"I need to backtest" — a phrase every trader hears from week 1. But almost no one explains which type of backtest to run. There's a huge difference between simulating the strategy automatically over 10 years of data and manually replaying how you would have traded candle by candle. And both have their place.
This guide explains what each does, when it makes sense to use it, and why you probably need both.
What automated backtesting is
You write the strategy rules in code (Python, Pine Script, MQL, etc.), the software runs the rules on years of historical data in seconds, and returns statistics: win rate, profit factor, drawdown, equity curve.
Example: "Go long when RSI(14) is under 30 and EMA(20) is above EMA(50). Stop loss 1.5 ATR. Take profit 2:1 R/R." The software finds every moment in which the condition occurred over the last 10 years and simulates the exact outcome.
Pros of automated backtesting
- Speed: 10 years of data in 30 seconds. You can test 50 variations of the strategy in an afternoon.
- Objectivity: zero emotional bias. The rule fires or it doesn't, no "it looked clear to me" or "this time I'll ignore the signal".
- Statistical edge: with thousands of trades you have solid numbers (statistical significance). You really know whether the strategy has an edge.
- Parameter optimization: you compare RSI 14 vs 21 vs 7 in a few clicks.
Cons of automated backtesting
- Overfit (problem #1): you tweak parameters until the curve looks perfect, but in live it doesn't work. If your backtest shows 75% win rate, it's almost always overfit.
- Doesn't see price action: the rule triggers on a numeric value but ignores context (e.g. news, gaps, ranging vs trending market).
- Doesn't train the trader: you never had to press "buy" or "sell". When you get to live, you don't have the discipline built up.
- Idealized execution: assumes perfect fill, no slippage, constant spread. Reality is worse.
What manual backtesting is
You move the chart forward candle by candle (or hour by hour, or tick by tick). At each new bar you decide: do I open? Close? Move SL? You are the trader, the system only simulates time flow and calculates P&L.
Example: you start on EUR/USD 15m on March 1, 2024. You advance bar by bar. You see a setup like sweep + engulfing → press BUY → set SL and TP → advance until it closes → log it in the journal. Repeat for 200 trades.
Pros of manual backtesting
- Trains discipline: you get used to seeing the setup, deciding, executing. When you get to live, your hands know what to do.
- Real pattern recognition: not just numbers, but "this is a sweep, this is a break of structure". The eye gets trained.
- Trade management testing: how do you handle a position in profit? When do you move SL? Only manual trains that.
- Emotional realism: watching the chart move in time (even accelerated) triggers the same emotions as live. You learn to manage them without losing real money.
Cons of manual backtesting
- Very slow: 100 trades require 8-15 hours of focused work.
- Hindsight bias: you try not to look ahead, but your brain "knows" the next candle is there. It takes hours of practice not to peek.
- Inconsistency: by trade 47 you're tired and take mediocre setups. Those trades don't represent your "real" strategy.
- Few samples: 100-200 trades aren't statistically solid. The result could be luck.
Which to choose and when
Automated backtest → use it for:
- Validating a statistical edge before investing time in it. If it doesn't work even in automated backtest, drop it.
- Comparing variants of a strategy (RSI 14 vs 21, SL 1.5 ATR vs 2 ATR, etc.).
- Stress testing: how does it behave in historical crisis periods (2008, 2020, 2022)?
- Monte Carlo: reshuffle trades randomly thousands of times to estimate the realistic worst-case drawdown.
Manual backtest → use it for:
- Learning to recognize setups in real time. The fundamental difference between theory and practice.
- Training execution discipline: entry, SL, no FOMO, no overtrading.
- Trade management testing: BE, partials, trailing stop. No automated tool replicates it faithfully.
- Validating the edge in specific markets (Asian session, high volatility, range trading) where automated is too coarse.
The combined approach (the one that works)
Professional traders rarely choose "one or the other". They use a 3-phase pipeline:
- Phase 1 — Strategy Builder + Automated backtest: you test the edge on 5-10 years of data. If it has win rate > 45% with RR > 1.5 and drawdown < 15%, move to phase 2. Otherwise discard.
- Phase 2 — Manual backtest: you replay the strategy manually for 100-200 trades. You verify that YOU can execute it in real time without seeing future candles. Add the journal.
- Phase 3 — Forward testing: live with minimum size (0.01 lots) for 1-2 months. If results match the manual backtest, you're ready to scale.
Skipping phases is the recipe for failure. See 65% win rate in automated backtest? Probably overfit. Jump straight to live? Emotions destroy every statistic.
How AlphaNex covers both
AlphaNex is a manual backtesting platform but also includes an Auto Strategy Builder with automated backtesting. You can:
- Build a strategy with AI starting from a description in plain English
- Automated backtest on years of data in 30 seconds
- Open a manual backtest session on the same data to "live" the strategy
- Automatic journal with screenshots, metric calculation, What-If simulator
All in the same ecosystem. No switching between 3 different pieces of software to do the same things.
In summary
- Automated = finds the numerical edge
- Manual = trains the trader to exploit it
- Winning traders use both, not just one
- Skipping manual = failed trader even with a profitable strategy
- Skipping automated = trader burning time on strategies without edge
Start with automated if you need to validate an idea. Start with manual if you have an already proven idea and want to learn to execute it.