Watch five thousand futures hit a written rule
Set the scene for
your survival.
On this page’s counted 5,000-path illustration, the trailing floor is the clause that closes the accounts first — then run your book.
Numbers first: 59 firm rule sets / 243 account-size configs · funded death 14.2%–75.2% · two-step 64.9%→70.5%. Answer first: the picture is the engine — a written floor, a mass of futures, the clause that binds first. Same arithmetic on your book, or on a firm’s dated one. Nothing is uploaded.
Working backwards: free 2,000-path run against your book or 59 firm rule sets — nothing uploaded.
Watch what actually closes the accounts · 5,000 paths · not your CSV
Illustrative Monte Carlo fan of five thousand typed-average futures against one prop rulebook — not your CSV, not a ranking. Mist paths stop where a rule closes them or where they reach the profit target; the dark line is one equity path; the wine line is the trailing floor that ratchets up and never down.
In this specimen, binds first: your floor. Of 5,000 illustrative paths, most that end hit the floor you named. Not a pass rate. Not your CSV.
ILLUSTRATIVE · REACH THE TARGET · THIS PROP SPECIMEN
ILLUSTRATIVE 5,000 PATHS · SEED 41 · NOT YOUR CSV · YOUR RUN USES YOUR NUMBERS
Quick answer: A trading account ends on a written rule — drawdown, daily loss, consistency, horizon — not on strategy. PropSurvival compiles the book you wrote, or a firm’s published clauses (currently 59 firm rule sets / 243 account-size configurations), into arithmetic, runs your own trade history against it entirely in the browser, and names the constraint that binds first. Free calculators shuffle typed averages against a static floor; this applies the clause as written. The dated public receipt for every firm last-read is the Rule Changelog. The free verdict stays free forever; four paid documents ($29 / $49 / $99 / $249 — see pricing and the homepage FAQ) deepen an account already in progress without changing free accuracy. First-party corpus × pricing depth publishes the 59/243 census beside that ladder. No signup and no card; your trades never leave the device. Run your book — free. Three studies open that path: funded account rules (14.2%–75.2% funded death among profiles one evaluation grades within 2%), two-step challenge probability (64.9%→70.5% conditioned), and broker export to expectancy (51.5% vs 65.2% win rate on the same $605.50).
How it works
It reaches a conclusion, and shows every step to it.
Five steps, in order. Each one is on the screen, so the verdict is something you can argue with rather than something you have to trust.
What is true about the account right now
Balance, high-water mark, equity, and the trades you actually took — imported from your platform’s CSV or entered by hand. Nothing is assumed that you did not supply.
The clause, as written — yours, or a firm’s
Your stated limits, a crypto book, or a dated reading of a firm’s own document — never paraphrased from a forum. Where the wording leaves a field undecided, the run records it as unresolved instead of choosing for you.
Five thousand paths against every binding clause
Trailing drawdown, daily loss, consistency cap and minimum days applied together — an account can survive each rule separately and fail the combination. Run from a stated seed, in your browser.
One number, and the clause that produced it
The modelled share of paths that respect every stated rule, and the constraint that ends most of the accounts that end — dated with the book, the rule-read date, the paths and the seed, so the same answer is still there tomorrow.
What the model does not know
Your win rate and your R are figures you supplied, not facts anyone measured for you — so the run states what it assumed, and the sensitivity view shows how far the answer travels when they are wrong.
Before you run the book
Three answers you do not have yet.
The rule corpus
Rules change. We keep the dates.
Anyone can list firms. The durable edge is knowing when a clause changed — because a figure computed against last quarter’s rulebook is a wrong figure, however confidently it is drawn.
Topstep’s own maximum-loss-limit article, read 2026-07-16, in three of its own sentences: “The MLL is a trailing limit” that “rises as your end-of-day balance grows” and, once it has risen, “never moves down.” Not our paraphrase — their wording, quoted.
Source: help.topstep.com/en/articles/8284204-what-is-the-maximum-loss-limit · last changed 2026-07-17 · last read 2026-08-16 · source sentence not found 2026-08-16
Every mechanic in this corpus is published per firm in the firm rules index, and the paid documents are built from the same dated corpus — read a full sample Case File before you buy one.
Each clause carries two dates. Last changed is when the number itself moved. Last read is when we last retrieved the firm's own sentence from their page — even if the number did not move. A look that cannot see the page does not advance the date.
Genuinely undecided clauses are recorded as unresolved rather than filled with the plausible answer, and the run says so next to the verdict — a number that quietly assumed the wrong basis is worse than no number.
We hold no affiliate relationship with any firm in this corpus, take no fee for inclusion, and publish no ranking of one firm over another. The table beside this card is a record of dates, not an endorsement.
| Recorded change | Read |
| Published firm count corrected to 59 / 243 | 2026-08-22 |
| e8-futures, top-one-futures added | 2026-08-17 |
| lark-funding added | 2026-08-16 |
| elite-trader-funding-eod, ment-funding-futures added | 2026-08-16 |
First-party depth · computeCorpus() × TIERS
59 firm rule sets · 243 configs · paid ladder from $29.
The certificate below is the live homepage citation unit for corpus census cells beside the TIERS paid ladder. Click it to open the free /app run — the same engine that names which written clause binds first. Dataset JSON and CC BY 4.0 licence sit under the panel. Census and ladder facts only — never a firm ranking.
Inconvenient, and true to the data: firm-count 59 is not ruleset-count 243 — conflating them invents a third figure. Coverage date 2026-07-22 is older than tip date 2026-08-27; dating the whole corpus to the tip silently re-dates firms last read on the older day. The free TIERS blurb omits a firm count on purpose. Paid terms are asymmetric: $29 and $49 are one-time; $99 and $249 renew yearly.
