Testing a structure idea honestly

A channel posts five winning screenshots of its order-block setup. Each one was chosen after the result was known. To find out whether the setup works, you need every trade the rule would have taken, losers included.

From a picture to a rule

Trend pullbackBreakoutRange fade

1 Setup2 Trigger3 Stop4 ExitAll

Any testable method has the same four parts: setup, trigger, stop and exit, each fixed before the trade.

The pullback above is decided in full before the trade: zone, trigger candle, stop and target. A structure idea needs the same treatment. Here is one written out as rules. It is an illustration of the level of detail, not a recommendation.

• Market and hours: EUR/USD, H1, 07:00 to 17:00 UK time.
• Swings and breaks: N = 2; breaks count on candle closes.
• Setup: a CHoCH down after at least two BOS up (the mirror for buys).
• Zone: the order block of the move that made the CHoCH, full candle range.
• Entry: sell limit at the block’s lower edge, first return only, within 20 candles of the CHoCH.
• Stop and target: stop 3 pips above the block; target the latest swing low; skip the trade if that is under 1.5R.
• Risk: 0.5% per trade while testing.

Then check the rule for gaps. Give it to someone else with the same chart. If you mark different trades, some step is still a judgement call.

Small stops make costs bigger

Order-block entries often use stops of 10 to 15 pips, so the spread and commission are a large share of each R. The testing a strategy lesson covers the three stages; the worked example shows what costs do to a typical structure result.

High R means long losing runs

Risk per trade: Win rate: Reward-to-risk:

Run 100 trades again

Final balance

Worst drawdown

Longest losing streak

Each run uses the same settings with different luck. Run it several times: the shape changes, the drawdowns don't go away.

These settings have a real edge: 0.25 × 4 − 0.75 = +0.25R per trade. Yet at a 25% win rate, a run of ten or more losses somewhere in 100 trades happens in roughly four runs out of five. Press the button a few times and watch how long the flat or falling stretches last. A screenshot of a 1:5 winner says nothing about the nine losers around it.

Biases to guard against

• Hindsight. Labels are easy on a finished chart. In a backtest, scroll forward one candle at a time and decide with the right side hidden.
• Selection. Posted examples are winners. Count every setup the rule produces.
• Shifting definitions. After a loss it is tempting to decide “that was not a real order block”. If the rule marked it, it counts.
• Too many variants. Try 20 versions of the rule and one will look good by luck. That is curve fitting.

Then forward test the same rules on a demo account before any money goes in.

Try it

Backtest log

+3R+2R+1R0R−1R−1.5R

Any result (R)Add

Undo lastClearLoad example

Trades

Win rate

Avg win

Avg loss

Expectancy (R/trade)

Total R

Longest losing streak

Max drawdown (R)

Plausible range of the true expectancy

Your backtest results, in R. The range shows how much the true result could differ from what this sample shows.

Start with the example, then tap −1R eight times in a row, the kind of streak a 25% method produces, and watch the expectancy and its plausible range fall.

Quick check

Quick check

1. Question 1 of 4

   Which rule is precise enough to test?

   ASell limit at the lower edge of the last green candle before the CHoCH, first return onlyBSell when price reaches a strong order blockCSell when smart money starts distributing

Go deeper: Trading Expectancy: The Number That Says If a Strategy Works · How to Keep a Trading Journal (Template Included) · Lesson: Testing a strategy