stats

name	stats
description	Compute and display trading performance statistics. Use when the user asks about their performance, win rate, P&L, or wants a trading summary. Also use proactively at natural boundaries (end of quarter, end of year, after a significant drawdown or winning streak).

Stats

Compute trading statistics from account data.

Quick use

lafmm stats                          # visual
lafmm stats --json                   # JSON (for reasoning)
lafmm stats --period 2026            # year
lafmm stats --period 2026-Q1         # quarter
lafmm stats --period 30d             # last 30 days
lafmm stats ibkr                     # specific account
lafmm stats --json --period 2026-Q1  # combinable

Use --json when you need to reason about numbers. Omit it when showing the user their stats.

JSON output fields

{
  "first_date": "2025-01-02",
  "last_date": "2025-06-30",
  "period": "all",
  "market_days": 125,
  "active_days": 80,
  "total_trades": 200,
  "buys": 95,
  "sells": 105,
  "round_trips": 45,
  "open_positions": [{"symbol": "NVDA", "side": "long", "qty": 50, "avg_price": 128.50, "open_date": "2026-03-15", "signal": "BUY 10(d)"}],
  "wins": 30,
  "losses": 15,
  "breakeven": 0,
  "win_rate": 66.7,
  "total_pnl": 3500.00,
  "avg_win": 125.00,
  "avg_loss": -100.00,
  "largest_win": 800.00,
  "largest_loss": -600.00,
  "expectancy": 35.00,
  "profit_factor": 1.88,
  "concentration_pct": 35.0,
  "start_capital": 10000.00,
  "end_capital": 15000.00,
  "total_deposits": 1500.00,
  "total_withdrawals": 0.0,
  "total_fees": 80.00,
  "total_dividends": 30.00,
  "total_tax": 5.00,
  "total_interest": -10.00,
  "total_platform_fees": 0.0,
  "trading_return_pct": 30.0,
  "max_drawdown_pct": 12.0,
  "max_drawdown_days": 20,
  "longest_win_streak": 8,
  "longest_loss_streak": 4,
  "sharpe": 1.50,
  "fees_pct_of_pnl": 2.3,
  "pre_system_trades": 20,
  "pre_system_win_rate": 55.0,
  "post_system_trades": 80,
  "post_system_win_rate": 62.0,
  "signal_trades": 50,
  "signal_win_rate": 65.0,
  "discretionary_trades": 30,
  "discretionary_win_rate": 50.0,
  "order_types": {"limit": 150, "market": 30, "stop": 20},
  "avg_hold_days": 7.5,
  "longest_hold_days": 30,
  "longest_hold_symbol": "AAPL",
  "symbols_traded": 12,
  "top_symbols": [{"symbol": "AAPL", "pnl": 1200.00, "round_trips": 8, "wins": 6, "losses": 2, "win_rate": 75.0}, ...],
  "monthly_pnl": [{"month": "2025-01", "pnl": 500.00}, ...],
  "rolling": [{"window": 10, "trip_number": 10, "win_rate": 70.0, "expectancy": 50.00, "profit_factor": 1.80}, ...],
  "robustness": [
    {"excluded": "AAPL", "reason": "best", "round_trips": 40, "wins": 25, "losses": 15, "win_rate": 62.5, "expectancy": 20.00, "profit_factor": 1.30},
    {"excluded": "TSLA", "reason": "worst", "round_trips": 43, "wins": 30, "losses": 13, "win_rate": 69.8, "expectancy": 45.00, "profit_factor": 2.10}
  ],
  "genome": [
    {"label": "N-S-K-U", "trades": 9, "wins": 8, "losses": 1, "pnl": 1201.21, "win_rate": 88.9},
    {"label": "W-S-K-U", "trades": 6, "wins": 3, "losses": 3, "pnl": -827.30, "win_rate": 50.0}
  ],
  "regime": [
    {"label": "RISK_ON", "trades": 32, "wins": 22, "losses": 10, "pnl": 3200.00, "win_rate": 68.8},
    {"label": "RISK_OFF", "trades": 6, "wins": 4, "losses": 2, "pnl": 520.78, "win_rate": 66.7}
  ],
  "spy_return_pct": 8.5
}

Key fields for analysis:

