manifoldbt-backtester

1# manifoldbt Backtester Skill
2 
3## Purpose
4 
5Execute what `backtest-expert` teaches. That skill grades a backtest on five
6dimensions, and its prerequisites say "metrics are user-provided": it scores
7numbers it never produces. This skill produces them. It runs a strategy over
8real bars and returns the eight inputs its evaluator asks for.
9 
10The two chain in one direction: spec, run, evaluate.
11 
12## When to Use This Skill
13 
14- A user describes a rule and wants it measured
15- `backtest-expert` is about to run and the numbers do not exist yet
16- A win rate, average winner, average loser or drawdown must come from bars
17- A strategy's parameter count must be established for scoring
18 
19Leave the verdict to `backtest-expert`. It owns the thresholds and the red
20flags, and this skill does not duplicate them.
21 
22## Prerequisites
23 
24- Python 3.9+
25- `pip install manifoldbt` (the Python package is Apache 2.0, the compiled
26  engine ships under the free Manifold-BT Engine License; the free tier covers
27  everything this skill does)
28- OHLCV bars as CSV or Parquet with columns `timestamp, open, high, low, close, volume`
29- No API key required
30 
31## Workflow
32 
33### 1. Write the strategy spec
34 
35A spec names indicators and one entry condition. Keep it to the smallest rule
36that states the hypothesis. Every added knob makes an in-sample fit easier to
37reach by accident, and the evaluator penalises the count.
38 
39```json
40{
41  "name": "sma_cross_costed",
42  "indicators": {
43    "fast": { "type": "sma", "period": 20 },
44…