trade-karne

Trade-Karne — Deterministic Trading Scorecard

What This Does

Trade-Karne grades your trading operation with a single weekly score — not just "did equity go up," but did you trade well. It computes a 3-axis Karne (report card) from your equity curve and trade log:

1. Result (40%) — Risk-adjusted return: Sharpe, Sortino, Calmar, max drawdown, alpha vs benchmark
2. Growth (30%) — Realized trade quality: equity change, profit factor, win rate, expectancy, average R-multiple from actual fills
3. Process (30%) — Discipline: R:R logging rate, capability progress, lessons written — minus penalties for idle cash, blind runs, and consecutive losses

The score is pure math — zero LLM tokens, zero API calls, zero external dependencies. It reads your existing ledger (SQLite, CSV, or in-memory dicts) and produces a JSON payload and a human-readable English report.

When to Use It

Use this skill when you want to:

• Build a trading quality-control system with measurable weekly grades
• Set up anti-stall guardrails that detect "standing still" (idle cash piling up)
• Implement a weekly scorecard across multiple trader books or personas
• Run post-trade analysis with deterministic MAE/MFE and behavioral tagging
• Get an honest assessment of whether your returns are luck or skill

When NOT to Use

This skill does NOT handle:

• Live order execution or broker integration — it's read-only analysis
• Real-time risk management — it's a weekly/periodic score
• Portfolio optimization or allocation decisions — it measures, doesn't prescribe

How to Use

Quick Start (Python API)

from trade_karne import DictDataSource, load_book_input, compute_book_karne, render_report_en

ds = DictDataSource({
    "equity_series": [10000, 10100, 10250, 10400],
    "cash": 1000,
    "hwm": 10400,
    "trade_pnls": [120, -50, 200],
    "trade_rs": [2.0, -1.0, 2.5],
})
inp = load_book_input(ds, "momentum", lessons_count=3)
karne = compute_book_karne(inp)
print(render_report_en(karne, [karne], week="2026-W26", generated_at="today"))

CLI

trade-karne score --db trades.db --book momentum
trade-karne postmortem --db trades.db --book momentum --out-dir ./reports
trade-karne metrics --list

Plug Into Your Existing System

Implement TradeDataSource (7 methods) and pass it to load_book_input(). The Karne never touches your files or DB directly — it only calls those 7 methods. See workflows/integration.md for a step-by-step guide.

Validation

After running a score:

1. Verify the Process axis — if it's below 40, idle cash or blind runs are dragging you down
2. Check the rolling Sharpe trends — a declining trend means skill is degrading
3. Run the postmortem and look for tag clusters (e.g., many premature_entry tags mean entries need tightening)
4. Compare Result axis across books — large variance means some personas need rebalancing or rethinking