Everything we changed, dated — scores, rules, statuses, payouts, reserves

Partial re-read: GT Funded's copy grows to $500K/90%, FundedPoly 402 at 63 days, Hypernova reserve $600,015 — no status change

Four of the ten tracked sites re-read in full on 26 September 2026; the other six covered by that morning's automated DNS/HTTP sweep. GT Funded: www.gtfunded.xyz answers 307 → 200 and its page metadata now advertises …

Hypernova reserve down $169,194.50 in four days: $958,399.39 (15 Sep) → $789,204.89 (19 Sep)

Daily 03:00 UTC on-chain readings of Hypernova's payout reserve on Arbitrum (vault + reserve wallet): $958,399.39 on 15 September, $925,592.74 on 16 September, $870,858.82 on 17 September, $849,504.79 on 18 September,…

Propr → us: $7,670.94 — Every payout our own account ever received — 48 transactions, each reconciled on Ethereum to the cent

48 payout requests between 20 May and 18 Sep 2026 ($9,588.65 requested, 80% split), read on the account's payout history and checked one by one on a public Ethereum RPC: receipt status OK and a USDC Transfer to one of…

Propr → us: $584.27 — Two payouts from our $200K Diamond funded account, 80% to the cent, settled in 13 and 43 minutes

We bought the largest product Propr sells — the Diamond 1-Step Classic, $200,000 — on 2026-09-14, passed it on 2026-09-18 and took two payouts the same day. Requested $444.23 and $286.11; received 355.3822 and 228.887…

Propr: assets are delisted below $1M open interest / $500K daily volume, and a delisted market goes reduce-only — stated by the CEO on X, absent from the rulebook

Two practical consequences. First, your catalogue is filtered twice: by what Hyperliquid lists, and then by Propr's own liquidity floor of $1M open interest and $500K daily volume — the rulebook only tells you about the first filter. Second, and this is the one that costs money: if a market you hold falls below that floor, it goes reduce-only, meaning you can close and only close — no adding, no reversing, no re-entering after you exit. On a funded account whose drawdown is measured against a balance you have to defend, that removes your ability to manage the position on your own terms, and it is triggered by the market's liquidity rather than by anything you did. None of this is unreasonable risk management — a dried-up perp is a genuine hazard, and $1M/$500K is not a harsh bar. The finding is narrower: a criterion disclosed in a reply to one trader is not a published rule, it can move tomorrow with no changelog, and anyone who bought a challenge last week has no way of knowing it exists. If you run a Propr funded account on a small-cap perp, watch open interest and 24h volume on Hyperliquid itself, where both are public per market.

Correction: three days of false 'no DNS record' readings withdrawn — full hand re-check of all 10 firms, no status change

From 13 to 15 September 2026 our automated daily sweep recorded 'dns: no-record' for every tracked domain, live firms included, and that reading reached registry.json and the register table. It was false: the sites we…

Hypernova reserve read: $977,923.42

Automated chain read (block 504954779). Wallet unchanged to the cent; vault −$885.23 in 12 hours (payout flow — the stats page lists a $562 payout dated 13 Sep). Stats page displayed $988,363 the same morning, +$10.4K over the chain. Lifetime payouts counter $822,890, up $91.5K since 7 Sep while the vault fell only $5.0K over the week: ~$86K entered the vault from an untraced source in the window.

Hypernova reserve read: $978,808.66

First automated chain read (block 504778439). Reserve wallet net +340,000.00 USDC since 7 Sep (to within 0.2 cent), vault −$4,145. The firm's 11 Sep post said '$440,000 added' and 'back to $1,000,000': corroborated in direction, not in amount — the wallet's NET movement is $100K short of the first figure and the total is $21.2K short of the second. Whether ~$100K also left the wallet in the same window is untraced.

Carrot: the evaluation breached on the daily limit — floating loss confirmed, score deliberately unchanged at 73.2

