Best On-Chain Prop Firms 2026

What is a decentralized prop firm?

In one sentence: a decentralized prop firm is a trading firm whose funding, rules and USDC payouts settle on a public blockchain instead of a private back office. As of 1 October 2026 our registry tracks 10 crypto prop firms, scores 7 of them on a grid revised 10 September 2026, and records 1 already defunct (GT Funded, offline since June 2026).

A decentralized prop firm (or on-chain prop firm) is a proprietary trading firm that funds you with its capital and settles your trades and payouts on a blockchain rather than through an internal back office. You pay a one-time evaluation fee, hit a profit target without breaching the loss limits, and get a funded account — but the rules, profit split and USDC payouts are verifiable on-chain.

On-chain vs traditional prop firms

Traditional prop firms (FTMO-style) run everything internally: you trust their dashboard for your P&L and their back office for payouts. On-chain firms move settlement and payouts to the blockchain, so risk parameters, profit splits and payout triggers are transparent and harder to manipulate. The trade-off: these firms are new and unregulated, so track record still matters. I unpack the full case for each model — custody, payouts, automation and fees — in on-chain vs traditional prop firms.

Why Hyperliquid became the home of on-chain prop trading

Hyperliquid is a high-performance L1 built for decentralized perpetuals (100k+ TPS, sub-100ms latency). That makes on-chain funded trading actually usable — fast execution, deep liquidity, and an open API that bots can plug into. Every serious on-chain prop firm in 2026 is building on it. If Hyperliquid is specifically your chain, I ranked the three firms building there in best Hyperliquid prop firms 2026.

Best decentralized prop firms in 2026 (quick comparison)

Propr.xyz leads our grid at 86.7/100 (revised 10 September 2026), one point ahead of Hypernova at 85.7/100 — the only two firms in this table that have ever paid us. Propr starts at $25 with a flat 80% split; Hypernova also starts at $25 (5K account, read in-app 17 August 2026), its 25K running $120 to $365 depending on risk profile.

The on-chain prop firms with a published rulebook today — plus the crypto-native firms searchers compare them against:

Propr.xyz — best on-chain prop firm overall

Propr.xyz scores 86.7/100 on our grid revised 10 September 2026 — the highest of the 7 firms scored. Challenges start at $25 (25K at $275), the split is a flat 80% up to $300,000 aggregate, and it voluntarily publishes a 15.2% pass rate on paid challenges (2,366 of 15,583, read 26 August 2026) most firms would hide.

Propr.xyz is the most product-complete and best-funded on-chain prop firm in 2026 ($1.5M seed, backed by SwissBorg, TGE planned). It’s our recommended pick for three reasons: the widest tradable universe (160+ markets, read 7 September 2026 — crypto, equity and commodity perps, plus prediction markets and memecoins), transparent on-chain payouts (USDC; median processing time 12 minutes in August 2026 after the anti-fraud review, self-reported by the firm), and — crucially — an open API that lets you automate your challenge.

• Profit split: 80% to you, on-chain USDC payouts — the full payout mechanics.
• Account sizes: $5K to $200K (aggregate cap $300K), four formats — the verified fee grid and ROI math.
• Rules: ~10% target, ~3% daily loss, ~6% max drawdown, no time limit.
• Edge: agent-friendly API → automatable execution (you set the direction, a bot handles DCA/TP/SL).

The challengers: Hypernova, Carrot Funding, Solana Funded, Breakout, HyperPNL — and GT Funded, gone dark

Hypernova is the closest challenger at 85.7/100: public since 14 August 2026, smart-contract payouts of 6 seconds on average (the firm's own homepage figure, read 21 September 2026), published pass rates of 19.2–27.9% (read 16 August 2026), and a payout reserve of $497,973.33 we read ourselves on Arbitrum on 1 October 2026. GT Funded, once the loudest advertiser in the category, has been offline since June 2026.

Hypernova raised $3M pre-seed (Lemniscap-led) and leans into the “trustless” angle: rules and payouts enforced by smart contracts, an 80% split (per its v1.0 rulebook — not the 90% some posts claim), and programmatic on-chain USDC payouts of 6 seconds on average (the firm's own homepage figure, read 21 September 2026). The strongest instant-payout narrative — and since 14 August 2026 it is public, with checkout open and pricing published ($120/$275/$365 on the 25K account); its public rules, read 16 August 2026, listed no bot restrictions, and its rulebook v1.1, read 5 September 2026, permits automated strategies built and run by the trader on a single account while banning copy trading and external signals. I read the whole thing in my Propr vs Hypernova comparison (funding source).

