Our Trading Algorithm: 2-Week Performance Review (The Good, Bad & Ugly)

Executive Summary

We launched our dual-signal trading algorithm on November 21, 2024. After 2 weeks and 74 closed trades, here's what the data shows:

✅ 68.9% win rate (51 wins, 23 losses)
✅ +1.33% expectancy per trade
✅ Technical patterns outperforming AI predictions (78.4% vs 59.5%)
⚠️ Small sample size - need 200+ trades for statistical confidence
⚠️ Only tested in bull market conditions

Performance Dashboard

Real-time performance metrics from our DynamoDB tracking system

✅ The Good

1. Strong Win Rate (68.9%)

Our algorithm is winning more than 2 out of 3 trades. This puts us in the top 20% of retail trading systems and competitive with professional algorithmic traders (typically 55-70% win rate).

2. Positive Expectancy (+1.33%)

The math works: (68.9% × 3.19%) + (31.1% × -2.79%) = +1.33%

This means on average, we make 1.33% per trade. Over 100 trades, that's 133% cumulative return (assuming equal position sizing and no compounding).

3. Technical Patterns Crushing It

Our technical pattern scanner is delivering a 78.4% win rate (29/37 trades). Classic chart patterns like consolidation breakouts and MA pullbacks are proving more reliable than our AI ensemble.

4. Tight Risk Management Working

Our 3% stop loss is doing its job: 65.2% of losses hit the stop, preventing larger drawdowns. The average loss is only -2.79%, while average wins are +3.19% - a favorable 1.14:1 reward-to-risk ratio.

⚠️ The Bad

1. AI Predictions Underperforming

Our ensemble ML model (combining 4 algorithms) is only hitting 59.5% win rate (22/37 trades). While still profitable, it's significantly behind technical patterns.

Why? AI predictions use fixed ±3% targets/stops, while technical patterns use measured moves based on chart structure. The AI needs dynamic target calculation.

2. MA20 Pullback Pattern Failing

The "MA20 Pullback" pattern has a dismal 20% win rate (1/5 trades). This classic "buy the dip" strategy isn't working in current market conditions.

Action: We're disabling this pattern until we can refine the entry criteria.

3. Small Sample Size

74 trades is not statistically significant. We need 200+ trades to be 95% confident these results aren't due to luck. Current results could be within normal variance.

❌ The Ugly

1. Return Distribution Shows Asymmetry

Our wins cluster around +3% (hitting targets), but losses are more spread out. Some trades are getting stopped out at -3%, while others drift lower before closing. This suggests our stop placement could be optimized.

2. No Bear Market Testing

We launched during a bull market rally. We have zero data on how the system performs during:

Market corrections (-10%+)
Bear markets (-20%+)
High volatility (VIX > 30)
Choppy, sideways markets

Risk: The system could fail spectacularly when market conditions change.

3. Execution Slippage Not Tracked

Our returns assume perfect execution at signal prices. In reality, you'd face:

Bid-ask spread: ~0.05-0.10%
Market impact: ~0.05-0.15% (depending on size)
Timing delays: Signals generated after close, executed next day

Real-world returns could be 0.2-0.4% worse per trade.

Equity Curve

Cumulative returns show steady growth with manageable drawdowns. The largest drawdown was approximately -8% (around trade 15-20), which recovered within 5 trades.

Key observation: No catastrophic losses. The 3% stop loss is preventing blow-ups.

🔧 What We're Improving

1. Testing 4% Stop Loss

Backtest shows 4% stops would have saved 50% of losing trades (55% improvement). We're testing this on new signals to see if it improves real-world performance.

2. Dynamic AI Targets

Instead of fixed ±3%, AI predictions will use support/resistance levels from technical analysis to set more realistic targets.

3. RSI Filter Refinement

Current filter rejects RSI > 60. Analysis shows winning trades average RSI of 47 vs losers at 52. We're testing RSI < 55 as a stricter threshold.

4. Confirmation Count Weighting

Signals that repeat multiple days (3x, 5x confirmed) should get higher position sizing. We're building a confidence-weighted allocation system.

Our Transparency Commitment

Unlike most trading services that cherry-pick winners or only show backtests, we're tracking every single trade in real-time in our DynamoDB database.

You can see our live performance on the Performance Page. Every win, every loss, every pattern - fully transparent.

We'll publish monthly performance reviews like this one, showing what's working and what we're fixing. No hype, just data.

Conclusion

After 2 weeks and 74 trades, our algorithm is showing promising but early results:

✅ 68.9% win rate is competitive with professional systems
✅ Technical patterns are delivering strong performance
✅ Risk management is preventing catastrophic losses
⚠️ Need more data to prove long-term viability
⚠️ Untested in bear markets or high volatility

Grade: B+ - Good start, but too early to declare victory. We're cautiously optimistic and committed to continuous improvement based on real data.

Check back next month for the 6-week review with 150+ trades.

Want Daily Signals?

Get our trading signals delivered to your inbox every morning before market open.

View Pricing →