimum sample size before recommending a tag change.
- Separate the mechanical outcome tag from the human judgment call — both should be preserved in the record.