Plenty of indicators draw levels on the chart and show how often price reaches them. 80% here, 90% there. Before you trust a number like that, ask one question: how far away was the level?
Often, much of a hit rate is distance. A level close to price gets reached more often than a level far away, whatever drew it. So the useful question isn't just "how often does price reach this level?" It's "how often does it reach it compared with any level at the same distance?"
Free: the distance check sheet. A ready-made spreadsheet that runs the check below on your own signals. Excel and Google Sheets, no sign-up. Download the sheet (.xlsx)
The coin-flip baseline
Start with a market that has no pattern at all. Flip a coin: heads, price moves up a point; tails, down a point. Put a target 10 points above and a stop 30 points below. Which gets hit first?
For a fair coin, the chance of reaching the nearer level first is the far distance divided by the total:
- Target 10, stop 30: 30 / (10 + 30) = 75%
- Target 5, stop 45: 45 / (5 + 45) = 90%
- Equal distances: 50%
No signal, no skill. Move the target nearer and the hit rate goes up, with nothing else changed. A level that hits 90% of the time may simply have been drawn close to price.
A high hit rate can still make nothing
The coin-flip version also shows why a hit rate isn't a result. Win 10 points 75% of the time and lose 30 points 25% of the time:
- 0.75 x 10 = 7.5 points gained
- 0.25 x 30 = 7.5 points lost
A 75% hit rate and zero expectancy. Commission and slippage then make it negative. A level is only worth something if it does better than its own distance says it should.
Measure distance in ATR, not points
Twenty points on a quiet day is a long way. On a busy day it's a small move. So measure each distance in units of the market's recent range, usually the average true range (ATR) of the chart's bars, taken at the moment the level is drawn: a 20-point target with a 40-point ATR is 0.5 ATR away.
Points hide this. A level that "works better in the morning session" may just be drawn closer to price, in ATR terms, in that session.
The trap: distance in disguise
This is where many false findings come from. Say you split an indicator's levels by condition and find condition A hits 80% of the time and condition B 52%. It looks like condition A is special.
Now look at the distances. In condition A the target sat 0.3 ATR away and the stop 1.2 ATR away; the coin-flip baseline is 1.2 / 1.5 = 80%. In condition B both sat 0.6 ATR away; the baseline is 50%. Almost all of the difference is geometry. Condition A drew its targets closer.
In our own testing on NQ, a level's distance from price, in ATR, was the biggest factor in whether price got there. Several patterns we tested, including session conditions and times of day, added nothing once we compared them at equal distance.
The check, step by step
You can run this in a spreadsheet on any indicator's signals, your own included, or use the free sheet above.
- Record every signal. Time, price, the ATR at that moment, the target, the stop, and which one price reached first, or neither. Every signal, not a selection.
- Convert the distances to ATR. Target distance divided by ATR, stop distance divided by ATR. The baseline in step 3 is a ratio, so it is the same in points or ATR; ATR matters for the buckets in step 5 and for levels with no stop.
- Work out the coin-flip baseline for each signal. Stop distance divided by (target distance + stop distance). The average across all signals is the hit rate a patternless market would give you at those distances. Leave out signals that reached neither level, and say how many; if there are a lot, the baseline runs a little low for near targets.
- Compare the real hit rate with that baseline. The gap is what the signal adds over distance. A gap near zero means the number is mostly geometry.
- Compare in distance buckets. Group the signals by target distance, for example under 0.25 ATR, 0.25 to 0.5, 0.5 to 1, and over 1. Check the gap in each. An average can hide a signal that only looks good where its targets happen to be close.
- For levels with no stop, use a mirror. Some indicators only say "price will reach this level today". For each one, take the same distance on the other side of price, from the same moment, and record whether price reached that too. In a market with no direction, the level and its mirror get hit about equally often. What the level adds is how much more often it gets hit than its mirror. Check levels above price and below price separately: a trending market alone makes one side get hit more often.
- Then turn it into R. A gap in hit rate isn't profit. Work out the result per trade in R, after commission and slippage, before you believe the gap matters.
What this check can't tell you
Real markets aren't coin flips. Trend and changing volatility bend these odds, most of all at wider distances, so the coin-flip baseline is a first check, not the last word. The mirror step helps, because the mirror sits in the same market at the same moment. The stronger version is a random benchmark: the same stops and targets, placed at random times in the same market, and the signal compared with that.
Sample size matters too. With 50 signals at around 80%, the hit rate itself is uncertain by roughly 11 points either way (a rough 95% range). A gap of a few points on a few dozen signals can't be told apart from noise.
And a gap that survives all of this still has to pay its costs. "What does it add over distance?" is the first check we run on any level, signal or indicator. One that clears it still faces the five tests every strategy is scored against before we offer it; the Tested page lists the results.
Not advice. Futures trading carries substantial risk of loss.