Qc Deep Feature Forensics

12-dimensional technical feature attribution engine — compares winner vs loser trade entry conditions using RSI, Bollinger, MACD, volume surge, gap, and more...

Security Scan

The skill's code, instructions, and resource requirements are consistent with its stated purpose (extracting features from an orders CSV and fetching historical data from Yahoo Finance) and do not request unrelated credentials or surprising system access.

Purpose & Capability

Name/description (12-dimensional feature attribution for winner vs loser trade entries) matches the code and SKILL.md. The script reads an orders CSV, reconstructs closed trades, downloads OHLCV from Yahoo via yfinance, computes indicators, and writes feature and report files. Required binaries (python3, pip3) and the listed pip dependencies are appropriate for the task.

Instruction Scope

Runtime instructions are narrow: install the listed Python packages, run `python3 deep_forensics.py <orders.csv>`. The code only reads the provided orders CSV, writes outputs and a per-ticker cache directory next to the CSV, and makes network calls to Yahoo Finance via yfinance. There are no instructions to read unrelated system files, environment variables, or to post data to unknown endpoints.

Install Mechanism

There is no packaged installer; SKILL.md instructs doing `pip3 install pandas numpy yfinance` and a requirements.txt is included. This is a normal approach for a Python script, but pip installs run arbitrary package code from PyPI — recommend using a virtual environment or isolated environment when installing. No downloads from unknown URLs or archive extraction are present.

Credentials

The skill requests no environment variables, credentials, or config paths. Its network use is limited to yfinance (Yahoo Finance) for historical data, which is consistent with the functionality. No unrelated secrets are requested or accessed.

Persistence & Privilege

The skill does not request always:true and is user-invocable only. It writes a local cache directory (yfinance_cache) and output CSV/markdown next to the orders CSV — typical and proportionate for caching. It does not modify other skills or global agent settings.

Assessment

This skill appears coherent and implements what it claims, but before running: (1) install dependencies inside a virtualenv/container to limit pip risk; (2) review the orders CSV you supply (it will be read and used as the sole input) and ensure it contains only the data you intend to analyze; (3) be aware the script will write a yfinance_cache/ folder beside your orders CSV and output files (feature CSV and feature_diagnosis.md); (4) the script requires internet on first run to fetch Yahoo data — if you need offline runs, pre-populate the cache; (5) if you handle sensitive trading/account data, inspect the full script locally (the included Python file appears to log only basic status messages) and run in an isolated environment. Overall the behavior is proportionate to the stated purpose.

Go beyond P&L analysis. This skill answers: "What market microstructure conditions were present when winning trades were entered vs losing trades?"

When to use

• "Why do some of my trades win and others lose?"
• "What entry conditions lead to profitable trades?"
• "Run a feature attribution on my backtest"
• "What-if analysis: would filtering out RSI > 70 trades help?"
• After running qc-order-forensics for high-level diagnosis, use this for deep-dive

How it works

python3 deep_forensics.py <orders.csv>

Pipeline

1. Order Reconstruction: Groups raw orders into closed trade pairs (buy group + sell group per contract)
2. Batch Data Download: Fetches historical daily OHLCV from Yahoo Finance per ticker (cached to local CSV to avoid re-downloading)
3. Technical Indicator Pre-computation: Calculates all 12 indicators on the full ticker history
4. Feature Extraction: For each trade, extracts a 12-dimensional feature vector at the entry date
5. Winner vs Loser Comparison: Statistical dual-sample comparison with diagnostic interpretation
6. What-If Simulation: Tests hypothetical filters and measures their net impact on total P&L

12 Feature Factors

Factor	Description	Format
gap_pct	Gap open % vs previous close	%
volume_surge	Volume / 10-day average volume	x
ma5_deviation	(Close - MA5) / MA5	%
ma20_deviation	(Close - MA20) / MA20	%
volatility_expansion	Intraday range / 5-day ATR	x
intraday_return	(Close - Open) / Open	%
rsi_14	RSI(14)	0-100
bb_position	Position in Bollinger Band (0=lower, 1=upper)	0-1
macd_hist_norm	MACD histogram / Close	ratio
consecutive_up_days	Count of consecutive up-close days	days
distance_from_20d_high	(Close - 20d High) / 20d High	%
prev_day_return	Previous day's return	%

Report Output

Section 1: Feature Mean Comparison (Winners vs Losers)

Table showing each factor's mean for winning vs losing trades, the delta, and a diagnostic interpretation. Example:

| Gap Open %    | +3.2% | +1.1% | +2.1% | Winners enter on stronger gaps |
| Volume Surge  | 2.8x  | 1.4x  | +1.4x | Volume confirmation helps      |

Section 2: What-If Filter Analysis

Simulates applying each filter rule to the full trade set and measures:

• How many losers would be avoided
• How many winners would be accidentally killed
• Net impact on total portfolio ROI
• Verdict: Shield (improves total P&L), Toxic (kills outlier wins), or Marginal

Filters tested include: gap > 2%, volume > 2x, below MA20, RSI > 70, BB > 0.95, consecutive up > 3, near 20d high, negative intraday return.

Combined filter: Stacks all "Shield" filters and reports the combined effect.

Section 3: Winner Entry Profile

Statistical percentile ranges (25th-75th) for each factor among winning trades. Defines the "ideal entry environment" envelope.

Output Files

• <name>_features.csv — Full feature matrix for all trades
• feature_diagnosis.md — Complete markdown report

Caching

YFinance data is cached per-ticker in yfinance_cache/ alongside the orders CSV. Subsequent runs on the same data skip downloads entirely.

Dependencies

pip3 install pandas numpy yfinance

Rules

• Do not modify files in yfinance_cache/ manually. The cache uses date-range coverage checks. Corrupted cache files will cause silent data gaps in indicator calculations.
• Do not change indicator parameters (RSI period, Bollinger window, etc.) without understanding that all 12 factors are calibrated together. Changing one shifts the entire winner/loser comparison baseline.
• What-If verdicts ("Shield" vs "Toxic") are based on total portfolio ROI impact, not win rate. A filter that improves win rate but kills outlier wins is marked "Toxic" because total P&L decreases. Do not override this logic.
• Internet access is required on first run for each ticker. Subsequent runs use cached data. If you run in an offline environment, pre-populate the cache directory.
• Minimum 20 closed trades required for statistically meaningful feature comparison. With fewer trades, the report will still generate but conclusions are unreliable.