GT Funded advertised the highest headline split in the category (up to 92%) — until it vanished: all its domains went offline in June 2026, with no statement. A live reminder that headline numbers mean nothing without a track record. Solana Funded is the biggest Solana-native option by market count (up to 90%, $100K — its site, read 5 September 2026). Both are worth a look if you want maximum split and aren’t tied to Hyperliquid. Carrot Funding is the “verifiable” option, and we stopped reading it and bought it: a 2-Phase $5,000 evaluation on 10 September 2026, $65 list, $52 paid with the code royaref, in USDC on Arbitrum. Read from the inside that day: execution on Hyperliquid, account on HyperEVM, payments on Arbitrum — three lines our earlier sheet had wrong. 80% split, accounts to $100K ($200K aggregated funded cap), payouts announced under 24h with a 100 USDC minimum and the full amount only. Its rulebook bans bots and copy trading (§23: “automated software, high-frequency trading systems, AI-driven tools, or mass order entry systems”) and caps leverage at 5× across every asset class (§14) — a dated prudential setting rather than a structural ceiling, since the firm told us on 10 September 2026 it has run up to 200× on gTrade and intends to raise leverage on Hyperliquid, announced and not shipped. On reserves, be precise, and the precision here is a correction of ours: this page briefly said that the vault promised by §19 could not be found. It can. Carrot’s GitBook publishes a full address table with named roles — Vault Custody Contract, Vault Integrations, a 2-of-4 payout multisig and the payout executor EOA — which we had not read, and having read it we checked it on 10 September 2026: the Vault Custody Contract on HyperEVM is a Safe holding about $15,016 in the Felix/Morpho USDC vault plus $98 on HyperCore, with funds stated to be mid-migration from Arbitrum to HyperEVM, so the on-chain picture is partial. The four published payout hashes are real Arbitrum transfers we checked one by one. The account itself is on-chain too: challenge #3955 is token #550 of the Funded Bunny Initiative collection (symbol FBI) on Arbitrum, contract 0x16f6ff310d113fe4b5a1c2e251e25530989e4c9c, minted at 05:00:25 UTC on 10 September 2026 — the hour we paid — and never moved since. The token ID is not the challenge ID, and no page of theirs documents that. I put it head-to-head with Propr in my Propr vs Carrot Funding comparison. If you’re weighing an on-chain firm against a traditional CFD one, the sharpest version of that trade-off is Propr vs FundingPips — cheaper entry and a refunded fee on one side, a flat 80% split, $20 on-chain payouts and the written right to run a bot on the other. And if the firm you’re comparing calls itself crypto-native but runs on MT5, read Propr vs Crypto Fund Trader first — 700+ pairs and 1:100 leverage against on-chain fills, plus the FINMA warning list entry most review pages leave out.

HyperPNL — two evaluations, and the corrections we owe you

HyperPNL sells two evaluations, not one. We opened a free demo account and read both checkout tabs from the inside on 9 September 2026 (app.hyperpnl.com — Start Challenge (login required), app v1.4.131). The 1-Step Flex costs $42 at 5K, $86 at 10K and $215 at 25K: 10% target, 3% max daily drawdown, 5% static max overall drawdown, and no minimum profitable days shown anywhere at checkout. The 2-Step Flex costs $50, $90 and $213 for the same sizes: 10% then 5% target, 5% max daily, 9% static max overall, 2 then 3 minimum profitable days — which makes the 2-Step 25K ($213) cheaper than the 1-Step 25K ($215). There is no time limit on any phase. 50K and 100K are still marked “Soon” with the slider locked — our eighth consecutive check — so the real ceiling you can buy is $25,000. That silence on profitable days turns out to be accurate, and we can now say so from the inside.

