How to Analyze a Trade Log From Your Backtest

The summary metrics of a backtest — net profit, Sharpe ratio, win rate — are the trailer. The trade log is the full movie. I've found bugs, discovered hidden edges, and killed strategies that looked profitable just by spending thirty minutes studying the raw trade-by-trade data.

Most traders glance at the equity curve, check if the number is green, and move on. That's like reading only the headline of a research paper and calling it due diligence. The trade log tells you why your strategy works (or doesn't), when it works best, and where its vulnerabilities hide.

What a Trade Log Contains

A proper backtest trade log includes these fields for every trade:

Time-Based Analysis

Sort your trades by entry time and look for patterns. Here's what I typically check:

Day of week performance. Many strategies have a day-of-week bias. Mondays often see gap fills and positioning. Fridays see profit-taking before weekends. I've seen strategies where removing Friday trades improved overall performance by 15% — Friday generated activity but no edge.

Session performance. For 24/7 markets like crypto, check performance by 4-hour blocks. Asian session (00:00-08:00 UTC) typically has lower volatility. London open (08:00-12:00) sees increased movement. A strategy that generates losses between midnight and 6 AM should probably sleep during those hours.

Monthly and seasonal patterns. Crypto has historically performed differently in Q1 (often strong) versus Q3 (often weak). If your strategy's profits are concentrated in a single quarter, you might be capturing a seasonal pattern rather than a persistent edge.

Holding Period Analysis

Plot trade outcome (profit/loss) against holding duration. This reveals whether your strategy works better as a quick scalp or a longer hold.

Common patterns I see:

• Best trades are the shortest. This suggests the initial signal is strong but the strategy stays in too long, giving back profits. Consider tightening take-profit or adding a time-based exit.
• Best trades are the longest. The strategy catches trends well but exits too many trades early. Consider widening take-profit or using a trailing stop instead of a fixed target.
• No correlation between duration and outcome. Holding period doesn't affect results — the edge is entirely in the entry signal, not the exit mechanism. Common with mean-reversion strategies where the entry level determines profitability.

Win/Loss Streak Analysis

Count your consecutive win and loss streaks. Compare them to what random chance would produce at your win rate.

At a 55% win rate over 200 trades, you'd expect approximately:

• Maximum winning streak: 8-10 trades
• Maximum losing streak: 6-8 trades
• Average streak length: ~2.2 trades

If your actual streaks are significantly longer than expected, your trades aren't independent — there's serial correlation. This could mean your strategy performs in clusters (which is exploitable) or that market conditions drive performance (which requires regime filtering).

If consecutive losses cluster around specific dates, overlay those dates on a market chart. You'll often find they correspond to specific events — FOMC meetings, earnings seasons, flash crashes. This tells you when your strategy is vulnerable.

P&L Distribution Analysis

Plot a histogram of trade P&L values. A healthy strategy shows a roughly bell-shaped distribution shifted slightly to the right (positive expected value). Watch for these warning signs:

Fat left tail. Most trades are small winners, but there are a few catastrophic losses that pull the average down. This is the signature of a strategy that picks up pennies in front of a steamroller — high win rate but vulnerable to tail events.

Bimodal distribution. Two clusters of outcomes rather than one continuous distribution. This often means the strategy behaves differently in two market regimes and should be split into two separate strategies.

Outlier dependency. Remove the top 5 winning trades. Is the strategy still profitable? If not, your edge is concentrated in a few outliers, making it unreliable. Robust strategies remain profitable even after removing the best and worst 5% of trades.

"The trade log is your strategy's autobiography. Read it carefully and you'll discover things the summary statistics intentionally hide." — From my internal training materials at a prop firm

Exit Reason Analysis

If your backtesting platform tracks why each trade closed (stop loss, take profit, signal reversal, time-based exit), this data is gold.

Calculate the average P&L for each exit type:

If stop-loss exits have much larger losses than the strategy's intended risk per trade, slippage or gaps are eating your stops. If take-profit exits are consistently followed by continued price movement in your direction, your targets are too conservative.

The trade log pairs well with understanding backtest metrics and equity curve analysis. Together they give you a three-dimensional view of strategy behavior.