Learn
How Do I Backtest?
Backtesting is how you validate a trading setup against real historical data before you risk a dollar. Done correctly, it tells you whether a strategy has statistical edge. Done incorrectly, it tells you what you want to hear.
Why It Matters
Most traders lose money because they trade setups they've never verified.
They saw a setup work a few times, felt confident, and started sizing up — without ever asking “across 100 trades, does this setup produce positive expectancy?” Backtesting forces that question. It separates the strategies worth trading from the ones that just felt right.
Why Manual Backtesting?
Automated backtesting runs code against historical data and outputs a result. It's fast, but it assumes your entry and exit rules can be perfectly encoded — and for discretionary traders, they can't. “Enter when the first 5-minute candle breaks out on heavy relative volume with a bullish catalyst” cannot be written as a simple rule.
Manual backtesting means pulling up every historical chart that meets your criteria and making the trade decision yourself — one candle at a time, without looking ahead. It's slower, but it develops the pattern recognition that makes you better at the setup. Automated backtests produce a spreadsheet. Manual backtests produce a trained eye.
Use manual backtesting for discretionary setups. Use automated backtesting for rule-based systems. Most active traders need both, but manual comes first.
The Backtesting Process
Define Your Setup Criteria Precisely
Write the exact conditions on paper before you look at a single chart. What makes a stock eligible? Gap up 8%+, dollar volume over $1M, price between $5 and $50? What is the entry trigger — a break of the high of a 5-minute candle, a reclaim of VWAP after a flush? What is your stop — below the low of the entry candle, below VWAP? Vague criteria produce meaningless results. You can't test 'looks like a good gap.'
Rule:Write your criteria as a checklist. If a stock doesn't meet every item, it doesn't count as a valid trade in your backtest.Select Your Historical Universe
Choose a starting date and a date range — 60 to 90 trading days is a good first pass. Test across different market conditions: a trending bull phase, a corrective period, a choppy sideways market. A setup that only worked during one regime isn't a durable edge. Keep your stock selection criteria identical throughout. If you'd scan for gaps over 8% today, only test gaps over 8% historically.
Rule:Test in at least two different market environments before calling the setup valid. Edge that disappears in corrections isn't edge — it's beta.Study the Chart Candle by Candle
This is where manual backtesting diverges from automated systems — and where it develops real skill. Pull up the intraday chart and advance it one candle at a time. Decide your entry, stop, and initial target before you see what happens next. This forces you to make the same decision you'd make live. If you look ahead and see that the stock ripped 40% that day before deciding your exit, you're not backtesting — you're fantasizing.
Rule:Cover the right side of the chart. Make your decisions on the left side only. This is the single most important discipline in manual backtesting.Log Every Valid Setup — Not Just the Winners
For every day your criteria fire, log the trade. Entry price, stop price, exit price, shares, R-multiple. Log the ones that hit your stop. Log the ones you would have exited early. Log the ambiguous ones. Cherry-picking only the clean setups is the most common form of forward bias — it inflates your win rate and makes a losing strategy look profitable. Your backtest is only as honest as your logging.
Rule:If you find yourself hesitating on whether to log a trade, log it. The edge case data is often the most informative.Measure Expectancy Across 50+ Trades
Once you have at least 50 logged trades, the math becomes meaningful. Calculate your win rate, your average R on winners, and your average R on losers. Expectancy = (Win% × Avg Win R) − (Loss% × Avg Loss R). A positive number means the setup has edge over a large sample. A negative number means adjust the criteria, tighten the entry trigger, or discard the setup entirely.
Rule:Expectancy matters more than win rate. A 40% win rate with 2.5:1 R is positive expectancy. A 65% win rate with 0.4:1 R is negative. Know which one you have.
Key Metrics to Calculate
The Biases That Ruin Backtests
A backtest that hasn't accounted for these biases is fiction. Learn to recognize them.
Hindsight Bias
You see a chart that ran 40%. Every entry looks obvious in retrospect. The fix: cover the right side of the chart and make decisions as if the outcome is unknown — because in the backtest, it should be.
Survivorship Bias
You only study the stocks that had big moves. The stocks that gapped and faded to -15% don't show up in your memory. Use the scanner to surface every stock meeting your criteria — including the ones that failed.
Cherry-Picking
Logging only the setups where entry and exit are clean and obvious. Real trading is messier. Log every valid setup, clean or not. The ambiguous setups tell you where your criteria need tightening.
Regime Overfitting
A strategy backtested only in a roaring bull market looks great. Test across multiple market environments. If the setup stops working in corrections or choppy periods, factor that into your trading plan.
Sample Size Delusion
Declaring edge after 10 trades. Ten trades is noise. Thirty is the minimum. Fifty is better. Two hundred gives you real confidence. The smaller your sample, the more likely you're trading variance rather than edge.
How Many Trades Do You Need?
With Noetic's 20-year data set, you can collect 200 historical examples of most setups in a single afternoon — before you risk a dollar in a live account.
Common Questions
- Does backtesting account for slippage and commissions?
- Not automatically. Noetic shows real historical price data. You need to account for your broker's typical slippage — for liquid stocks, assume 0.05–0.10% per entry and exit. For illiquid or thinly traded names, slippage can be 0.3–0.5%. Subtract this from your R calculations on every trade.
- What's the difference between backtesting and forward testing?
- Backtesting uses historical data. Forward testing (paper trading) tests in real-time without risking capital. Backtest to filter out bad setups. Forward test to confirm execution and psychology before going live. Both are necessary — backtesting validates the setup, forward testing validates you.
- My backtest shows a 70% win rate. Is that real?
- Question it carefully. A 70% win rate on 20 trades means almost nothing. On 200 trades with consistent criteria, it's more meaningful — but still ask: did you log every valid setup including the ugly ones? Did you study the chart without looking ahead? Cherry-picked backtests can produce win rates above 80% on a losing strategy.
- Should I use 1-minute or 5-minute bars for backtesting?
- Depends on your trading timeframe. If you're making decisions based on 5-minute patterns, backtest on 5-minute charts. If your entries are driven by 1-minute price action, use 1-minute bars. Noetic provides 1-minute resolution, so you can work at whatever granularity matches your live trading.
- What data does Noetic Traders provide for backtesting?
- 20 years of US equities at 1-minute bar resolution, including pre-market and after-hours sessions. The gap scanner lets you screen by gap %, dollar volume, price, market cap, and RVOL to surface every historical day that matches your setup criteria.
Ready to Test Your Setup?
20 years of intraday data. Gap scanner. Playbook. Everything you need.
$49/month. No contracts.