ETH-only Freqtrade futures strategy based on a double moving-average compression breakout system.
This repository is for research and backtesting only. It is not investment advice and is not a live-trading guarantee.
- Pair:
ETH/USDT:USDT - Main timeframe:
1h - Higher timeframe filter:
4h - Direction: long and short
- Exposure: effective
2x - Max open trades:
1 - Entry idea: 5/10/30 MA + EMA compression, breakout, then first pullback confirmation
- Stop idea: planned structure stop through
custom_stoploss - Take-profit idea: structure target, minimum
2.3R, preferred3.5R - Trailing: ordinary trades off; strong 4h trend only, starts after
+5%
Red = strategy, light blue = ETH buy-and-hold, light orange = BTC buy-and-hold.
Backtest uses Binance USD-M futures 1h data with an OKX futures adapter config.
| Period | Trades | CAGR | Total Profit | Winrate | Profit Factor | Max DD | Avg Win | Avg Loss | Payoff |
|---|---|---|---|---|---|---|---|---|---|
| 2021-05-11 to 2025-01-01 | 215 | 21.46% | 100.84% | 34.88% | 1.345 | 17.72% | 5.24% | 2.09% | 2.51 |
| 2025-01-01 to 2026-05-11 | 76 | 36.94% | 53.04% | 39.47% | 1.591 | 18.54% | 4.77% | 1.95% | 2.44 |
| 2021-05-11 to 2026-05-11 | 291 | 20.78% | 154.24% | 36.08% | 1.404 | 13.51% | 5.11% | 2.05% | 2.49 |
Same period, normalized to 10,000 USDT.
| Curve | Final Equity | Total Return | CAGR | Max DD |
|---|---|---|---|---|
| Strategy | 25,424.16 | 154.24% | 20.80% | -17.72% |
| ETH buy-and-hold | 9,204.63 | -7.95% | -1.66% | -81.38% |
| BTC buy-and-hold | 22,710.60 | 127.11% | 18.07% | -77.24% |
The stress test generates fat-tailed, volatility-clustered, calendar-seasonal, jump-risk, BTC/ETH bootstrap, and black-swan synthetic 1h paths.
Main takeaway: the simpler stable strategy is more robust than the optional trend-continuation variant in stress tests.
- Report:
reports/futures_backtests/MONTE_CARLO_STRESS_TEST.md - Script:
scripts/monte_carlo_stress_test.py
- Strategy:
user_data/strategies/VideoDoubleMaFuturesStrategy.py - Config:
config/config_okx_futures_eth_only_binance_data.json - Backtest summary:
reports/futures_backtests/eth_2x_false_breakout_refinement_results.json - Full note:
reports/futures_backtests/ETH_2X_FALSE_BREAKOUT_REFINEMENT.md
conda create -n env_freqtrade python=3.12 -y
conda activate env_freqtrade
pip install freqtradeOptional plotting/data dependencies:
pip install pandas pyarrowFrom this folder:
freqtrade backtesting `
--userdir user_data `
-c config\config_okx_futures_eth_only_binance_data.json `
--datadir user_data\data\binance `
--strategy VideoDoubleMaFuturesStrategy `
--timeframe 1h `
--timerange 20210511-20260511 `
--export trades `
--cache noneThe current version improved the false-breakout problem by converting the planned structural stop into a real Freqtrade custom_stoploss. This reduced delayed body_line_invalid exits and raised realized payoff to about 2.5R.