TL;DR: In a prop firm challenge, position size decides which rule you hit first and how long the attempt takes. Size so that your worst day, with every correlated stop hit, fits under the daily limit with room to spare; keep leverage around 1–3x, because at 10x costs alone use about 1.7% of the account per trade. Smaller size makes you slower and safer from the daily limit, but without an edge it does not raise your odds of passing.
Why is position sizing the main decision in a challenge?
A challenge account has a target above and two floors below: a daily loss limit and a maximum drawdown. Your strategy decides the direction of each trade. Your size decides how fast equity moves toward the target or the floors. The same signals at twice the size reach the target or a floor much sooner (about a quarter of the time with no edge, about half with a steady edge), and hit the daily limit on days that would have been survivable at half size.
Our replays of Trigr's house plans for Propr show the trade-off. Each plan was replayed with a fresh challenge on every UTC day since 2023-07-24 (since 2024-04-02 for Conservative), up to 365 days each, under Propr's exact rules. Classic 1-Step:
| Plan | Size vs recipe | Backtest pass | Median days to pass | Failed by daily loss | Failed by drawdown |
|---|---|---|---|---|---|
| Fast | 1.4x, 3 strategies | 60% | 20 | 36% | 4% |
| Balanced | 0.5x, 1 strategy | 87% | 53 | 2% | 11% |
| Conservative | 1x, 2 strategies | 100% | 193 | 0% | 0.2% |
The Fast plan's failures are almost all daily-limit breaches. Bring size down and those nearly vanish, at the cost of more than twice the time to pass. Balanced runs a different, smaller strategy set, so Fast to Balanced is not a pure size change, and the Conservative plan also uses different strategies.
Read these numbers with care
They are in-sample: the plans were chosen with the 2025 data visible, and overlapping daily starts share most of their path, so 798 starts are not 798 independent tries. In out-of-sample research (choosing on 2020–2023 starts, scoring on 2025–26), picks that looked like 40–80% in-sample plateaued at about 25–34%. Our realistic planning figure for the slowest, smallest-size tier is 40–55%, not 90–100%. The pass rates post has the full detail.
What do size multiple and leverage mean?
Two numbers set your exposure, and they are easy to confuse:
- Size multiple scales a strategy's normal position size. In Trigr's Propr plans, Fast plans run their strategies at 1.4–2x their recipe size, Balanced at 0.5–1x, Conservative at 0.7–1x.
- Leverage is notional exposure divided by the margin behind it. At 3x, $1,000 of margin controls $3,000 of position.
What reaches your equity is notional exposure. A $3,333 slot at 2x has $6,667 of notional; a 2% move against it costs $133 whatever you call the settings. Propr caps leverage at 10x for BTC, ETH and SOL, and at 2x for other crypto, so on altcoins the cap does some of the sizing work for you.
How much leverage should you use?
Our best challenge results came at 1–3x. The reason is cost. A round trip of about 17 bps of notional is small at low leverage and large at high leverage, measured against the account:
| Leverage (whole account deployed) | Round-trip cost as % of account | Share of a 3% daily limit |
|---|---|---|
| 1x | about 0.17% | about 6% |
| 2x | about 0.34% | about 11% |
| 3x | about 0.51% | about 17% |
| 5x | about 0.85% | about 28% |
| 10x | about 1.7% | about 57% |
At 10x, two round trips in a day cost more than the whole daily limit before price moves at all. Costs also push you below the zero-edge ceiling with every trade. Our posts on trading fees in perp backtests and slippage and funding show how quickly they compound.
How do you size for the worst day?
Start from the daily limit and work backwards. On Propr's 1-Step challenges the limit is 3% of the balance the day began with (the 00:00 UTC realised-balance snapshot), judged on equity, with a momentary touch enough to fail.
Worst-day loss ≈ Σ (notional × stop distance) + Σ (notional × round-trip cost) + open loss carried over midnight + a gap allowance
Worked example, $10,000 Classic 1-Step, three slots on three altcoins, each $3,333 at 2x ($6,667 notional), stops 2% away:
- At the stop: 3 × $133 = $400
- Costs: 3 × about $11 = about $34
- Worst day: about $434, or 4.3%. Over the $300 limit.
Halve the leverage to 1x and the worst day falls to about $217, or 2.2%, inside the limit with room. Then add a buffer for gaps: stops can fill past their level in a fast market, and on equity-judged rules a wick that reverses still counts. The daily loss limit guide works through the midnight case, where an open loss eats the next day's allowance.
A target for the sum: keep the planned worst day well below the limit, and treat the worst day in a backtest replayed from every start date as a minimum estimate, not a ceiling.
Why do correlated slots count as one position?
Because they lose on the same days. Three strategies on three different altcoins look diversified on a normal day, but on a sharp market-wide drop they often all hit their stops together. For daily-limit purposes, assume correlation is 1 unless you have strong evidence otherwise. Two practical consequences:
- Adding a slot on a correlated asset adds to the worst day almost in full.
- A strategy on a different timeframe or a different signal family (a breakout next to a trend follower) diversifies more than a second asset with the same signal.
Our multi-strategy portfolio agent guide covers how slots combine.
Does smaller size raise your odds?
Not on its own. With no edge, equity behaves roughly like a random walk, and the chance of reaching the target before the drawdown floor is about drawdown ÷ (target + drawdown): about 37.5% on the Classic 1-Step, 25% on the Turbo 1-Step, about 29.4% on the Pro 1-Step, and on the Classic 2-Step about 32% (the simple formula's 44% ignores the trailing floor). Those odds do not depend on size. Trading at half size makes the same random walk take roughly four times as many steps to reach either side, so the attempt is slower, not more likely to pass.
Smaller size does two useful things. It stops the daily limit from ending attempts that the drawdown would have survived. And it gives a real edge time to show, since an edge accumulates over many trades. The Kelly criterion makes the same point from the other side: optimal size depends on a known edge, and an overestimated edge gives sizes that are far too large. The odds math post works through the baseline.
How does Optimize sizing work?
When you deploy your own agent on a Propr account, Trigr replays the agent's combined backtest with a fresh challenge started on every UTC day, against that account's real daily-loss and drawdown rules, and reports the worst day, maximum drawdown and pass statistics. If the backtest breaches or leaves less than a 0.5% buffer, Optimize sizing scales every position down and re-runs it (up to three rounds per click) until it fits with that buffer. House plans show their measured record instead of this check.
Treat 0.5% as a minimum. Live trading has costs and gaps a backtest misses, and the replay measures a position held across midnight one day at a time. The backtest vs live gap post covers the usual differences.
Sizing checklist
- Write down the daily limit and drawdown floor in dollars.
- List every position that can be open at the same time, with notional and stop distance.
- Assume correlated positions all stop out on the same day.
- Add round-trip costs at your leverage and a gap allowance.
- Keep the total clearly under the daily limit.
- Prefer 1–3x leverage; check the venue's caps.
- Check the replayed worst day and worst drawdown against the limits, with a buffer.
- Forward-test on paper at that size before paying a fee.
How Trigr fits in
Trigr's house plans on Discover › Propr are fixed sets of published strategies with weights and a size multiple, and the server sizes every slot when you deploy. Each plan shows its backtest pass rate, days to pass and failure split, so you can see what a given size does. The live record began on 2026-09-30; as of 2026-10-05, each plan has six forward challenges running, none breached and none finished. For your own strategies, the rules check and Optimize sizing do the fitting, and a paper agent gives you a forward test at live Hyperliquid prices. If you then take a challenge, you can open one through our Propr referral link.