On 10 September 2026 we published one thing as untested and refused to infer it: what Carrot's balance-based Objectives counter displays while a position floats at a LOSS, which is the case that actually matters. The account answered the question the same evening by dying, and the dashboard read on 11 September carries the whole sequence. Two BTC/USD longs at 5x, no stop loss, about $24,930 of notional on a $5,000 account, held overnight. BTC moved roughly 1.1% against them. Both were force-closed at 19:30:42 at the same price, 77,420.0, and the day closed 4 trades at -$288.78: $5,000 to $4,711.22. THE CHART IS THE PROOF. BALANCE stayed flat at 4,986.169 the entire way down while EQUITY fell 5,054.693 -> 5,018.706 -> 4,922.21 -> 4,917.752 -> 4,879.217 -> 4,839.727 -> 4,836.86, then dropped vertically to 4,711.225 on the forced close. The realized balance never moved more than $13.83, so it is the equity that crossed the 4,750 daily floor and the floating loss that took the account — exactly as §05 and the public FAQ state, and exactly against the §24 cheatsheet cell reading "Balance-based (does not include floating P&L)". Our 10 September reading is confirmed end to end, and that cheatsheet cell remains the one wrong line in the documentation. TWO FURTHER FACTS, BOTH OBSERVED. The daily floor sat at 4,750 and the account settled at 4,711.22, i.e. $38.78 THROUGH the stated limit: the forced close slipped past it. And the Max loss counter stopped at $288.78 of $500.00, so $211.22 of the total drawdown budget was never used — on this product it is the DAILY limit that ends accounts, not the total one. ONE FIGURE IS DERIVED RATHER THAN PHOTOGRAPHED and is labelled as such, because we have no capture of the counter mid-drawdown: since the counter reads 5,000 minus the balance, and the balance was frozen at 4,986.169, it would have displayed $13.83 of $500.00 while equity sat at 4,836.86, roughly $87 from death. It is arithmetic, not an observation, and it is published as arithmetic. ATTRIBUTION, because it decides how this is written: the breach is OURS, not the firm's. No stop loss was set and 5x of notional was carried overnight. The engine enforced a published rule correctly and the interface showed the floor in absolute dollars on screen the whole time. Nothing here is scored against Carrot, and the score does not move: 73.2/100, on the grid of 10 September 2026. The rules criterion was already raised to 78 that day on exactly this reading; a first-hand confirmation of a finding already priced in moves no criterion, and this grid does not re-score a firm for having been right twice. What IS reported, in figures and without adjectives, is the screen that follows a breach: "Challenge Failed", then 15% off a new challenge behind a countdown reading 10:16:12.

Carrot Funding: the daily limit is measured on EQUITY — confirmed by our own account breaching, and the §24 cheatsheet cell is still wrong

If you trade a Carrot evaluation, the number that can end your account is the EQUITY reading, floating P&L included, and the limit that will realistically end it is the DAILY one, not the total. Do not size from the §24 cheatsheet cell: it describes how the limit LEVEL is set, from the balance high-water mark per §07, and misstates what crosses it. Two operational consequences from our own case. First, the Objectives counter on the dashboard is BALANCE-based, so it can read almost untouched while your equity is a few dollars from the floor — watch the chart and the absolute dollar floor, not the counter. Second, a forced close is not guaranteed to land exactly on the floor: ours settled $38.78 below it. The breach itself is OURS and not the firm's — no stop loss was set and 5x of notional was carried overnight, the engine enforced a published rule correctly, and the interface showed the floor in absolute dollars on screen the whole time. Nothing here is scored against Carrot and the score does not move: 73.2/100, grid of 10 September 2026. Recorded in figures and without adjectives: the post-breach screen reads Challenge Failed, then offers 15% OFF a new challenge behind a countdown reading 10:16:12.

Two of our Carrot findings retracted the same evening — score deliberately held at 73.2

