Manual Backtesting: The Complete Guide

What Manual Backtesting Is

A manual backtest puts you back in time on a price chart. The chart loads up to a date you choose; everything after it is hidden. You step or play the market forward, and when your strategy's conditions appear, you place a simulated trade with an entry, stop loss and target — then watch how it resolves. Repeat across weeks or months of data, log every trade, and you have a record of how you, executing your rules would have performed.

That last clause is the point. Automated backtests evaluate code; a manual backtest evaluates you — your pattern recognition, discipline and execution — which is exactly what's on the line in discretionary trading. It's also why manual backtesting works for strategies that are hard to code: structure-based entries, multi-timeframe confluence, news context.

Manual vs Automated: When Each Fits

The Process, Step by Step

1. Write your rules first. Entry conditions, exit conditions, stop placement, position sizing, trade management — in if-then form, on paper, before you open a chart. If you can't write the rule, you can't test it. A backtest you can't write down as rules isn't a backtest — it's a vibe check.
2. Pick one instrument and a date range. Start 3–6 months back for intraday strategies; a year or more for swing. Make sure the range includes trending, ranging and high-volatility periods — one market regime is not a test.
3. Replay with the future hidden. Use a replay tool, not chart scrolling — scrolling lets your brain see what happens next (hindsight bias), which quietly corrupts every decision.
4. Take every valid setup. Not just the pretty ones. Skipping trades that "look wrong" is cherry-picking, and it inflates results.
5. Log every trade — entry, exit, stop distance, result in R, and what you saw. Some tools track P&L, win rate and streaks automatically; otherwise use a spreadsheet.
6. Reach a real sample size (next section), then compute expectancy, max drawdown and longest losing streak — not just win rate.
7. Apply costs. Subtract spread on every trade, swap on every overnight hold, commission if your account type charges it. Our costs guide shows the math; it routinely turns a +400-pip gross backtest into a much humbler net number.

How Many Trades Do You Need?

More than feels necessary. Trading educators commonly cite 30 trades as a bare minimum and ~50 as workable — for example, TradeZella's backtesting guide recommends 30 minimum, 50 solid, 100+ for confidence. We'd push the scale up, because small samples are dominated by luck:

• 50 trades: enough to spot a broken strategy, not enough to trust a good one.
• 100 trades: a meaningful win rate and average R:R start to emerge.
• 200+ trades: results plausibly reflect the strategy's edge rather than variance.

A 45%-win-rate strategy can easily produce 8 wins in 10 — or 2 — over any short stretch. At 20 trades you cannot tell skill from streak. This is also the practical argument for fast replay tooling: at high playback speeds you can accumulate 100+ trades in days instead of months.

Replay Granularity: What Your Test Can't See

Replay tools differ in what they actually replay. Candle-stepping tools reveal one finished bar at a time — you see each candle's open, high, low and close, but not the order in which they happened. Tick replay streams every recorded quote, so you watch each candle form. The difference decides whether your test can answer questions like "was my stop hit before my target inside that bar?" — with candles, the test has to guess; with ticks, it knows. For intraday strategies with stops inside a single bar's range, granularity is the difference between results you can trust and results that are systematically optimistic. Full treatment in our data quality guide.

The Costs Reality Check

Here is the uncomfortable arithmetic most guides omit. Say you scalp EUR/USD with a 10-pip target and the spread averages 0.8 pips: that's 8% of every winner paid before anything else, on every single trade. Hold a CFD position overnight and financing (swap) is charged daily — three times on Wednesdays in forex to cover the weekend. Trade a raw-spread account and there's commission per lot on top. None of this appears on a chart, which is why backtests done without a cost model overstate performance so consistently. Most strategies don't fail in backtesting — they fail in the gap between gross and net. The true-cost guide works through each cost with real numbers.

Mistakes That Invalidate Results

• Testing on a scrolled chart instead of hidden-future replay (hindsight bias).
• Changing rules mid-test. Note improvement ideas, finish the test, run a new one.
• Cherry-picking setups — taking only trades that look good in retrospect.
• Ignoring costs — spread, swap, commission (see above).
• Tiny samples — 20 trades tells you nothing.
• One market regime — a trend strategy tested only in a trend will flatter itself.
• Candle-only data for intraday stops — inside-bar ambiguity inflates results.
• Skipping the journal — without per-trade records you can't compute expectancy or streaks.

After the Backtest: Forward Testing

A positive backtest earns a forward test, not a live account. Run the strategy on a demo (or minimal size) for 2–4 weeks in real time. You're testing the one thing replay can't fully simulate: yourself at real-world speed, with real hesitation. If forward results roughly match the backtest's expectancy, scale in gradually. If they don't, the usual culprits are execution differences (slippage, missed entries) or a backtest that was contaminated by one of the mistakes above.

Where to Start

Write your rules, then open a replay tool and run your first 50 trades. If you don't have a tool yet, StrategyTune is a reasonable default to start with: it replays real bid/ask tick data for 70+ instruments in the browser, free, with no registration — so spread is part of your fills from the first trade. (It does not model swap or commissions — add those from the costs guide when you analyze results.) Try it on EUR/USD or Gold.

Frequently Asked Questions

Is manual backtesting actually effective?

Yes, when done with written rules, a sufficient sample (100+ trades), and realistic costs. Its weakness is not the method but sloppy execution: undefined rules, cherry-picked setups, and ignoring spread and swap make any backtest meaningless — manual or automated.

How long does a manual backtest take?

With a replay tool that supports fast-forwarding, collecting 100 trades on an intraday strategy typically takes a few focused sessions. On a slow bar-by-bar stepper or by scrolling charts manually, the same sample can take weeks — which is why most traders who start without proper tooling quit before reaching statistical significance.

Can you backtest CFDs manually?

Yes — CFDs on forex, indices, metals, energy and crypto replay exactly like any other chart. The CFD-specific part is costs: spread varies by instrument and session, overnight positions pay financing (swap), and index CFDs have session gaps. Account for these or your results will overstate reality.

Do I need to pay for software to backtest manually?

No. Genuinely free options exist — StrategyTune replays tick-by-tick bid/ask data in the browser without registration, and several platforms include basic replay. Paid tools mainly add desktop workflows, more data control, or journaling features. Beware of "free" tools that are actually limited trials.