TL;DR: We backtested 12 house plans across Propr's four challenge types, starting a fresh challenge on every day since mid-2023, each given up to a year. In-sample, they pass 58–100% of starts. Out of sample, comparable picks passed about 25–34% within 180 days, and a realistic planning figure for the slowest, smallest-size tier is 40–55%. This post shows all the numbers and explains why the gap exists.
How did we measure pass rates?
A single backtest from one start date tells you little about a proprietary trading challenge. The outcome depends heavily on when you start: begin the week before a sharp selloff and even a good strategy can breach. So we measured each plan the way a challenge actually runs, many times over:
- A fresh challenge on every UTC day since 2023-07-24, giving 798 daily starts. The Conservative plans use strategies with shorter histories, so they start on 2024-04-02 (545 starts).
- Up to 365 days per attempt. Propr has no time limit, but we cap the window at a year so every start can finish or be counted as unfinished.
- Propr's exact formulas. Daily loss from the 00:00 UTC realised balance, static floors on 1-Step challenges, the trailing floor on the 2-Step, equity checked hour by hour, using each hour's low and high for breaches (Propr checks every moment, so a live account can breach inside an hour), and the profit target judged on equity too. The rules come from Propr's rules page.
- Costs. Trading fees, plus 3 bps of slippage per side and funding.
- Data through 2026-09-29.
What do the backtests show for all 12 plans?
| Plan | Pass (backtest) | 90% interval | By start year | Median / avg days to pass | Failed by daily loss | Failed by drawdown |
|---|---|---|---|---|---|---|
| Classic 1-Step Fast | 60% | 54–69% | 64 / 59 / 59% (2023/24/25) | 20 / 24 | 36% | 4% |
| Classic 1-Step Balanced | 87% | 78–95% | 92 / 89 / 83% | 53 / 69 | 2% | 11% |
| Classic 1-Step Conservative | 99.8% | 99.8–100% | 2024 100%, 2025 100% | 193 / 185 | 0% | 0.2% |
| Turbo 1-Step Fast | 59% | 54–67% | 68 / 61 / 52% | 19 / 25 | 13% | 28% |
| Turbo 1-Step Balanced | 73% | 62–87% | 74 / 70 / 76% | 44 / 52 | 5% | 23% |
| Turbo 1-Step Conservative | 90% | 89–99% | 86 / 94% | 222 / 230 | 0% | 7% (3% unfinished) |
| Pro 1-Step Fast | 58% | 53–66% | 68 / 59 / 53% | 18 / 24 | 30% | 12% |
| Pro 1-Step Balanced | 76% | 65–88% | 75 / 74 / 79% | 44 / 52 | 10% | 15% |
| Pro 1-Step Conservative | 93% | 93–100% | 88 / 99% | 210 / 219 | 0% | 6% (1% unfinished) |
| Classic 2-Step Fast | 63% | 56–71% | 60 / 60 / 70% | 20 / 22 | 7% | 30% |
| Classic 2-Step Balanced | 87% | 79–97% | 96 / 89 / 81% | 49 / 53 | 3% | 10% |
| Classic 2-Step Conservative | 100% | 100% | 100 / 100% | 193 / 185 | 0% | 0% |
Days to pass count passing starts only. The Conservative rows cover 2024 and 2025 starts.
What is in each plan, described generically: the Fast plans mix three or four strategies (4-hour return-direction and momentum strategies on PEPE, WLD, ETH and AVAX) at 1.4–2x their recipe size. The Balanced plans run one or two strategies (a WLD 4-hour strategy, sometimes with DOGE intraday momentum or an ETH volatility breakout) at 0.5–1x. The Conservative plans run an ENA swing-momentum strategy and an OP volatility-breakout strategy at 0.7–1x.
What patterns stand out?
Speed and pass rate trade off cleanly. Every Fast plan passes about 58–63% of starts with a median around 20 days. Balanced plans pass 73–87% in roughly 44–53 days. Conservative plans pass 90–100%, but their medians run from 193 to 222 days. The speed vs pass rate post digs into that choice.
The failure mode depends on the rules. On Classic and Pro 1-Step, the 3% daily limit does the damage at aggressive sizes: 36% of Classic Fast starts and 30% of Pro Fast starts failed on daily loss, against 4% and 12% on drawdown. On Turbo, whose 3% static drawdown is as tight as its daily limit, drawdown dominates (28% of Fast starts). On the 2-Step, with its 5% daily limit and 8% trailing floor, drawdown also dominates (30% of Fast starts). At Balanced sizes, drawdown fails more starts on Classic and Pro too (2% versus 11%, 10% versus 15%). Our post on the daily loss limit explains when it binds first.