This entry retracts part of the two entries below it. They stay published, because this file does not rewrite its own history; read them knowing the corrections here supersede them. Before publishing two open questions about Carrot Funding, we put them to the firm. It answered the same evening and both of our findings were wrong. FIRST, THE VAULT. We had written that rulebook §19 promises an on-chain vault and that we could not locate it, and we published the formula 'the payouts are verifiable, the reserve is not' across the site. That was a reading failure: Carrot's GitBook documentation publishes a complete address table with named roles — Vault Custody Contract, Vault Integrations, a 2-of-4 payout approval multisig and the payout executor EOA — and we had not opened that page. Having opened it we checked it: the Vault Custody Contract on HyperEVM is a Safe holding roughly $15,016 in the Felix/Morpho USDC vault plus $98 on HyperCore. We were then about to publish that figure against the $788,521 of 'funding capital' shown on their Analytics page as a solvency gap. The firm's answer removes that framing entirely: funding capital is an internal risk metric — the capital still free before they stop accepting new funded traders — and not a reserve balance at all. Two different quantities; there is no gap. They add that funds are mid-migration from Arbitrum to HyperEVM, which is why the on-chain picture is partial. The address from which the four verified payouts leave is documented by them as a distribution EOA, so our description of it as an empty wallet holding 607.05 USDC against a large claim was unfair and has been removed everywhere. SECOND, THE NFT. We found no token #3955 in the Funded Bunny Initiative contract on Arbitrum — 551 tokens, IDs 0 to 550 — and were ready to publish that a challenge bought that day mints nothing. The firm answered that tokens are minted on-chain but that the token ID does not equal the challenge ID, because the product was not NFT-native at the start. That checks out: our challenge #3955 is token #550, minted on Arbitrum at 05:00:25 UTC on 10 September 2026, the hour of our purchase, with exactly one Transfer event since — the mint itself. What survives is narrower and still real: §20 and the FAQ both state that every challenge is minted as its own NFT, the mapping between challenge and token is documented nowhere, and the firm told us that clarifying it 'wasn't our priority'. That is a documentation defect confirmed by the source, not an absent product. WHY THE SCORE DOES NOT MOVE. Separately these pull in opposite directions — rules should fall, because a headline claim in §20 and the FAQ does not describe what the contract actually does, and survival should rise, because our reserve criticism was a reading error rather than a finding. Net movement is roughly a tenth of a point. A fourth revision of the same firm in a single day, for 0.1, would cost this grid more credibility than the precision would buy. The facts are corrected on every page today; the score waits for the next real input. TWO DATED COMMITMENTS, RECORDED AND NOT SCORED: on 10 September 2026 Carrot stated it will remove AI tooling from the §23 ban, and that it previously ran up to 200x leverage on gTrade and intends to raise leverage on Hyperliquid once comfortable. Neither is shipped. This grid has never scored a promise; freedom stays at 66 until the rulebook itself changes.

Carrot rules 72 → 78 (total 71.7 → 73.2) — we tested the contradiction instead of publishing it

Earlier the same day we marked Carrot's rules criterion down on one charge: that its rulebook contradicts itself on whether floating P&L can breach you, leaving a reader unable to compute their own drawdown. Rather than leave that standing, we tested it on the paid account. Raphael opened two BTC/USD positions and we read the live dashboard against them. Balance and equity are tracked as separate live values — 4,986.169 and 5,054.693, the 68.52 gap being the floating P&L — and the dashboard chart plots EQUITY in the same coordinate space as the two limit lines it draws at 4,750 for the daily loss and 4,500 for the maximum drawdown. The app shows you visually which of the two numbers is measured against the floor. The public FAQ then says it in words, before purchase and without a login: drawdown limits are calculated on equity and not balance, open losing trades count toward the limits in real time, and a breach fires the moment equity touches a limit even temporarily. Three sources agree — §05, the FAQ, and the app's own chart. Exactly one line disagrees, the §24 cheatsheet cell reading 'Balance-based (does not include floating P&L)', and that cell is describing how the limit LEVEL is set, from the balance high-water mark per §07, while misstating what crosses it. That is a drafting defect in a summary table, not a hidden trap, and we priced it as a trap. We also under-rated the interface itself: it publishes the limits as absolute dollars rather than percentages to interpret — 'Max loss: $4,500.00', 'High watermark: $5,000.00', 'Max daily loss: $4,750.00' — with a countdown to the UTC reset. Nothing to compute is the top of this criterion, not the bottom. Why the upgrade stops at 78: the test produced four new defects. The FAQ puts Maximum Loss at 10% from the high-water mark with no 1-Phase exception, against 8% in §03 and §24, so the contradiction moved rather than disappeared. The FAQ's forbidden-practice list adds copy trading and hedge trading, neither of which appears in §23 — two documents, two lists. The dashboard carries a 'verified by OASIS' badge while the FAQ describes the firm as still integrating ROFL, the same present-tense overstatement of an unfinished feature we docked HyperPNL for. And on that same screen the chart is equity-based while the Objectives counter is balance-based: it read 'Max loss $13.83 of $500.00', which is 5,000 minus the balance, while equity sat above the starting balance consuming nothing. One thing we could not test and do not claim: what that counter displays while a position floats at a LOSS, which is the case that actually matters. Testing it would mean losing Raphael's money on purpose, and we do not trade his accounts to manufacture a data point. It is recorded as unknown, not inferred.

