Research & Methodology

1. How this is run

PropSurvival is a focused research-and-engineering operation with a systematic-trading and quantitative-risk background: the same discipline that writes the simulation engine reads the firm documentation, builds the rule data, and stands behind the result. There is no outsourced pass-rate scraping, no forum-sourced numbers, and no third-party ranking service reselling a "best firm" list under this name — every claim on this site traces back through the one process described below, not through a marketing team's summary of it. Questions, corrections, or a rule that looks wrong: support@propsurvival.com.

2. What "verified" means here

Every modeled rule — a profit target, a daily loss limit, a drawdown mechanic, a consistency clause — is transcribed from that firm's own published documentation, not from a forum post, a review site, or a competitor's summary. Each rule set carries a structured provenance record alongside its values: for the fields it covers, a source URL, the exact quote that supports it, the date it was retrieved, and a confidence label (sourced, derived, reconciled, or ambiguous) describing how directly the firm's own words support the stored value. That coverage is not uniform field-by-field — a dollar figure that is exact arithmetic on an already-cited percentage carries its source through that percentage rather than as a second copy of the same quote, and a handful of structural or not-yet-independently-cited fields are still catching up — but it is real and checkable, not decorative: open any rule file and the record is there to compare against the firm's own page. A rule that changes on the firm's site is a rule that goes stale here until it is re-read — see the changelog for the dated history of every re-read.

3. When a firm's own documentation does not settle it

Firms sometimes describe the same rule two different ways, or publish a value that admits more than one reading. When that happens, it is recorded rather than guessed past: each ambiguous field lists the candidate readings and states which one the model uses, with the reasoning attached. The default stance is conservative — the reading that makes passing harder, so the model never quietly flatters an outcome. One real example from the rule data: Topstep publishes its daily-loss-limit reset only as a local time ("5 PM CT"), which maps to two different UTC values depending on the time of year the reset falls in; both candidates are on record, and the later one is what the model uses. A second field in that same rule set is disclosed the other way around — a stricter reading is on record as the more accurate one but deliberately left un-applied, because switching to it would silently disable an unrelated check elsewhere in the simulation engine. That trade-off is written into the rule data itself, not smoothed over.

4. How a number gets checked twice

Beyond the initial read, rule sets are periodically re-verified with a two-pass blind method: the rule is re-derived independently, without looking at the stored value first, and only then compared against what is live. Where the two passes agree, the field is confirmed. Where they disagree — including cases where a firm's own page contradicts itself between two explanations of the same rule — that disagreement is recorded, not silently resolved in whichever direction is more convenient. Every recheck is dated.

5. Corrections

Corrections happen. The changelog is the dated record of every one — rule re-reads, pricing-language fixes, engine corrections that changed a number. The practice is a dated changelog entry describing what was wrong and what changed, not a silent edit to the original text: a log that gets rewritten after the fact stops being a log. If you find a rule that no longer matches a firm's own published page, email support@propsurvival.com with a link to the firm's documentation — confirmed corrections are credited there.

6. What this product is, and is not

PropSurvival is independent analytical software. It is not financial, investment, or trading advice; not trading signals; not a broker; not a prop firm; and not affiliated with, endorsed by, or sponsored by any prop firm. It publishes no pass rates, rankings, or comparative claims about any named firm — firm rules are facts about published documents, and how a given trader fares under them is computed locally, from that trader's own statistics, never asserted by us about a firm's business.

There is no advertising on this site and no affiliate placement, and no firm pays for a listing or a favorable reading of its own rules — the network requests this product actually makes are named in full on the privacy page, and a rule corpus paid for by the firms it describes would not be independent analysis.

7. Data and reproducibility

Simulations use a seeded, deterministic pseudo-random generator, so a published figure — given its seed and path count — is exactly reproducible, not just directionally similar. Where original research is published, such as the static-vs-trailing drawdown study, the dataset behind it ships as machine-readable JSON under a CC BY 4.0 licence: reuse it, quote it, check it, with attribution.

8. Machine access

This site's robots.txt explicitly allows the major AI and search crawlers by name — GPTBot, ClaudeBot, Claude-User, PerplexityBot, Google-Extended and others — and llms.txt gives any model a curated summary of what is published here and what it means. The rule data is written to be checked, not just read.

9. Contact

Questions about a rule, a correction, or the methodology itself: support@propsurvival.com.