The State of Prop-Firm Drawdown Rules: a census of 178 configurations across 43 verified rule sets
Most prop-firm drawdown floors move. Free calculators still model the ones that don't.
And "moves" is not one rule wearing different clothes: whether the floor ratchets on every tick, only at each day's close, or only when a trade closes changes what a losing position in progress can do to it.
This is a census, not a simulation: every number below is a count of published clauses across the full verified rule corpus — 43 rule sets, 178 account-size configurations — computed directly from rules/** by a build script, never hand-typed. It publishes no pass rate and ranks no firm; the subject is the shape of the rules themselves.
Key findings
65.2% of configurations (116 of 178) use a floor that moves. Of those, end-of-day trailing is the most common single archetype at 42.7% of the whole corpus (76 configurations, 22 rule sets) — more common than the static floor most free calculators assume. Intraday trailing, the variant that ratchets on unrealized profit and is checked continuously, is 15.2% (27 configurations, 6 rule sets). Closed-balance trailing — the least common and structurally most forgiving trailing variant, since an open position cannot move the floor or trigger a breach — is 7.3% (13 configurations, 2 rule sets).
A static floor does not mean a forgiving one. Of the 62 static-floor configurations, 80.6% (50) are still checked against live, unrealized equity — an open drawdown can still breach a static floor before you close the position. Only 14.5% (9) test exclusively against closed-trade balance, and 3 do not record a checking basis at all. "Static" describes whether the floor's own value moves, not whether an open position is safe.
A daily loss limit is common but not universal, and a consistency rule is a minority feature. 67.4% of configurations (120 of 178) carry a daily loss limit; the rest, including 3 explicitly optional at checkout, do not. 25.8% (46 of 178) carry a consistency/best-day rule on top of the drawdown floor. Neither is implied by the other — the drawdown mechanic alone does not tell you the whole rule set.
Re-checking the corpus surfaces real disagreement, and this window's corrections were ours. 53.6% of recheck entries (111 of 207) came back with at least one live two-reader disagreement, which the corpus records but does not silently resolve. Separately, in the 2026-07-22 to 2026-07-31 recheck window, 28 of 207 entries (13.5%, across 8 of 43 rule sets) were logged as "changed" — and every one, on inspection, was a correction to our own prior transcription or a disclosure clarification, not a firm changing an already-published rule. See Table 2.
Table 1 — Drawdown-mechanic archetypes across the corpus
Classified by two published fields per configuration: whether the floor's own value moves (type), and, where it does, what equity value moves it (basis). A third field, the cadence at which a breach is actually tested, is disclosed separately in Methodology — its provenance was not consistent enough across the corpus to carry a headline of its own.
| Archetype | Configs | Share of 178 | Rule sets | How the floor behaves |
|---|---|---|---|---|
| ArchetypeStatic floor | Configs62 | Share of 17834.8% | Rule sets13 | How the floor behavesFixed at a set distance from the opening balance; never moves |
| ArchetypeIntraday trailing | Configs27 | Share of 17815.2% | Rule sets6 | How the floor behavesRatchets on every new equity peak, including unrealized profit |
| ArchetypeEnd-of-day trailing | Configs76 | Share of 17842.7% | Rule sets22 | How the floor behavesRatchets only at each day's close; intraday swings don't move it |
| ArchetypeClosed-balance trailing | Configs13 | Share of 1787.3% | Rule sets2 | How the floor behavesRatchets only on realized balance; open trades don't touch it |
Table continues → scroll horizontally
178 published configurations. How does each one's floor behave?
116 of 178 move in some way. Only the closed-balance variant — the smallest slice — ignores an open position entirely.
Bar length is each archetype's share of all 178, read from Table 1. Classified from each configuration's own published type and basis fields · rules verified 2026-07 · no pass rates.
Why there are no pass rates in this table. A drawdown mechanic is a fact about a document a firm publishes, and we cite it. A pass rate is a claim about that firm's business, and we do not publish one — however it was computed. The mechanic is the subject; the firm is a configuration of it. The free engine computes your own probability from your own statistics, locally, and tells no one.
"43 rule sets" counts directories, not businesses. 7 operators publish more than one rule set in this corpus — for example Apex Trader Funding (2: intraday and end-of-day products) and Alpha Futures (3). 43 rule sets trace to 33 distinct operators. No single rule set spans more than one archetype across its own account sizes.
Table 2 — What the 2026-07-22 to 2026-07-31 recheck window found
207 recheck entries were logged in this window across all 43 rule sets. 28 of them, across 8 rule sets, came back with an outcome of "changed." Every row below is one of those 8 — what changed, and whether it was a firm's rule or our own prior reading.