Carrot Funding bought and traded — desk cap lifted, and four of our own numbers corrected (72.9 → 71.7)

We paid for a 2-Phase $5,000 evaluation on 10 September 2026 with our own money ($65 list, $52 with royaref, USDC on Arbitrum) and traded it the same day. `tested` goes to true. The 85-point desk cap was never binding at 72.9, so nothing below comes from lifting a ceiling. Start with our error, because it is the largest single correction in this revision. Our 5 September sheet published the 1-Phase $5,000 at $65 with a 5% daily loss and a 10% maximum loss, and documented no consistency rule at all. The configurator shown before payment, and rulebook §03 which is public and requires no login, both read $75, 4% daily, 8% maximum, and a Best Day Rule of 50%. We had copied the 2-Phase row onto the 1-Phase line and missed an entire rule. We also wrote that the fee is the same per account size whatever the structure; it is not — 1-Phase $5K is $75 against $65 for the 2-Phase. Carrot published all of this correctly and in the open. The mistake is ours, it is corrected in public rather than edited away, and it does not mark the firm down. (1) Rules 65 → 72. Three independent sources now agree line by line: the pre-payment configurator, the public rulebook, and the objectives actually applied to our live account — $250 profit target, $250 daily loss, $500 maximum loss on a $5,000 2-Phase. That is precisely the consistency whose absence cost HyperPNL points, and it earns some here. Carrot also discloses outright that it routes orders A-book or B-book at its own discretion (§22) and labels every trader on their own dashboard, a disclosure most firms in this table simply decline to make. What holds the upgrade at 72 is that the rulebook contradicts itself on the one number that decides whether you survive: §05 states equity limits apply to "both realized and unrealized losses", while the §24 cheatsheet states Max Loss is "Balance-based (does not include floating P&L)". §15 says Max Loss "stays the same" once funded while the cheatsheet shows it moving from 8% to 10% on the 1-Phase, and the §07 formula box multiplies the starting balance by the max DAILY loss. A reader cannot compute their own drawdown from this text, which is the criterion itself. (2) Payouts unchanged at 82. All four published payout hashes verify on-chain as genuine Arbitrum USDC transfers — 733.22, 384.38, 399.24 and 799.41, the dashboard rounding down — and published hashes that actually settle are not the norm here. We have not requested a payout ourselves, so there is no first-hand fact to move this in either direction, and the reserve question below is a solvency finding rather than a payment one. Same rule the grid has enforced since 12 August and that Raphael enforced again on 9 September: measurements move criteria, inferences do not. (3) Survival 68 → 55. The Points page, read 10 September 2026, states 279 traders in total and a weekly prize pool of $69; one $52 purchase placed us third on the weekly board. This is a materially smaller firm than desk research suggested. Separately, §19 promises that 50% of evaluation revenue flows into a vault and that the allocation is "verifiable on-chain" — we could not locate that vault. All four verified payouts leave the same externally-owned address, which carries no contract code, and it held 607.05 USDC against the $788,521 of funding capital published on the Analytics page. A distribution wallet is not a treasury and we do not claim the firm holds $607; the finding is narrower and worse for the Proof-of-Reserves claim than that: the payouts are verifiable, the reserve behind them is not. (4) Freedom 70 → 66. §12 is genuinely permissive on paper — no time limit, no minimum trading days, no stop-loss requirement, no minimum hold, weekend and news trading allowed — and Hyperliquid's central limit order book is real price discovery rather than oracle pricing. But two hard ceilings are now documented that we did not previously have: leverage caps at 5x across every asset class (§14), and §23 forbids "automated software, high-frequency trading systems, AI-driven tools, or mass order entry systems". A firm cannot headline trading freedom and ban automation in the same rulebook; Propr permits bots and APIs and keeps its 95. (5) Value 80 → 82. §17 claims Hyperliquid's own fee schedule with "no markup", and that is an arithmetic claim, so we measured it on two round trips: HYPE/USD cost $4.40 on $9,786.60 of round-trip notional, or 0.04496%, and ETH/USD cost $4.77 on $10,606.40, or 0.04497%, against a 0.045% base taker rate. Two assets, two directions, matching to the fourth decimal — the claim is exact, and this is the only measured fee verification in the table. Also creditable: no monthly, platform or withdrawal fees, an 80% split, a 100 USDC payout floor, and a challenge NFT the holder can transfer to another address. The caveat for readers is the arithmetic consequence rather than a criticism: a flat round trip still costs about 0.09% of notional. Our ETH position opened at 2,475.90, closed at 2,475.45, and finished at −$3.49. Net 72.9 → 71.7: a first-hand pass that confirmed the firm's documentation is accurate and found ours was not, while revealing an operation smaller and more restricted than desk research implied.

