Most people set take profits by picking a round number that feels right, or by copying a two to one reward to risk ratio out of a book that never saw the instruments they trade. Then they wonder why so many trades run to within a whisker of the target and reverse, or why the target never gets hit at all.
The feed already contains the data to replace guessing with measurement, and it sits in two columns most people scroll past. Run-Up and Draw. PnL tells you what a trade paid. Run-Up tells you the best it ever got to and Draw tells you the worst. Together those two bracket the path, and the path is what your exit rules are actually managing.
Outcome and path are different questions
Take the two closed rows from the capture below.
The silver long went in at 64.71 and out at 69.02 for a realised 6.67 percent, and its Run-Up column read +6.92 percent. So the position reached 6.92 at its best and delivered 6.67. That is a capture ratio of about 96 percent. Whoever exited that trade left almost nothing on the table.
The Ford short went in at 14.43 and out at 14.13 for a realised 2.08 percent, with a Run-Up of +3.12 percent and a Draw of -4.37 percent. Capture ratio of about 67 percent. A third of the available move was handed back at the exit, and before any of that happened the position was more than four percent underwater.
Two trades, both winners, and the exits were of completely different quality. Looking at the PnL column alone, the first was better because the number was bigger. Looking at the path, the first was better because the exit was nearly optimal, and that is a repeatable property whereas a large PnL is not.
What the columns look like on the screen
Two practical points come straight off that screenshot. The columns populate on the closed rows in this capture and show dashes on the open ones, so plan on building your table from closed history rather than watching it live. And the TP and SL columns being empty on every row means none of these positions had a resting target or stop working. That is a legitimate way to trade, but it makes the exercise below considerably harder, because you cannot audit a level you never wrote down.
The Draw column is the stop you should have set
Start with the stop, because it is the half of this that actually protects you.
The Ford short finished at plus 2.08 percent after being 4.37 percent underwater. If your standing rule is a three percent stop, that trade is a loser in your account and a winner in the feed, and the difference has nothing to do with signal quality. You were stopped out by ordinary path noise on a signal whose normal behaviour includes that much heat.
So the stop distance is not a statement about your risk appetite. It is an empirical question about how much drawdown the signals you trade typically take before they work, and the Draw column answers it directly. Pull the draws of your winning trades, look at where the bulk of them sit, and put the stop outside that cluster.
The obvious objection is that a wider stop means more risk, and it does, unless you fix the other half. A stop at five percent instead of three means the position has to be smaller by forty percent to keep the same dollar loss. If you are risking a hundred dollars per trade, a three percent stop supports a 3,333 dollar position and a five percent stop supports a 2,000 dollar position. Same hundred dollars at risk, fewer stop outs on trades that were going to work. That trade is nearly always worth making, and the mistake to avoid is widening the stop while leaving the size alone, which is just quietly tripling your risk budget.
Building the table from rows you actually traded
The filters above the table are what make this tractable. Set STATUS to closed so you only get rows with populated path columns. Set SOURCE to the engine you actually trade, because a momentum entry and a reversal entry have completely different path shapes and averaging them together produces a table that describes neither. Set the timeframe filter to the horizon you trade.
Then record four numbers per row. Direction, realised PnL, Run-Up, Draw. A spreadsheet is fine. From those, compute two figures.
- The median Run-Up of trades that finished positive. That is your candidate take profit zone, and it is usually a lot closer than the target people set by instinct.
- The eightieth to ninetieth percentile of Draw among trades that finished positive. That is your minimum stop distance, the point past which you are cutting winners rather than cutting losers.
Be honest about sample size. Four rows in a screenshot is not a sample, and neither is fifteen rows of your own. Thirty closed trades on one source at one horizon is roughly where the numbers stop moving every time you add a row, and even then the right posture is that this is a rough guide rather than a parameter. If your table tells you to move your stop by a tenth of a percent, do nothing.
And having computed the take profit, place it. The TP column exists. A target you intend to honour in the moment is not a target, it is an intention, and the whole reason the Ford short gave back a third of its run-up is that somewhere a decision got made in real time instead of in advance.
Three ways this method misleads you
The first is regime. Run-up and draw distributions are conditional on volatility, and a table built during a quiet quarter will set take profits that stop triggering the moment ranges expand, and stops that get hit constantly the moment they contract. At minimum, segment your table by period so you are not blending a 4h signal with a 1day one. Better, normalise the figures against a volatility measure so the numbers travel between regimes.
The second is that the table describes you, not the signal. You only have path data for the trades you took, and you took those for reasons, including some bad ones. If you systematically skip the setups that look frightening, your draw distribution is flattered and your stops will be too tight for the population of signals you will trade in future.
The third is the one that quietly destroys expectancy. A take profit truncates the right hand tail. Many signal-driven approaches make their money from a small number of trades that run far past anything typical, and if you set a target at the median run-up you will capture more of the ordinary winners and systematically amputate the rare enormous one. Before you place a fixed target on everything, check what your total result looks like with the largest two or three winners removed. If the answer is that the strategy stops working without them, then your take profit rule needs to let something run, and the run-up table is telling you where to trail a stop rather than where to exit.