How to Backtest Using Your Own Trading History

The Concept: Filter, Don't Simulate

You have 180 trades in your journal from the last 3 months. Those trades are facts — real entries, real exits, real P&L. Now ask:

• What if I remove the 35 trades that had no tagged setup?
• What if I remove the 28 trades taken after 14:00 GMT?
• What if I remove the 15 trades taken on Fridays?
• What if I remove trades where I risked more than 1.5%?

Each filter produces a new equity curve. Compare it to your original curve. If the filtered version is better — the trades you removed were hurting you. If it's worse — those trades were actually helping.

This is the equity curve comparison applied as a backtesting methodology.

Six Filters That Reveal the Most

Filter 1: Best Setup Only

Question: "What if I only traded my highest-PF setup?"

How: Remove all trades except your top setup by profit factor (minimum 30 trades for reliability).

What traders often find: Total P&L can improve significantly — sometimes 30% or more. Your best setup carries the portfolio; everything else dilutes it.

Action if better: Gradually shift toward trading only this setup. Don't go cold turkey — start by reducing B/C setups to 50% size for a month.

Filter 2: Best Session Only

Question: "What if I only traded my most profitable session?"

How: Remove all trades outside your best 3-4 hour window. See session analysis.

What traders often find: Fewer trades, but P&L stays similar or improves. The extra sessions were adding volume without adding edge.

Action if better: Set a hard stop time. Trade your best session and do journal review / analysis during the rest.

Filter 3: Remove Revenge Trades

Question: "What if I never revenge traded?"

How: Remove trades tagged as revenge, or trades entered within 10 minutes of a loss with no setup tag.

What traders often find: P&L improves noticeably once revenge trades are removed. Revenge trades often show profit factors well below 1.0 in journal data — removing them tends to improve overall performance.

Action if better: Implement the 10-minute cooldown rule.

Filter 4: Cap Daily Trades

Question: "What if I stopped after 4 trades per day?"

How: For each day, keep only the first N trades (chronologically). Remove trades 5, 6, 7+.

What traders often find: Early trades tend to have higher profit factors than later ones. Removing the excess improves monthly P&L and dramatically smooths the equity curve.

Action if better: Set a hard trade count limit per day. See overtrading analysis.

Filter 5: Remove Worst Day of Week

Question: "What if I never traded Fridays?"

How: Remove all trades on your worst day (identified from day-of-week breakdown).

What traders often find: Removing the worst day of the week often improves P&L and reduces drawdown — the exact impact depends on your data.

Action if better: Skip that day entirely for one month. Measure the real impact vs the backtested impact.

Filter 6: Remove Oversized Trades

Question: "What if I never risked more than 1% per trade?"

How: Remove or cap all trades where position size exceeded 1% of account.

What traders often find: Oversized trades tend to have worse win rates (emotional sizing) and bigger losses. Removing them reduces volatility dramatically while P&L often stays flat or improves.

Action if better: Lock your position size at 1% with no exceptions — use the position calculator to verify before every trade. Mechanical sizing removes the temptation.

Step-by-Step Process

Step 1: Export Your Trade Data

You need at minimum: entry time, exit time, direction, instrument, P&L, and ideally setup tag and position size. Export from your broker, MT4/MT5, exchange, or journal. The more fields, the more filters you can test.

Step 2: Calculate Your Baseline

Run the full dataset unfiltered. Note:

• Total P&L
• Win rate
• Profit factor
• Max drawdown
• Equity curve shape

This is your "before" — the benchmark every filter will be compared against.

Step 3: Apply One Filter at a Time

Don't stack filters. Test one variable at a time so you know which change drives the improvement. If you remove Fridays AND revenge trades simultaneously and P&L improves, you don't know which change caused it.

Step 4: Compare the Numbers

60% fewer trades, 75% more money, 51% less drawdown. The numbers make the case clearly.

Step 5: Overlay the Equity Curves

Numbers tell the story. The equity curve overlay makes it visceral. When you see the filtered (green) line climbing smoothly while the total (gray) line barely moves — you can't unsee it. The visual is more motivating than any spreadsheet.

Step 6: Implement ONE Change

Pick the single filter that showed the biggest improvement. Implement it for 30 days. Don't change anything else. After 30 days, re-run the backtest with the new month's data and compare.

Advanced: Multi-Variable Backtests

Once you've tested individual filters, you can combine the best ones:

1. Start with your best setup filter (biggest single improvement)
2. Add session filter (second biggest)
3. Add trade count cap (third biggest)
4. Compare the triple-filtered curve to your original

The combined impact can be larger than any single filter — but be careful of overfitting. If your triple-filtered dataset has fewer than 30 trades, the results might not be reliable. Keep each filtered group at 20+ trades minimum.

Timeline Backtesting: The Replay

A different approach: instead of filtering by variables, replay your trades chronologically and ask different questions at each point:

• Day 5: You're up $400. What if you reduced size by 25% here to protect gains?
• Day 12: You had a -$300 day. What if you'd stopped after the first loss (-$120)?
• Day 18: You switched from EUR/USD to GBP/JPY. What if you hadn't?

Timeline backtesting is more subjective but helps identify decision points where different choices would have changed outcomes. It's especially powerful for understanding your tilt patterns — you can see exactly where discipline broke down and what it cost.

Common Backtesting Mistakes

• Cherry-picking the best filter to confirm a bias. Test ALL reasonable filters, not just the one you want to see. The data might disagree with your assumption about what's holding you back.
• Overfitting to small samples. 15 trades in a filter group is noise. 30+ starts to become meaningful. Don't make career-changing decisions on 12 data points.
• Stacking too many filters. Filter → setup + session + day + direction + time = maybe 5 trades left. That's not a backtest, that's an anecdote. Keep it to 1-2 filters max.
• Ignoring the trade count reduction. A filter that improves P&L by 50% but removes 80% of your trades might leave you with too few trades to make a living. Check that the remaining trade count is viable.
• Backtesting once and never again. Markets change. Your trading changes. Re-run backtests monthly to ensure your filters are still valid.

The Bottom Line

Hypothetical backtests have their place, but they're not the only option. You need to understand what your EXISTING trades are telling you — which ones help and which ones hurt. Filter-based backtesting on your own journal data is a practical, low-bias way to answer that question. For more on backtesting methodology, see Investopedia's backtesting guide.

The data is already there. You just need to ask the right what-if questions. Start with the six filters above, find the one that makes the biggest difference, and implement it. One filter, one month, one re-evaluation. Repeat until your equity curve looks like a staircase.