5f7702469e
- Removed v10 TradingView indicator (moneyline_v10_momentum_dots.pinescript) - Removed v10 penalty system from signal-quality.ts (-30/-25 point penalties) - Removed backtest result files (sweep_*.csv) - Updated copilot-instructions.md to remove v10 references - Simplified direction-specific quality thresholds (LONG 90+, SHORT 80+) Rationale: - 1,944 parameter combinations tested in backtest - All top results IDENTICAL (568 trades, $498 P&L, 61.09% WR) - Momentum parameters had ZERO impact on trade selection - Profit factor 1.027 too low (barely profitable after fees) - Max drawdown -$1,270 vs +$498 profit = terrible risk-reward - v10 penalties were blocking good trades (bug: applied to wrong positions) Keeping v9 as production system - simpler, proven, effective.
65 lines
3.8 KiB
Markdown
65 lines
3.8 KiB
Markdown
# Backtester Status (Dec 2025)
|
||
|
||
## Current Scope
|
||
- Python package under `backtester/` (isolated from trading bot)
|
||
- CSV loader (`data_loader.py`) validates OHLCV inputs with timestamp filtering
|
||
- Math utilities (`math_utils.py`) implement RMA, ATR, ADX helpers
|
||
- Money Line indicator module (`indicators/money_line.py`) provides:
|
||
- Configurable inputs for v9/v10 parity (`MoneyLineInputs` dataclass)
|
||
- Primary + momentum signal detection (flip threshold, MA gap, ADX/volume filters)
|
||
- Output `MoneyLineSignal` records with full metric context
|
||
- Strategy simulator (`simulator.py`) that consumes OHLCV data + indicator signals and reproduces the ATR-based TP1/TP2 runner logic (including ADX runner stops, trailing stop, MAE/MFE tracking, and placeholder quality filters)
|
||
- Command-line runner (`cli.py`) to execute quick tests against CSV data, print performance stats, and export trade logs
|
||
- Binance OHLC exporter (`scripts/export_binance_ohlcv.py`) for refreshing SOL/USDT candle sets straight from the public REST API
|
||
|
||
## Next Steps
|
||
1. Wire production quality scoring + adaptive leverage gating into the simulator so backtests align with live trade filters
|
||
2. Build batching utilities for multi-symbol / multi-parameter sweeps and persist the results to CSV/Parquet for analysis
|
||
3. Compare Binance-derived backtests vs. Drift trade history to calibrate slippage/latency assumptions
|
||
4. Extend CLI to accept saved indicator scores from the database once we export them (blocked signals, 1m data feed)
|
||
|
||
## Simulator Plan
|
||
- **Core engine (`backtester/simulator.py`)**: iterate through OHLCV rows, feed them to the Money Line indicator, and execute trades using the same ATR TP1/TP2 runner + adaptive leverage stack used in production.
|
||
- **Position model**: store entry price, size, ATR snapshot, TP1/TP2 levels, runner state, and MAE/MFE tracking so analytics match the bot’s database fields.
|
||
- **Execution hooks**: reuse config helpers (quality scoring thresholds, adaptive leverage tiers) to determine whether signals become trades, ensuring apples-to-apples results for v9 vs v10.
|
||
- **Metrics output**: aggregate PnL, win rate, profit factor, MAE/MFE distributions, and per-parameter summaries so sweeps can compare settings quickly.
|
||
- **Parameter sweeps**: now that the simulator is deterministic, the remaining work is orchestration (CLI wrapper + reporting) so we can explore grids quickly.
|
||
|
||
## Data Acquisition
|
||
- Use the helper script to pull candles (example grabs ~4 months of SOL/USDT 5m bars):
|
||
|
||
```bash
|
||
source .venv/bin/activate
|
||
python scripts/export_binance_ohlcv.py \
|
||
--symbol SOLUSDT \
|
||
--interval 5m \
|
||
--start 2025-08-01 \
|
||
--end 2025-11-28 \
|
||
--output data/solusdt_5m.csv
|
||
```
|
||
|
||
- The script paginates Binance’s REST API (1000-candle chunks), pauses between requests to respect rate limits, and writes timestamped OHLCV columns that `backtester.data_loader` already expects.
|
||
- Re-run the export whenever we want fresh data; adjust `--start/--end` to widen the window or target different regimes.
|
||
|
||
## CLI Usage
|
||
Run a standalone backtest from the project root:
|
||
|
||
```bash
|
||
python -m backtester.cli \
|
||
--csv data/solusdt_5m.csv \
|
||
--symbol SOL-PERP \
|
||
--timeframe 5 \
|
||
--start 2025-09-01 \
|
||
--end 2025-09-30 \
|
||
--position-size 2500 \
|
||
--max-bars 300 \
|
||
--export-trades results/sol_sep.csv
|
||
```
|
||
|
||
The script uses the CSV loader (timestamped OHLCV data), Money Line indicator defaults, and simulator to print summary metrics. Exported CSVs contain per-trade fields (entry/exit, PnL, TP flags, MAE/MFE) for further analysis.
|
||
|
||
## Notes
|
||
- Package intentionally avoids dependencies beyond pandas/numpy for portability
|
||
- No trading bot files modified; all work lives under `backtester/`
|
||
- Update this document as modules (scoring, simulator, reporting) come online; immediate next edits should cover quality filters + sweep tooling once implemented.
|