Later start years are not uniformly worse. Some plans decay from 2023 to 2025 starts (Turbo Fast from 68% to 52%), others hold or improve (Classic 2-Step Fast from 60% to 70%). Year-to-year swings of this size are a reminder of how much the starting period matters.
Why aren't 798 starts 798 independent tries?
A challenge started on Monday and one started on Tuesday share almost the whole path after Tuesday. If a big rally on day 10 carries Monday's start to the target, it probably carries Tuesday's too. So 798 daily starts behave like a much smaller number of genuinely independent attempts, and a pass rate built from them is less precise than the count suggests.
The 90% intervals in the table come from a block bootstrap, which resamples whole stretches of starts to respect that overlap. Even so, it understates the uncertainty, because it cannot account for the second problem.
That second problem is selection. These plans were in-sample: they were chosen with the 2025 data visible. When you pick the best of many candidates on the same data you score them on, part of what you measure is luck, as our post on selection bias in backtests explains.
What happens out of sample?
To estimate the real number, our team ran an out-of-sample study in September 2026. We picked candidate strategies using only 2020–2023 challenge starts, then scored them on 2025–26 starts they had never seen.
| Out-of-sample result | Pass rate |
|---|---|
| Picks scored on a 180-day window | about 25–34% (median ~30%), even when the backtest said 40–80% |
| One-year window, small sizes, passing in 2–5 weeks | ~34–45% |
| One-year window, small sizes, passing in 5–12 weeks | ~54% |
| Very fast: aggressive SOL breakout sizing, pass in under a day | about 7% |
| Very fast: daily SMA50 trend on BTC/ETH at high size, 1–2 days | about 22–25% |
Two lessons follow. First, a realistic planning figure for the slowest, smallest-size tier is 40–55%, not 90–100%. Second, the fast "lottery" options held up out of sample, but at low rates: they are honest about being long shots.
We also tested a published community bundle that claimed around 94%. Under Propr's real rules it passed 0–44% depending on size and never passed a 2025 start within a year. Claims without a method are worth very little.
How do the results compare with the zero-edge ceiling?
If your trading has no edge at all, your chance of hitting the target before the drawdown floor is roughly drawdown ÷ (target + drawdown), whatever your size. That is the gambler's ruin result.
| Challenge | Zero-edge ceiling |
|---|---|
| Classic 1-Step | 6/(10+6) ≈ 37.5% |
| Turbo 1-Step | 3/(9+3) = 25% |
| Pro 1-Step | 5/(12+5) ≈ 29.4% |
| Classic 2-Step | about 32% simulated (formula: 8/(10+8) ≈ 44%) |
For the Classic 2-Step the no-edge odds are about 32% in our simulations; the simple dd/(target+dd) formula gives 44%, but the trailing floor follows the peak until it locks at the starting balance, which lowers it. The out-of-sample results cluster near these numbers, which is the sobering part: much of what looks like skill in a backtest is the base rate any coin-flip strategy would get. The odds math post works through this in detail.
What is the live record so far?
Each plan also runs forward challenges on its strategies' paper record, which started on 2026-09-30. As of 2026-10-05, every plan has 6 challenges running, 0 breached and 0 finished. There is no live pass rate yet. A live pass rate appears once 90 forward challenges have run their full 365-day window, so not before late 2027. A plan is hidden if that rate is under 50%, or earlier if more than half of at least 90 started challenges have already breached. We will publish it on the plan pages as it arrives.
Which number should you plan with?
Use the backtest pass rate to compare plans with each other, and the out-of-sample figures to decide whether a fee is worth paying. For a Classic 1-Step attempt, our planning estimate works out to about $390 per pass; for a fast Turbo 1-Step attempt it is about $200–220 per pass ($50 fee at roughly 23–25% odds). Budget for several attempts, not one.
How Trigr fits in
Every plan above is offered on Discover › Propr while its strategies stay published on the measured version, with its backtest pass rate, start-year range, days to pass, failure breakdown and live record side by side. When you deploy your own agent on Propr, a similar daily-start replay runs as a rules check against your account (a house plan skips this step: Trigr shows the plan's measured record and sizes every slot itself), so you can see how your strategy would have fared before paying a fee. The backtesting docs explain how costs enter the engine. Past results do not predict future ones, and the honest number is usually the smaller one.