Strategies · Lesson 18 · Advanced
From gut feeling to playbook: defining and testing setups
You put your setups into if-then form, give them a measurable target, and test them in the bar replay before real money is even up for debate.
With a free account: interactive chart exercises, quizzes with answers, progress and XP.
What makes a setup testable
A gut feeling sounds like this: the coin looks strong. A testable setup sounds like this: IF price closes above the upper edge after at least three weeks of range AND volume is above the 30-period average, THEN long on the retest of the edge. The difference is the if-then form.
The test is simple: could a stranger with your rule read the same yes or no off the same chart? If yes, the setup is testable. If your setup contains words like strong, beautiful or overdue, it is still an opinion.
From gut feeling to a tested playbook
Interactive exercise: here you learn right on the chart, with feedback on every click. Sign up freeto start it.
Two cleanly defined example setups
Setup A, range breakout with retest: IF price has been swinging between two horizontal edges for at least three weeks AND closes above the upper edge AND the breakout volume is above the 30-period average, THEN long as soon as price tests the broken edge from above and holds. Stop below the edge, measured target see below.
Setup B, EMA pullback in a trend: IF the 4h chart shows higher highs and higher lows AND the 13-EMA is rising AND price has retraced 33 to 50 percent of the last impulse move to the EMA, THEN long when a 15min candle closes above the last minor high. Stop below the pullback low.
The measured target instead of an opinion
Bulkowski gives every pattern a price target via the measure rule: measure the formation height and add it to the breakout. The conservative variant uses half the height, it hits much more often in his data. For setup A that means: measure the range height, add half the height to the upper edge, that is your first target.
Important and honest: Bulkowski's average-rise numbers assume perfect trades without costs. They serve to compare setups, not as a return promise. With leverage, spread, slippage and funding eat into it on top. The measured target is a minimum that you align with real resistance, not a price oracle.
20 runs in the bar replay before real money is on the line
Before you even think about real money, you test every setup in the bar replay under /simulator/replay. There the chart runs candle by candle, you decide blind as in the live market. Goal: at least 20 clean runs per setup. That is not yet a statistically hard sample, but enough to see whether you can even recognize and execute the rule consistently.
20 trades are deliberately few for a verdict on the edge (that takes hundreds, see the edge lesson). But they are enough to find craft errors: do you recognize the setup reliably? Do you stick to stop and target? That is exactly what you check first.
The journal columns for the test
Keep the same columns for every replay trade: setup name (A or B), date/coin, entry, stop, target, result in R, did you follow the rule exactly (yes/no), a note.
The most important column is rule followed exactly. Separate process from result: a won trade with a rule break is a bad trade, a lost trade by the plan is a good trade. Only once 20 runs are done and you have executed the rule almost always cleanly is the discussion about real money open at all.
Test yourself
Which of the two is a testable setup?
- IF close above the upper range edge AND volume above the 30-period average, THEN long on the retest
- If the coin looks overdue and feels strong, then long
You ran your setup 20 times in the bar replay: 9 winners, 11 losers, rule followed almost always cleanly. What is the right conclusion?
- 20 trades are too few for an edge verdict, but I can execute the setup cleanly, so keep testing
- The setup is worthless, it had more losers than winners
- Get in big with real money right away, it ran roughly even after all
Sign up free for the answers with an explanation for each option.
The key points at a glance
- A setup is testable when it stands as an if-then rule with no room for interpretation.
- Every setup gets an objective measured target, not an opinion.
- Bulkowski thinking: argue with success rates and statistics, not with conviction.
- First 20 runs in the bar replay, then talk about real money.
Deep dive
How many trades you really need to prove an edge
The 20 runs in the bar replay test your execution, not your edge. They show whether you recognize the setup and stick to stop and target. Whether the rule makes money across the series is only revealed by hundreds of trades, since Bulkowski builds his statistics on more than 38,500 samples.
- 9 to 11 after 20 trades is no reason to scrap the setup.
- Below a 50 percent hit rate can be clearly profitable with a good reward-to-risk ratio.
- Dennis: 95 percent of profits come from 5 percent of trades.
Reading the measure rule honestly: only a minimum, not an oracle
Bulkowski's measure rule adds the formation height onto the breakout, but the full height is hit disappointingly rarely. Half the height is statistically the better choice.
- Double Top: full height only 40 to 44 percent, half height 68 to 79 percent.
- Cup with Handle: hit rate jumps from 50 to 76 percent.
- Average-rise numbers assume perfect trades with no costs.
- Treat the measured target as a minimum, align it with real resistance.
Backtesting without fooling yourself: the most common errors
On a finished chart every setup looks obvious because you know the outcome. That is why the bar replay runs candle by candle, so you decide blind as in the live market and read nothing into it in hindsight.
- Overfitting: Hite looks for the hardiest method, not the optimal one.
- Robust beats optimal, test across multiple coins and market phases.
- Confirmation bias: put every trade in the journal, even the ugly ones.
- A breakout only counts on the close outside the formation, not on the wick.
1R = your risk unit: the loss you have budgeted per trade (distance from entry to stop-loss). +2R means: you won twice as much as you risked.
R/R (reward-to-risk ratio) = potential profit target divided by planned risk. 2:1 means: double the reward against the risk.
Sources: Bulkowski, Thomas: Encyclopedia of Chart Patterns; Douglas, Mark: Trading in the Zone; Schwager, Jack: Market Wizards
Make this lesson interactive
Sign up for free and learn with click exercises right on the chart, quizzes with explanations and saved progress. Then you practice everything risk free on the demo exchange.
100% free, no payment details.