• total_trades: execution count (individual fills). round_trips: completed positions (flat→position→flat). open_positions: array of currently held positions with symbol, side, qty, avg_price, open_date, signal
• monthly_pnl, top_symbols P&L, concentration_pct, symbols_traded: computed from per-execution realized P&L (includes partial closes on open positions)
• wins, losses, win_rate, expectancy, profit_factor: all computed from round trips, not individual executions
• rolling: sliding window metrics over round trips (default window=10). shows edge stability over time
• robustness: leave-one-out analysis. excludes best/worst PnL symbols and recomputes metrics. reason is "best" or "worst"
• trading_return_pct: time-weighted return (TWR), matches IBKR
• sharpe: flow-adjusted (deposits/withdrawals subtracted before computing daily returns)
• profit_factor: gross wins / gross losses. >1.5 is good, >2 is excellent
• concentration_pct: % of absolute P&L from top symbol. >50% is risky
• pre_system_trades + post_system_trades = total round trips. pre-system = before tracked_since, post-system = after
• signal_trades + discretionary_trades = post-system. signaled = opened on a system signal, discretionary = opened without one
• order_types: dynamic dict. Keys are whatever order types appear in trades (limit, market, stop, stop_limit, trail, etc.)
• avg_hold_days / longest_hold_days: position hold duration from open→close reconstruction
• genome: trade genome classification. Each bucket has a 4-letter type code across 4 axes:
  • Trend: W (With-trend) / N (Neutral) / A (Against-trend). SMA 50/150/200 alignment at entry
  • Cadence: F (Flash <1d) / S (Swing 1-20d) / P (Position >20d). Actual hold duration
  • Setup: B (Breakout, within 1% of 50d high) / K (Pullback) / R (Reversal, within 5% of 50d low or RSI<30). Price structure at entry
  • Volume: C (Confirmed, rel_vol >1.4x) / U (Unconfirmed). Volume at entry vs 50d average
  • Example: W-S-B-C = With-trend Swing Breakout Confirmed. ? = no OHLCV data for that ticker
  • Edge = top 3 by P&L (your strengths). Leak = bottom 3 by P&L (your weaknesses)
  • Populated automatically when data/ exists in the workspace. Empty if no OHLCV data available
• regime: market regime at trade entry. 2-state: RISK_ON (default) / RISK_OFF (VIX/VIX3M > 1.10 for 7 consecutive days). Based on VIX term structure inversion, the only regime signal that survived OOS validation on 20 years of data. Requires VIX and VIX3M data in data/us-indices/_ref/
• spy_return_pct: benchmark, null if SPY data unavailable

What it computes

Performance: executions, round trips, win rate, P&L, avg win/loss, expectancy, profit factor, order type distribution Capital: start/end capital, deposits (exact USD via FXRateToBase), TWR Risk: max drawdown, drawdown duration, win/loss streaks, Sharpe ratio (flow-adjusted) Costs: trading fees, platform fees, dividends, tax, interest, fees as % of P&L Behavior: pre-system vs post-system split, post-system further split into signaled vs discretionary, hold duration Genome: 4-axis trade classification (Trend/Cadence/Setup/Volume) with edge/leak breakdown Exposure: top symbols by P&L, concentration risk, monthly P&L breakdown Robustness: leave-one-out analysis excluding best and worst performing symbols Rolling: sliding window win rate, expectancy, profit factor over round trips Benchmark: TWR vs SPY over the same period

Your role

The script computes numbers. You interpret them in context of:

• profile.md: does performance match their goals and risk tolerance?
• insights/: do the numbers confirm or contradict patterns you've observed?
• journal observations: what was the human feeling during their best and worst trades?

Writing the summary

When asked for a full summary (or at natural boundaries), write to insights/{YEAR}-summary.md. Structure:

1. The Numbers: cold facts from the compute output
2. The Patterns: what the data reveals about behavior
3. The Evolution: how the trader is changing, compared to their goals

Keep it honest. If discretionary trades underperform systematic ones, say so clearly.

Using genome data

The genome reveals the trader's behavioral DNA. Use it to:

• Edge/Leak: edge types (top P&L codes) = natural strengths. Leak types (bottom P&L) = where they bleed. Flag when they're about to enter a leak-type trade.
• Drift: compare genome across periods (--period 2026-Q1 vs --period 2026-Q2). If edge types disappear or leak types grow, the trader is drifting.
• Mirror: if the trader claims to follow a system, check if their genome matches. "You say trend-following, but 60% of your trades are A-S-R-U (against-trend reversal)."
• Anchor insights to codes: "Your N-S-K-U trades are 89% win rate, your strongest pattern. Your W-S-K-U trades are 50%, same setup in different trend context, loses edge."

When to offer

• User asks: "How am I doing?" / "Show me my stats"
• End of month/quarter/year
• After a significant event: large drawdown, big win, strategy change
• When you notice a behavioral pattern that the stats would illuminate