Forward Testing a Trading Strategy

Backtesting tells you whether a strategy could have worked. Forward testing tells you whether it works when you have to make the call in real time, without knowing what the next candle does.

Most strategies look better in the backtest than in the forward test, and a lot of that gap comes from how differently you trade when you can't see the right side of the chart. I do this with my own trading too. Every new system I trade goes on a funded prop account first, and it only touches my own money after it has held up there.

What Is Forward Testing in Trading?

Forward testing means taking a strategy with fixed, written rules and trading it in the live market, on a simulator or with the smallest size you can, and logging every single trade. You decide in advance how many trades the test will run for, and you don't change the rules until it's done.

It's sometimes called out-of-sample testing, because the strategy is being tested on market data it has never seen. Paper trading is one way to do it, but paper trading without written rules, a set sample size and a log of every trade is just practice.

Backtesting vs Forward Testing

You need both. Backtest first to find out whether an idea is worth your time, because it's fast and costs nothing. Then forward test before any real money goes on it.

And even after both, you still need structure when you trade it live. A strategy with an edge does nothing for you if you don't follow it. That part is in trading discipline.

Why Backtests Look Better Than Live Trading

When you look back at a chart, you already know what happened, and it's very hard to unknow it. Look at a losing trade afterwards and you'll find a million reasons why it was obviously a bad trade. That hindsight leaks into manual backtests in a few ways:

• You skip the messy trades. Setups that would have been confusing in real time get quietly left out.
• You get perfect fills. A backtest fills your stop at the stop price. In a fast market the live fill can be a few ticks worse.
• You fit the rules to the past. Every tweak that makes the backtest look better also makes it more likely you're describing old data instead of finding an edge.
• You leave out costs. Commissions and fees on every round trip add up quickly for a day trader. In one study of US day traders, the average trader was up more than $8,000 before costs and down about $750 after them (more in what percentage of day traders are profitable).
• There's no fear. Nobody hesitates on a backtest. Live, you'll see the setup and freeze, or take it early.

How to Forward Test a Strategy, Step by Step

1. Write the rules as a checklist

Entry, stop, target, size, max trades a day, and when you don't trade. If a rule can't be checked with a yes or no after the session, rewrite it until it can.

2. Decide the sample size and end date first

Write down how many trades the test will run for before you take the first one. If you don't, a losing streak in week two will talk you out of a strategy that might have been fine, or a lucky week will talk you into one that isn't.

3. Pick where to test

A simulator is fine for learning the mechanics. TradingView paper trading works for futures, and MotiveWave, the platform I chart on, has a simulator too. As soon as you can follow the rules on sim, move to the smallest live size, like one micro contract. Real fills and real nerves are part of what you're testing.

4. Log every trade, with a rule check

For every trade, record the setup, entry, stop, exit and result in R, plus one more column: did you follow every rule, yes or no. That column is what makes the results readable later. A trading journal that grades execution does this for you.

5. Don't change anything until the sample is done

No new filters, no new indicator, no moving the stop after two bad days. The moment you change a rule you've started a new test, and everything before it stops counting.

6. Include costs

Use your real commission and fees per round trip, and assume a tick or two of slippage on stops. If a strategy only makes money when trading is free, it won't make money for you.

How Many Trades Do You Need to Test a Strategy?

More than most people think. Win rate is a percentage, and small samples swing a lot. Say you measure a 50% win rate. This is the range the real win rate probably sits in, depending on how many trades you took:

95% range using the normal approximation for a measured 50% win rate. Smaller or larger win rates give slightly narrower ranges.

Now put that next to your breakeven. With a 2 to 1 reward to risk, you break even at a 33% win rate. After 20 trades at 50%, the real number could be below that, so you can't even tell a winning strategy from a losing one. After 100 trades the range is 40% to 60%, which clears breakeven comfortably.

My rule of thumb is at least 50 trades before you judge a strategy, and 100 or more before you trust it with more size. How long that takes depends on how often you trade. If you take ten trades a day, 50 trades is one week, and it may all come from one kind of market. If you take two a day, it's spread over a month or more, which covers more conditions. A high-volume trader should collect a bigger sample for that reason.

How to Read Your Forward Test Results

When the sample is done, look at the trades where you followed the rules and work out:

• Win rate and average win and loss in R (R is the amount you risked).
• Expectancy: win rate × average win, minus loss rate × average loss. Win 45% of the time with an average win of 1.8R and an average loss of 1R, and you make 0.45 × 1.8 − 0.55 × 1 = +0.26R per trade on average.
• Max drawdown: the worst run from a peak, in R. It tells you how big your account needs to be to survive the bad stretches.
• Rule-following rate: the share of trades where you followed everything.

Then put the result in one of four boxes:

The second row is the one people struggle with. If you followed your rules for 50 trades and lost money, the strategy has no edge in its current form. That's painful, but it's also the most useful thing a test can tell you, because now you know exactly what has to change. If you never follow your rules, you'll never get that answer.

And if the plan was right but you kept breaking it, the problem isn't the strategy at all. In my experience coaching, it usually is. That's covered in trading discipline.

Forward Testing Mistakes to Avoid

• Changing the rules in the middle of the test.
• Judging the strategy after 10 or 20 trades.
• Testing only in one kind of market, like a strong trend week.
• Leaving out commissions, fees and slippage.
• Not logging the trades you'd rather forget.
• Mixing trades where you broke the rules in with the ones where you didn't.
• Quitting after a losing streak that's well inside what the sample size allows.