Exits Decide the Equity Curve: Scaling Out vs. All-In-All-Out, and Why Exit Design Moves Results More Than Entry Tweaks
Most traders spend their weekends tuning entries. They shift the breakout trigger a tick, add a filter, move the moving average from 10 to 12, and rerun the backtest hoping the equity curve steepens. It rarely does — not by much. Then they change one thing about how the trade ends, and the whole curve changes shape.
That's not an accident. Your entry decides whether you're in a trade. Your exit decides what that trade is worth. On an automated strategy running in NinjaTrader 8, the exit logic is doing the heavy lifting on every single fill — win, loss, and the messy ones in between.
The same entry, three different strategies
Take one opening-range breakout signal. Same instrument, same session, same trigger. Now attach three different exit plans to it:
- All-in, all-out at a fixed target. Two contracts on, both off at +20 points or at the stop. Clean, symmetrical, easy to model.
- Scale out. One contract off at +10, stop to breakeven, the runner trails until it's stopped or the session ends.
- Trail from the start. No fixed target at all — the position rides a trailing stop from entry.
Backtest those three over the same trade list and you will not get three versions of the same result. You'll get three genuinely different strategies: different win rates, different average winners, different losing streak lengths, different depth of drawdown. Same entry. The market gave all three the identical opportunity. The exit decided what came of it.
This is why we treat exit design as a first-class part of a bot, not a setting you tack on at the end. It's also why two automated strategies can post nearly identical annual returns and feel completely different to hold — a point we broke down in Two Algos, Same Annual Return, Completely Different to Trade.
What scaling out actually buys you (and what it costs)
Scaling out is popular for a reason that has almost nothing to do with math and everything to do with staying in the seat.
What it buys:
- A higher hit rate on something. Taking a partial at a modest target converts a much larger share of trades into "at least partly green." That changes what your trade list looks like day to day.
- Smaller give-back. Moving the stop to breakeven after the first partial caps the classic frustration: watching a +30 trade round-trip to a loss.
- Shallower equity dips. Because more trades book something, the curve tends to grind rather than lurch — which matters enormously on a funded account with a trailing drawdown watching your every peak.
What it costs:
- Your biggest days get smaller. The trades that pay for the whole month are the ones that run. Scaling out sells the front half of exactly those trades.
- Your losers stay full size. A scale-out plan reduces winners; it does nothing to the trade that hits the stop with the full position on. That asymmetry is the hidden cost.
- You need the contracts to do it. One micro can't be scaled out of. Sizing and exit design are the same conversation — see Micros or Minis for Your Eval.
All-in-all-out is the mirror image: bigger winners, more give-back, a lumpier curve, and fewer decisions to get wrong. Neither is "correct." They're different trade-offs, and the right one depends on the strategy's edge profile and the account rules you're trading under.
Match the exit to what the strategy is actually catching
Here's the part that separates a thought-out lineup from a pile of bots: the exit should reflect what kind of move the strategy is trying to capture.
Momentum and breakout strategies
An ORB or breakout bot earns its keep on the days the market decides to go somewhere. Those days are a minority of the sample, and they carry the year. Cap the runner and you've amputated the edge. Breakout strategies generally want some portion of the position left to run — a trail, a session-close exit, a wide secondary target.
Mean-reversion strategies
A Snapback-style bot is betting on a rubber-band move back to a level. That move has a natural destination. Once price gets there, the edge is spent — holding for more is just an open-ended bet with no thesis behind it. Mean-reversion generally wants a defined target and a prompt exit, not a trail hoping for a trend that the strategy was never designed to catch.
Run both kinds together and the exit designs are supposed to look different. That's the mechanism behind the smoothing effect we covered in Breakout vs. Snapback — and it's also why identical exit rules pasted across a whole lineup quietly homogenize strategies that were supposed to behave differently. If your bots are already correlated, matching exits makes it worse; that's the same failure mode described in Correlation Is the Hidden Risk in Your Bot Portfolio.
Time is an exit too
The exit nobody codes and everybody needs: the clock.
A strategy built on the energy of the first hour has no business holding a position into the lunch chop. A session-end flatten isn't a compromise — it's a rule that says "the conditions my edge depends on are gone." Time-based exits also do something a price-based exit can't: they bound your exposure. You know the position will be flat by a certain minute regardless of what the tape does.
The same goes for a hard flatten before a scheduled news release. It's not a prediction. It's a decision made in advance, executed without a human deciding whether today feels different.
How to test an exit change without fooling yourself
Exit changes are the easiest place in all of strategy development to overfit, because moving a target a few ticks visibly "fixes" the worst trades in your sample. Some discipline:
- Change one thing. Target, or trail, or scale ratio — not all three. If you can't say which change caused the improvement, you haven't learned anything.
- Judge the curve, not the total. Net profit is one number. Look at max drawdown, longest losing streak, average winner vs. average loser, and time under water. An exit tweak that adds 8% to the total and doubles the drawdown made the strategy harder to trade, not better.
- Use enough sample. Our bots are built against roughly 4.5 years of backtest data for exactly this reason — an exit rule that only shines in one regime isn't an exit rule, it's a memory of 2023.
- Check the tails. Pull the ten biggest winners. Did your new exit still capture them, or did it clip every one? If it clipped them, be honest about what you gave up for the smoother curve.
- Prove it in sim before it touches a funded account. Every exit change is a new strategy. Treat it like one — the sim-to-live checklist applies just as much to a modified bot as to a brand-new one.
The Bot Portfolio Analyzer exists to make step 2 fast — comparing drawdown, streaks and curve shape side by side instead of squinting at two profit numbers and calling the bigger one better.
The takeaway
Entries get the attention because they feel like the skill. Exits get the results because they decide the size of every outcome the entry produced. If you have a weekend to spend on your lineup, spend it here: pull your trade list, look at what your exits did to your ten best and ten worst trades, and decide — deliberately, in writing — whether you're trading for a smoother curve or a bigger tail. Then let the bot execute that decision without you renegotiating it at 9:47 in the morning.
That's the whole point of automating it. The decision gets made once, when you're calm, with the data in front of you. No coding required to run our strategies — but the thinking about how a trade should end is still yours to do.
Want to see how our bots handle exits? Every strategy in the lineup — ORB, Snapback, Scalper, 10 EMA, NR7 and Volume Spike — ships with its exit logic built and backtested, and the Bot Portfolio Analyzer lets you compare curve shape before you commit a dollar. Start the 14-day free trial at pushbuttontrading.co.
Trading futures involves substantial risk of loss and is not suitable for every investor. Backtested results are hypothetical and are not a guarantee of future performance.