Carrot Funding: the 1-Phase prices and drawdowns we published were the 2-Phase ones — our error, not a rule change

The correction is ours, not the firm's: nothing in Carrot's rules moved between 5 and 10 September 2026. If you were sizing a 1-Phase from our sheet, the entry price is $10 to $100 higher than we printed and the room you have is smaller, not larger — 4% a day and 8% overall instead of 5% and 10% — with a 50% Best Day Rule on top, meaning a single day cannot account for more than half your total profit. The same read documents three ceilings we had never published: leverage caps at 5x on every asset class (§14), §23 forbids "automated software, high-frequency trading systems, AI-driven tools, or mass order entry systems" — so bots and APIs are out at Carrot — and §22 states outright that the firm routes each trade A-book or B-book at its own discretion, labelling every trader on their dashboard. Read §03 yourself before paying; it is free to read and it is the document that would settle a dispute.

Carrot Funding says it will drop AI tooling from the §23 ban and raise leverage on Hyperliquid — announced, not shipped, and not scored

Nothing changes for a live account today. If you automate, §23 still forbids it and you can still be breached for it; if you need more than 5x, §14 still caps you there. Treat both statements as intentions with a date attached, not as rules you can trade on, and re-read §14 and §23 yourself before assuming either has moved. The one thing the 200x figure does establish is that 5x is a prudential setting the firm chose, not a technical ceiling of its infrastructure.

Propr: we are re-checking the rulebook version references in our own July and August entries

Nothing changes for a live account. What changes is what we are willing to assert: until this check is finished, treat any version number attached to a Propr rule on this site as unverified, and read the published rulebook on Propr's own site. Screenshot the rules the day you pay — a dated capture is the only version history you control.

HyperPNL bought, funded and traded — desk-research cap lifted, 61.8 → 64.5

We paid $42 for a 1-Step Flex $5,000 evaluation, funded it in USDC on Arbitrum and traded it the same day. `tested` goes to true and the 85-point cap comes off. Four criteria move, and the firm ends up higher than it was, because most of what the money established is in its favour. (1) Rules 64 → 68. Two of our own criticisms fell. The 1-Step requires NO minimum profitable days: the counter on the paid account reads 0/0, so the checkout was right to show no such line and our disclosure complaint of that morning was wrong. The app also states its limits as figures a buyer can check rather than percentages to interpret — a $4,850.00 daily floor and a $25 profitable-day trigger on a $5,000 account — and it publishes a real breach post-mortem: our $25,000 demo was closed automatically at 12:18 on 9 September with the cause named in the interface, Daily Loss Limit, at −3.3%. Against that, a new disclosure finding costs it points: the docs name three contracts — Payment Router, Payout Treasury, Data Recorder — publish an address for none of them, and the purchase we traced end to end touched none of them either. The single on-chain event was a deposit into RelayDepository, a third-party bridging contract identified by the string in its own bytecode; the $42 fee was then debited off-chain in HyperPNL's internal ledger. The one field that looks like an on-chain record, labelled Tier, is not an Arbitrum transaction or block. Documenting contracts you never expose is a disclosure failure, and that is what this criterion measures. The rulebook still covers only the 2-Step, and the homepage still sells "No restrictions" over an 18-jurisdiction exclusion list. (2) Payouts UNCHANGED at 60 — and this is the part we got wrong first. The initial version of this revision cut payouts to 45 on the strength of the purchase trace. Raphael challenged it the same day: no payout was requested, so no payout evidence exists, so the payouts criterion has nothing new to price. He is right, and the objection is the same one this grid has enforced since 12 August — dated verifiable facts move scores, inferences do not. What we traced is the PURCHASE rail, not the PAYOUT rail; that the fee was debited off-chain does not establish how a withdrawal would settle, and we said as much in our own audit note before scoring as though we had said the opposite. The finding was real and it stays published in full — it simply belongs under rules, where disclosure lives, and it has been moved there. Payouts will move when we request one. (3) Survival 58 → 62: the product is real, delivered and enforcing, which a landing page cannot prove. (4) Freedom 70 → 74: 5x asset leverage carrying $24,594 of notional on a $5,000 account, at least 92 instruments on a live Hyperliquid book, no consistency rule, no time limit and no profitable-day requirement — against an order ticket that will not accept a stop-loss while a position is open. (5) Value 58 → 60: execution cost measured at about 0.055% of $60,720 of volume, in line with the advertised 0.045% taker plus spread, so no hidden markup — though a winning trade of +$6.94 still left the account $26.65 down for the day. Still untested and still the whole question: the 80% split and the "instant" payout claim. The account is in Phase 1 and negative.