| Rule set | Recheck date(s) | What the recheck found | Kind |
|---|---|---|---|
| Rule setAqua Futures | Recheck date(s)2026-07-22, 2026-07-30 | What the recheck found25K daily loss corrected from our $625 (a flat-ladder assumption) to the firm's published $600. | KindTranscription corrected |
| Rule setEarn2Trade | Recheck date(s)2026-07-22 | What the recheck foundDrawdown-lock field corrected: our prior reading held it null ("never locks"); both independent readers found the firm's own documentation states verbatim that the threshold stops moving once equity reaches the starting balance. | KindTranscription corrected |
| Rule setFTMO (1-Step) | Recheck date(s)2026-07-31 | What the recheck foundDaily-loss reset time unified with FTMO's own 2-step product: 22:00 UTC corrected to 23:00 UTC. | KindTranscription corrected |
| Rule setPhidias (Fundamental) | Recheck date(s)2026-07-22 | What the recheck found100K profit target corrected from 8% ($8,000) to the firm's published 6% ($6,000); a 50K quote had been carried across sizes in error. | KindTranscription corrected |
| Rule setTake Profit Trader | Recheck date(s)2026-07-22 | What the recheck foundMinimum trading days corrected from 0 to the firm's published 5-day requirement, across all five account sizes. | KindTranscription corrected |
| Rule setTradeify | Recheck date(s)2026-07-30 | What the recheck foundThe 35% consistency rule was found to apply to the funded/payout stage only, not the evaluation this corpus models; the evaluation-stage field was cleared. | KindTranscription corrected |
| Rule setMaven Trading | Recheck date(s)2026-07-31 | What the recheck foundA targeted re-read confirmed the existing trailing-5% reading was correct; an apparent site self-contradiction turned out to describe a different Maven product. | KindDisclosure clarified, no value change |
| Rule setTopstep | Recheck date(s)2026-07-31 | What the recheck foundAn unresolved reading from a prior check was formally logged as a disclosed ambiguity. The live value did not change. | KindDisclosure clarified, no value change |
Table continues → scroll horizontally
6 of the 8 rows are corrections to PropSurvival's own prior transcription (a copy-paste error, a stale assumption, a missing field); 2 are disclosure bookkeeping with no value change. This describes one window at one corpus, not a claim about how often any firm changes its rules — firms may do so at any time without notice, and this page states no monitoring cadence.
The inconvenient number: two-reader disagreement is common, not rare. 53.6% of all 207 recheck entries in this window (111 of them) came back with at least one field where two independent blind readers disagreed. Per the two-reader doctrine, a disagreement is recorded and left unresolved rather than silently arbitrated — 55.8% of rule sets (24 of 43) currently carry at least one such open flag. This is disclosed, not smoothed over, because a corpus that hides its own uncertainty is less useful than one that states it.
Methodology — how every figure above was produced
- Corpus 43 verified rule sets, 178 account-size configurations, each field sourced with a URL, a verbatim quote, a retrieval date and a confidence rung (reconciled / sourced / derived / ambiguous / unknown) recorded in
rules/<slug>/<size>.json. Field-level citability across the corpus: 21 point-cited, 22 published as a range, 0 neither. - Generation Every count on this page is computed at build time by
tools/build-drawdown-landscape.jsreadingrules/**directly — no figure here is typed by hand, and a mechanical self-check regenerates the dataset and confirms every numeral on this page traces to it before release. - Classification The Table 1 archetype is
type(does the floor move) crossed withbasis(what equity value moves it, when it does). A third field,evaluation_frequency— how often a breach is actually tested — exists in the corpus but was excluded from this classification: a spot check of the 9 configurations where it disagreed with whatbasiswould predict found its own provenance rung "ambiguous" or "derived from basis" for 6 of the 9, not solid enough to carry a published finding. - Two-pass blind recheck Rule sets are periodically re-verified by two independent readers who do not see the stored value or each other's output first. Agreement confirms a field; disagreement is recorded as a live flag, never silently resolved in whichever direction is convenient. Every recheck is dated in
rechecks[]; Table 2 is a complete accounting of the 2026-07-22 to 2026-07-31 window's "changed"-outcome entries, not a curated sample. - Denominators Every count on this page states its own denominator (178 configurations, 43 rule sets, or 207 recheck entries, as applicable) rather than a bare percentage — a config-level cut and a rule-set-level cut answer different questions and are never presented as interchangeable.
- Limitations "43 rule sets" counts verified directories, not distinct businesses — 7 operators publish more than one. The recheck-window findings describe that window only, not an ongoing schedule. Firms change published rules at any time without notice; each firm's own current documentation is always the final authority, and a rule confirmed here on a given date can be different today.
Data license: CC BY 4.0 — reproduce these figures freely with attribution to PropSurvival and a link to this page. The full dataset — every count, denominator and the complete recheck-window accounting — is published as machine-readable JSON: research-drawdown-landscape-data.json.
These are counts of published rules. See what they mean for your own trading.
Type your win rate, average R and risk size — or import your trade CSV — and get your pass probability, dominant failure cause, expected attempts and fee spend under any of these rule sets' real, current rules. Free, no signup, computed entirely in your browser.
Run your numbers against all 43 rule sets — freeNothing you enter leaves your device — methodology. The mechanic isolated and simulated: static vs trailing drawdown study. Prefer to see it? Scroll the interactive tour.
PropSurvival is independent analytical software and is not affiliated with, endorsed by, or sponsored by any firm named above. Rule descriptions reflect our verification dated above; firms change rules without notice — each firm's own documentation is always the final authority. Nothing here is investment advice.