18.2
Crypto ANN
PYTHON PLUGIN
Crypto
Machine Learning
Description
Strategy 18.2: ANN proxy using EMA crossovers + RSI + vol regime for crypto
Strategy Logic
Strategy 18.2: ANN-based BTC Price Prediction (proxy implementation).
The original strategy uses a neural network trained on normalised returns,
EMA (30m, 1h, 3h, 6h), EMSD (same windows), and RSI (3h, 6h, 12h) to
predict K quantiles of future returns.
Since actual NN training requires separate infrastructure, this
implementation uses a composite signal from EMA crossovers, RSI regimes,
and volatility as a proxy for the ANN output.
Features computed (daily bar equivalents):
- EMA crossovers at periods 2, 5, 12, 24 (proxy for 30m/1h/3h/6h)
- Exponential moving std-dev at same periods
- RSI at periods 12, 24, 48 (proxy for 3h/6h/12h)
Parameters
| Parameter | Default Value | Type |
|---|---|---|
| ema_fast | 5 | int |
| ema_slow | 24 | int |
| ema_mid | 12 | int |
| rsi_period_short | 12 | int |
| rsi_period_long | 48 | int |
| rsi_overbought | 70 | int |
| rsi_oversold | 30 | int |
| vol_lookback | 20 | int |
| signal_threshold | 0.3 | float |
| quantile_k | 5 | int |
Risk Configuration
| Risk Parameter | Value |
|---|---|
| Max Position Pct | 10.0% |
| Stop Loss Pct | 8.0% |
| Take Profit Pct | 15.0% |
| Max Drawdown Pct | 20.0% |