HyperPNL revised on four criteria after a session inside the product (62.3 → 61.8)

We opened a free demo account on 9 September 2026 and read the checkout, the Performance Guard and the Terms of Use from the inside. Four changes, in both directions. (1) Rules 60 → 64: the homepage-versus-docs contradiction we published on 5 September was our own error — HyperPNL sells two evaluations, 1-Step and 2-Step, and each set of numbers belongs to one of them. Credit also for a named operator (Hyperpnl Ltd., Cayman Islands, reg. CO-425378) and for the docs' explicit "if a rule is not written here, it is not enforced". Held back because the rulebook documents only the 2-Step, and because the 1-Step checkout hides a minimum-profitable-day rule the app then enforces at $125 on a $25K account. (2) Payouts 70 → 60: the Terms of Use subject payout eligibility to "compliance, anti-abuse checks" and guarantee no timing or availability, while the homepage sells "Even we can't reject your payout"; the funded programme is a "Trade Ideas" arrangement in which the trader owns nothing and the firm keeps execution rebates it calls a possible conflict of interest. (3) Freedom 65 → 70: 92 instruments verified live on a real Hyperliquid order book, with Limit / Market / Stop Loss / TP / Reduce-Only and no consistency rule. (4) Value 60 → 58: $215 for a $25K 1-Step is roughly 1.7× Propr's Turbo at the same size, and deposits are USDC-on-Arbitrum only with an instruction to send $1 more than you intend to deposit. Net movement half a point, because the transparency gained and the payout guarantee lost very nearly cancel. Still capped at 85: we have never bought a HyperPNL evaluation.

HyperPNL: a 2-Step evaluation sold beside the 1-Step, with its own rule grid — first observed 9 September 2026

Which rules govern your account depends on the tab you clicked at checkout, and only one of the two products has a written rulebook. If you buy the 1-Step — the cheaper ticket at every size except $25,000 — the checkout screen is the whole of your documentation. Screenshot that screen the day you pay. No profitable day is required on the 1-Step: the app tracks the metric, it does not impose it.

HyperPNL: our own $42 purchase left no on-chain trace of the firm — observed 9 September 2026

Buying a HyperPNL evaluation puts your fee into a third-party bridge and then into an internal ledger. Nothing about the purchase, the split or the reserve can be verified by a third party, which is the opposite of what the homepage advertises. We score that where it belongs: as a disclosure failure under the rules criterion, and NOT against payouts. The purchase rail is not the payout rail — that a $42 fee settled off-chain establishes nothing about how a withdrawal would be paid — so the payouts sub-score did not move at all on the day the desk-research cap was lifted, and it stays at 60 until we request a withdrawal and have something to measure. An earlier version of that day's revision did cut it to 45 on this trace alone; the objection was raised and accepted the same day, and the sub-score was restored. The honest comparison, since we have been asked for one: at Propr and at Hypernova we have been paid and have reconciled the money on-chain — Hypernova's 80% split to the cent on Arbitrum, and Propr's Payout Distributor a verified contract on Ethereum with published source. At HyperPNL we have not requested a withdrawal at all, and no address is published for any of the three contracts. No payout of ours has been tested — the account is in Phase 1 and negative — so the 80% split and the "instant" claim remain untested rather than disproved.

Hypernova reserve read: $642,953.63

Direct chain read (block 502546138). Reserve wallet net −106,139.997356 USDC in seven days — 106,140.00 to within 0.3 cent, close to round but not to the cent like the 128,000.00 of 31 Aug; vault up $4,523.80. Total down $101,616.20 vs 31 Aug while the homepage lifetime-payouts counter rose $94.5K over the same window — ~$7.1K untraced. Homepage displayed $649.9K (+$6.9K vs chain); the stats page still showed $668,297 with 5 Sep as its latest payout, so it looked stale.

Propr: the public rulebook shows v1.0 and a changelog ending 18 March 2026, while we read v1.0.4 and v1.0.5 in July

You cannot check today which version governs the account you are about to buy, and the two versions we read in July are not retrievable from the public page. Screenshot the rules the day you pay: a dated capture is the only version history you control. We keep our July reads dated rather than deleting them — they were read at source.

