Here are some early observations after 3 weeks of forward-testing of the Candela North trading system algorithm:
- closed trades - early performance is weak: 28% win rate from 19 trades vs ~66% in backtests. The sample size is small, so the results are not yet conclusive
- active trades - currently +1.2R across 39 open positions, with the portfolio positioned on the long side.
Current exposure is heavily long, so outcomes will depend on broader market direction over the next period. The backtest shows that 3-4 weeks of drawdowns are common for this model, so this phase is within expected variance.
Positive takeaway - trade management behaves as designed with taking partial profits and adjusting stop losses consistently.
Next steps:
- allow open trades to develop and evaluate the outcomes
- narrow down the scope of crypto assets to the top 30 by market capitalization to reduce the noise and focus on more liquid assets
- change provider for the stock klines, rebuild the model and restart forward testing for equities