How to Read AI Analysis Report: Understanding Strategy Insights

Requesting the Analysis

Open any completed backtest and click “AI Analysis.” The initial analysis takes 15–30 seconds as Claude Opus 4.5 processes your full trade history, equity curve data, and the underlying market conditions. The result appears as a structured report with multiple sections. You can then ask follow-up questions in the chat for deeper exploration.

Key Sections of the AI Report Explained

Market Regime Classification

The AI classifies market conditions into regimes: trending (bullish/bearish), ranging, high-volatility, and transitional. It shows what percentage of your trades occurred in each regime and how performance differed. If 70% of your profits came from a single bull-trend period, the regime section makes this dependency visible.

Performance Patterns

Statistical analysis of winning and losing trades. Do winners cluster at certain times or conditions? The AI examines trade duration distributions, time-of-day patterns, and sequential dependencies. These patterns often reveal that a strategy's edge exists only in specific conditions — knowledge you can use to add filters.

Event Study

For each trade, the AI examines what the market did before entry (pre-event) and after exit (post-event). This reveals timing insights: are you entering too early, too late, or at the right moment? Event study data is presented as aggregated patterns, not individual recommendations.

Risk Assessment

Analysis of drawdown periods, consecutive losses, and risk-adjusted metrics. The AI identifies the worst historical period and what market conditions caused it — essential for setting realistic expectations and proper position sizing.

How AI Identifies Strategy Weaknesses

The AI looks for structural patterns that surface hidden vulnerabilities:

Regime dependency. If your strategy shows 1.8 profit factor during uptrends but 0.7 during ranging periods, the AI highlights this divergence. The overall 1.3 PF looks acceptable, but the regime breakdown reveals profitability depends entirely on trend continuation.

Timing inefficiency. The event study might show that price initially moves against your entries for 2–3 bars before turning profitable — suggesting entry conditions fire slightly too early.

Drawdown clustering. The risk assessment examines whether largest drawdowns occurred during extraordinary events or ordinary conditions. Drawdowns during normal trading indicate a systematic weakness in strategy logic.

Trade duration anomalies. If winning trades average 12 bars but losing trades average 45 bars, your strategy holds losers far too long — invisible in summary statistics but devastating to equity curve smoothness.

What the AI Will NOT Say

The AI provides observations and data. You interpret and decide. This is by design — regulatory compliance requires that AI tools don't provide trading advice.

Actionable Follow-Up Steps After Reading the Report

The report is most valuable when it drives concrete next steps:

Step 1 — Identify the weakest regime. Find the market condition where your strategy underperforms. If it loses money during ranging markets, consider adding a trend strength filter (ADX above a threshold) to skip those periods.

Step 2 — Examine event study for timing clues. If price moves against your entry for the first few bars, experiment with a confirmation candle. If price continues in your favor after exit, consider a trailing stop or wider take profit.

Step 3 — Use follow-up chat for deep dives. Ask specific questions: “What happened during the largest drawdown period?” The follow-up chat uses Sonnet 4 for fast responses and maintains full context from the initial analysis.

Step 4 — Modify and re-test. Make one change at a time based on the report's findings. Run a new backtest, then request a new AI analysis to compare.

Limitations of AI Analysis

Historical data only. The AI analyzes what happened, not what will happen. Market regimes shift, correlations break, and past patterns may not repeat.

Pattern does not equal causation. When the AI identifies that winning trades correlated with high volume, this is a statistical observation. Testing the hypothesis (adding a volume filter) is the only way to determine if the relationship is genuine.

Single-strategy scope. The analysis examines one strategy in isolation. It doesn't consider portfolio interactions or how adding this strategy to your existing trading would affect overall risk.

FAQ