Hypernova: the $25K Low Risk assessment is $275 on the homepage configurator and $280 in the rulebook fee schedule

Five dollars is trivial; two prices for one product is not, because the fee schedule is the document that would settle a dispute. Screenshot the checkout total you actually pay, and treat the rulebook figure as the one the firm can point to later.

Breakout: the split we described as an 80 to 95% ladder is published today as 80-90%, with the 90/10 sold as a checkout upgrade

The correction is ours, not the firm's: what is published today is 80-90%, and the 90% costs extra at purchase rather than being earned over three profitable months. If you were comparing firms on a 95% headline, that number does not exist on Breakout's site today.

Try Propr −30 %: personal code after proof → public code OWNTHEHOUSE

Any switcher can take −30 % on a first challenge immediately; our 5 % cashback applies to the amount actually paid after the code. Pages describing the proof-and-email mechanic must be updated.

Hypernova reserve read: $668,336.00

Split not read; stats page total ($668,336 USDC, cold reserve + operational vault), displayed "$668.3K" — not a direct chain read. Down $76,233.83 in five days, against $715,417 of lifetime payouts published the same day; outflow not traced, so the two figures are reported side by side. Homepage counters were unreliable at this read: $671.3K and $536.4K for the same reserve on two loads minutes apart.

Full registry re-verification — all 10 firms re-checked, no status change

Full DNS + HTTP + manual page re-check of the 10 tracked crypto prop firms, closing the gap since the 10 August verification (the daily automated monitoring ran throughout; this entry publishes it). No status change: …

Hypernova payouts 92 → 96 (total 85.7)

Nine payouts taken on our own funded account 28–31 August 2026, the largest reconciled on-chain at exactly 80% to the cent. Remaining gap to the top sits in the rules criterion.

Hypernova → us: $1,932.09 — Six withdrawals on our $25K funded account F2, each reconciled on Arbitrum

The six 'Process Withdrawal (F2)' entries of the account's on-chain log (29 Aug → 1 Sep 2026), read 7 Sep 2026 and checked on Arbitrum's public RPC: status OK and a USDC transfer (0xaf88…5831) to our wallet of 101.48,…

Hypernova: 141 perpetual markets become 99 in six days, with no announcement

If your strategy lives on a thin or exotic pair, check the markets page before buying — a listed instrument on 25 August was not guaranteed on 31 August, and the firm did not say which went.

Hypernova reserve read at $744,569.83 — down $116,947.51 in six days, including a round $128,000 outflow

On-chain reading of 31 August 2026, 03:18 UTC: Hypernova's payout reserve stands at $744,569.83 on Arbitrum (vault $27,174.79 + reserve wallet $717,395.03), down $116,947.51 from the 25 August top-up level. The reserv…

Hypernova → us: $3,494.69 — Nine payouts in four days on our own $25K funded account, all Paid with public hashes

Nine payout requests between 28 and 31 Aug 2026 ($102.54 to $699.43 each), every one marked Paid on the dashboard with an Arbiscan transaction. Spot-checked on-chain: the $699.43 request of 31 Aug settled 559.546701 U…

Hypernova reserve read: $744,569.83

Reserve wallet out exactly 128,000.00 USDC — reads like a treasury movement, not payout flow; destination not traced. Homepage displayed $765.5K: a $20.9K display lag, the widest yet.

Propr payouts 84 → 88 (total 86.7)

Propr's transparency dashboard now publishes a verified Payout Distributor proxy contract on Ethereum (0x6e81…7acf24, 3,864 payout txs) — the fact behind the 12 August markdown ('payer is an ordinary wallet') expired, so the markdown moved. Not 92: still no pre-funded reserve to read.

Hypernova rulebook v1.0 → v1.1: the third-party bot ban is gone, and so is the hidden consistency requirement

Automation on your own account is now inside the rules — the beta's §14.2 no longer applies. Copy trading is still a closing offence: automate your own strategy, never someone else's signals.

Hypernova reserve topped up to $861,517.34 — up ~$117K on the 17 August reading

On-chain reading of 25 August 2026: Hypernova's payout reserve stands at $861,517.34 on Arbitrum (Vault + reserve wallet), up from $744,562.56 on 17 August — a top-up of roughly $117K. Status unchanged: active. Record…

Hypernova reserve read: $861,517.34

Down $139.5K in a week from the $1.0M level; homepage display ($861.7K) matches the chain to ~$0.2K.

