Backtesting and Journaling

Backtesting means walking through historical charts one bar at a time — as if you are living through that day — and taking every setup your plan describes, with the exact entry, stop, and target rules you wrote in TF-107. No skipping the ugly ones. No "I would have known." If your written rules say take it, you record it.

Sample size is everything. Twenty trades proves nothing — anyone can get lucky twenty times. You want at least 100 trades across different market conditions: uptrends, downtrends, choppy stretches, calm and volatile years. A plan that only works in a roaring bull market is a plan that will betray you the moment the market changes. If your setup is rare and 100 trades takes too long to collect, the honest answer is that you cannot validate it yet — pick a setup that occurs often enough to test.

Be brutally mechanical about it. For each trade record the date, the setup, entry, stop, target, and result. Spreadsheets work fine; specialized replay software works better because it hides the future from you (charting platforms show you the whole chart, which quietly tempts you to "know" what came next). The goal is a clean record of what your rules — not your hindsight — would have produced.

Here is where most backtests lie to their owners: they forget costs. Every trade pays the spread (TF-105) — the gap between bid and ask — plus any commission your broker charges. On paper, a strategy that wins $50 per trade on average looks great. Subtract a $5 spread cost and a $2 commission each way, and the real edge shrinks or vanishes entirely.

Always compute results net of costs: subtract the spread and commissions from every single trade in your test. This matters most for active strategies — day trading and scalping — where dozens of small costs compound into a headwind that can exceed the strategy's edge. A plan that is profitable before costs but unprofitable after is not a trading plan. It is a donation plan.

Be pessimistic on purpose. Assume slightly worse fills than the perfect chart price (this is slippage — your order filling a little worse than expected, covered in TF-105). If the plan survives pessimistic costs, it is robust. If it only works with perfect fills and zero commissions, it does not work.

There is a classic trap called curve-fitting (or overfitting): tweaking your rules until they perfectly explain the past. Change the moving average from 20 to 23 days because it backtests better, drop the losing months as "unusual," add a filter for every bad trade — and you will eventually produce a backtest that looks flawless and a plan that fails on the first live trade. You did not discover an edge. You memorized the answers to last year's test.

The defense is out-of-sample testing: split your data. Develop and tune your rules on one period (say, 2020–2023), then lock the rules and test them on a period you never touched during development (2024–2025). If performance collapses out-of-sample, your rules were fitted to noise — go back and simplify. Simpler rules survive; complicated rules memorize. A plan with three clear rules that works modestly in both samples beats a plan with twelve optimized rules that works perfectly in one.

Backtesting validates the plan. Journaling validates you — whether you actually follow it. After every live or simulated trade, log: the date, the setup tag (which setup was this?), entry, stop, target, result, a chart screenshot marked with your lines, and one line on how you felt and whether you followed the plan. Thirty seconds per trade. No essays.

The screenshot is the journal's superpower. Six months later you will not remember what the chart looked like — but the screenshot shows you exactly what you saw, and whether your entry trigger was really there or something you imagined. Tag every trade by setup so you can later ask: which setup actually makes money, and which one just feels good?

Review the journal weekly with three metrics:

• Win rate — what percentage of trades profit. (Alone, this means little — see below.)
• Average win vs. average loss — are your winners bigger than your losers? A 40% win rate with winners twice the size of losers is a profitable trader.
• Plan compliance — what percentage of trades followed every rule? This is the number that predicts your future. A profitable plan traded sloppily loses; a mediocre plan traded with discipline can survive.

Journaling feels like homework because it is homework — and it is the homework that separates people who improve from people who repeat the same expensive mistakes for years. The market charges tuition either way. The journal is how you make sure you actually get the education.

Tested plan, honest journal — now comes the hardest part: you. TF-109 "Trading Psychology" covers why your brain is badly designed for trading, how to judge yourself on process instead of outcomes, and the routines that keep emotions from spending your money.