Citable partition: 30 point-cited + 29 range-only = 59 firm rule sets (neither = 0). This block publishes census and ladder facts only.
Where your data goes
Nowhere. Airplane mode is the test.
A trade history is the most revealing file a trader owns, and checking a rule should not cost you one. Disconnect the network, run a full analysis, and watch every number still resolve — not a promise to take on faith, a test you run yourself.
What leaves this page, and what never does.
Your trades
Parsed in the browser. Never transmitted.
Account
No account required.
Login
No sign-in. No session.
Registration
None.
Nothing that identifies you leaves this page — only a licence check and the information you explicitly provide.
The ladder · free tier, then $29 / $49 / $99 / $249
Four questions, in the order traders actually ask them.
The free tier is the whole instrument, not a trial — the same arithmetic, the same rule corpus, and no accuracy gate. What you buy is a different question, asked from inside an account you already own.
Paying changes the question, never the arithmetic.
Your first run
Three numbers is enough to start.
You do not need a CSV to get an answer. Pick a book — yours, or a firm's — set three figures, and read the verdict — then import your real trades when you want the precise one.
Choose Personal, Crypto, or a prop-firm rulebook
Your limits, a crypto book, or a named firm. The rulebook loads with the door — own rules, or trailing basis, daily loss, consistency cap, minimum days.
Set three figures, by dragging
Risk per trade, win rate, average win over average loss. Drag them sideways — or tap any figure and type it exactly.
Read the verdict, then import the real trades
A CSV from your platform replaces the estimates with what you actually did. The arithmetic does not change — only the inputs get sharper.
Three numbers is a start. A CSV is your book. The 5,000 paths above are neither — they are illustrative. Import stays on this device.
Organic studies · Numbers first · Answer first
Three published answers that open a free run.
Numbers first: funded death rates 14.2%–75.2% among profiles one evaluation grades within 2% · two-step pass 64.9%→70.5% once you condition instead of multiplying · broker export win rate 51.5% vs 65.2% on the same $605.50. Answer first: each study below opens one free /app path — the published figure is the reason to run, not an adjective. Run your analysis — free — no signup, no card.
- Funded account rules — 14.2%–75.2% funded-year death among near-identical eval profiles · Run this study's numbers — free
- Two-step challenge probability — 64.9% product is the floor; conditioning lifts to 70.5% · Run this study's numbers — free
- Broker export to expectancy — same 91 fills read 51.5% as trades vs 65.2% as rows (13.7 pp) while both rebuild $605.50 · Run this study's numbers — free
Working backwards: a free verdict names the clause that binds. Run your analysis — free.
FAQ · Numbers first
Answer first: is this just another prop calculator?
Numbers first: funded death 14.2%–75.2% · two-step 64.9%→70.5% · 59 firm rule sets / 243 configs. Answer first: the open question below is the high-intent one — free calculators shuffle averages against a static floor; this applies the written clause. Working backwards: free path before another typed average — Run your analysis — free.
Is this just another prop calculator?
Free calculators take typed averages and shuffle them against a static drawdown. This engine imports your actual trade history, applies the written rulebook — your limits, or a firm’s real drawdown mechanics (intraday trailing vs. EOD vs. static, freeze-at-initial, consistency rules) — decomposes failure by cause, and computes the risk band that changes the outcome. Run the free analysis — nothing is uploaded.
Can a simulation really predict whether I'll pass?
No. Monte Carlo gives you the probability distribution your statistics imply under the firm's rules, assuming your future resembles your sample. What it does well: showing when the fee is mathematically premature, and which lever moves your odds most.
What happens to my trade data?
It is parsed and stored inside your browser only. There is no server that could receive it. Two things can leave the device, and only these two: a licence key if you activate one, and an email address if you type one into a form.
Is PropSurvival free?
The verdict engine is free and always will be — every firm in the corpus, a 2,000-path Monte Carlo with no accuracy gate, the Firm Fit comparison, CSV trade-history import, and your run history, with no signup and no card, entirely in your browser. Four paid tiers ($29 / $49 / $99 / $249, see pricing.html) add deeper documents for an account already in progress; none of them change the free verdict's accuracy.
Can I reproduce a result later?
Yes. Every run records the firm, the date that firm's rules were read from its own documentation, the number of paths and the seed. Given the same inputs and the same firm rules, the model returns the same numbers. That is a property of the engine, not a claim about the market.
What happens when a clause is ambiguous?
It is recorded as unresolved rather than guessed. Where a firm's own worked example does not settle a field — for instance whether a drawdown floor tracks the intraday equity peak or the end-of-day balance — the run says so next to the verdict and points you back to the firm.
Cite-embed · Numbers first · SMTP-less
Answer first: paste the live corpus, or embed the calculator.
Numbers first: 59 firm rule sets / 243 account-size configs · funded death 14.2%–75.2% · two-step 64.9%→70.5%. Answer first: directories get a copy-ready blurb with the live corpus figures — no signup, no outbound send. Working backwards: copy the blurb below, or Run your analysis — free.
Read next · first-party findings, each dated
Same fills, two win rates: 51.5% vs 65.2% articles/broker-export-to-expectancy.html · published study 2026-08-26 Where the trailing floor stops following apex-trailing-drawdown.html · §freeze 2026-07-22 The consistency cap, with the arithmetic topstep-consistency-rule.html · §rule 2026-07-14 See what $99 buys, before you pay sample-case-file.html · full sample, real engine 2026-07-31 All research and explainers, grouped articles/index.html · the full library 2026-07-28