Propr → us: $153.91 — Our own payout, reconciled on-chain to the cent

$192.39 requested, 153.908936 USDC received on 22 Aug 2026 — exactly 80% ($192.39 × 0.80 = $153.91). Sent by Propr's Trader Payout Sender through its verified Payout Distributor contract on Ethereum, tx 0xea78acbf…ca1…

Hypernova reserve read: $1,001,060.62

Top-up: the reserve wallet took in exactly 260,395.00 USDC. The firm announced 'topped up to $1,000,000' — accurate, slightly understated.

Hypernova reserve read: $744,562.56

Reserve wallet unchanged to the cent in two days; all movement in the vault (payouts drain the vault).

Carrot Funding: advertised account ceiling drops from $500K to $100K

The headline number a firm advertises is not a commitment. A quiet 5× reduction of the ceiling is exactly the kind of change this register exists to date.

Correction: Carrot Funding reclassified active — our 10 Aug 'dead' verdict checked the wrong domain

On 10 August we classified Carrot Funding dead because carrotfunding.com 302-redirects to a HugeDomains parking page. That was the wrong domain: the firm brands itself carrotfunding.io and is fully operational there —…

Hypernova re-checked: public launch announced 14 August, reserve read at $746,229.76

Full re-check of Hypernova. On 14 August 2026 its founder announced the public launch: the invite-only beta is over and the checkout is open on app.hypernova.xyz, with pricing published ($120/$275/$365 on the 25K acco…

Hypernova reserve read: $746,229.76

First direct chain read. Homepage counter (~$750.9K, rounded) no longer matches the chain to the cent.

Grid revision after Hypernova's founder challenged it line by line — four changes

(1) Hypernova's desk-research cap lifted: we had bought and traded an account. (2) Payouts: two Hypernova receipts reconciled against the Vault at exactly 80% (28 July) scored above Propr's then-unverifiable ordinary-wallet payer. (3) Propr's rules score marked down for a 20% price rise shipped without announcement (2-Step 100K, $749 → $899). (4) Survival now rewards verifiable on-chain solvency. Net: the two firms one point apart, for opposite reasons.

Propr: the purchase grid moved ahead of the rulebook — Pro 1-Step, $200K tier and Free Trial not described in v1.0.5

You can buy a product whose rules are not in the published rulebook. Before paying for Pro or the $200K tier, get the format's exact target/daily/drawdown from the checkout screen and keep a dated screenshot — the rulebook will not settle a dispute it does not mention.

Solana Funded: advertised ceiling halves from $250K to $100K

Same lesson as Carrot: size the account you can actually buy today, not the one in last month's screenshot.

HyperPNL: the $200K tier disappears from the pricing page; $50K and $100K still 'Soon'

The only account you can buy is the $25K. Treat every 'Soon' tier as non-existent for planning.

Full registry re-check: GT Funded reclassified relaunching (unproven), FundedPoly still 402

Second dated pass over all tracked firms. GT Funded: the original gtfunded.com has had no DNS record since June 2026; a new site on gtfunded.xyz is a waitlist landing page whose own figures contradict each other ($250…

Hypernova reserve read: $770,556.00

Figure published on the firm's homepage; taken as the starting point, not a chain read.

Propr: floating P&L no longer counts toward daily loss and max drawdown

An intraday wick against you no longer ends the account; only where you close (and the start-of-day balance) matters. Materially looser for scalpers and anyone holding through news. Until the rulebook is re-versioned, treat it as observed behaviour and re-check the rules page before relying on it.

Propr → us: $119.79 — Our first payout, requested and received

119.79 USDC, settled in 2 minutes on 31 Jul 2026 — the first Propr payout that touched our own wallet.

Hypernova: $7,663.03 — Largest single payout we verified

Announced gross $9,578.79 × 80% split = $7,663.03 — exactly the USDC transfer that left the Vault on 28 Jul 13:19 UTC. Matched to the cent, twice (second: 8,812.44 → 7,049.95).

Propr rulebook v1.0.5: the daily-loss floor added six days earlier is removed

A losing streak compounds: each day's allowance is computed on a smaller balance. Plan the daily limit in dollars from the current balance, not from the account's nominal size.

Registry opened: 10 crypto prop firms under dated DNS/HTTP monitoring

First dated checks recorded for the crypto prop firm status registry: every tracked firm — living ones included — gets DNS resolution, HTTP status and a manual read of the live page, each check dated. GT Funded entere…