We stopped reading and bought one. On 9 September 2026 we paid $42 for a 1-Step Flex $5,000 evaluation, funded it in USDC on Arbitrum and traded it the same day; the account is in Phase 1. On that paid account the Performance Guard writes the daily floor as a balance rather than a percentage — $4,850.00, which is 3% of $5,000, a $150 budget for the day — next to a 5% static overall drawdown ($250) and a 10% target ($500). The profitable-days counter reads 0 / 0: the 1-Step requires none. The $25 figure beside it is a trigger — 0.5% of $5,000, the threshold above which a day would count as profitable — not a day you have to produce; the 2 then 3 profitable days belong to the 2-Step alone. The leverage is where the real risk sits: the app let us carry $24,594 of BTC/USD notional on $5,000 of capital, an asset leverage of 5× — 4.9 times the account. At that size the $150 daily budget is 0.61% of BTC movement and the $250 total budget is 1.02%. Friction is measurable too: across $60,720 of volume the balance went from $5,000.00 to $4,973.35, −$26.65, while the single closed trade booked +$6.94 gross — roughly $33.59 of execution cost, about 0.055% of volume, consistent with the 0.045% taker fee advertised.

Three corrections, made here in the text rather than in a footnote. The newest is the one just above, and it is ours: from an earlier reading of a free demo account the same day we saw a “Min Profitable Day Trigger” line and concluded the 1-Step enforced a profitable-days rule it never disclosed at checkout. It does not. The trigger is a counting threshold; the requirement, read on the paid account, is zero out of zero. We withdraw that reproach — the second time in two days that looking closer made us take a criticism back. The other two concern what we published on 5 September 2026. First, we described a “contradiction between the homepage and the docs” on drawdowns and phase count — 5% versus 9%, 3% versus 5%, one phase versus two. That was our error: these are two different products. The GitBook rulebook documents only the 2-Step and never the 1-Step, which does leave whoever buys the 1-Step without written rules, but it is not a contradiction. Second, we wrote that HyperPNL publishes no entity, no company and no jurisdiction. That was false. Its Terms of Use — last updated 29 January 2026, readable only once logged in, on a noindex page rendered client-side that no fetch or crawler can read — name Hyperpnl Ltd., registration number CO-425378, registered office CO Services Cayman Limited, P.O. Box 10008, Pavilion East, Cricket Square, George Town, Grand Cayman KY1-1001, Cayman Islands, contact business@hyperpnl.com. §12 puts the agreement under Cayman Islands law, with binding arbitration seated in the Cayman Islands, a single arbitrator, in English.

The contradictions that remain are the firm’s own. Its homepage says “No restrictions — Anyone can participate freely”, while §3.1 of those same Terms restricts 18 jurisdictions: the United States of America and its territories, Canada, Iran, North Korea, Syria, Cuba, Russia, Belarus, Ukraine (Crimea, Donetsk, Luhansk and other occupied or sanctioned regions), Venezuela, Afghanistan, Iraq, Yemen, Libya, Somalia, Sudan, South Sudan and Myanmar (Burma). The list can change “at any time, without prior notice”, and using a VPN to get around it is “a material breach” that can trigger “immediate suspension or termination of your account and permanent ineligibility for payouts”. Same shape on payouts: the homepage says “Even we can’t reject your payout”, while §6 makes them “executed directly by smart contract” yet subject to “compliance, anti-abuse checks, and rule adherence”, and states that HYPERPNL “does not guarantee payout timing or availability”. §2 calls the evaluation accounts “hypothetical and have no monetary value”; §4.1 makes the fee “non-refundable once trading activity begins”. The split is 80%, the minimum withdrawal 1% of the account, paid in USDC. No documentary KYC at this date — its own FAQ, read 9 September 2026: “No. At this time, HyperPNL does not require standard KYC verification for payouts” — but the geographic filter is written into the contract.

What is genuinely good is also written down: the rulebook’s “No Additional Rules” section bans multiple accounts and copy trading, then states there is no consistency rule, no maximum risk per trade and no hidden or discretionary condition, with the sentence “If a rule is not written here, it is not enforced.” News trading is allowed. At least 92 distinct instruments sit in the selector, on Hyperliquid’s real order book. And the rule engine demonstrably runs: the free $25,000 demo account we had open was closed automatically on 9 September 2026 at 12:18 app time, the interface naming the cause “Daily Loss Limit”, final balance $24,170.48 against $25,000.00 — −3.3%. A failure stated, its cause named, its moment timestamped, in a post-mortem you can reopen. The reservation is that it was a demo: the rule was enforced, the financial consequence was not.

The on-chain promise is the one place where buying counted against HyperPNL — and it counts as a disclosure failure, under the rules criterion, not against payouts. The docs already concede that “Trading is handled off-chain for performance and scalability”, with only fund routing and result anchoring on-chain. We traced our own $42 end to end on Arbitrum from a public RPC, and the only on-chain event is a deposit into RelayDepository, the deposit contract of Relay, a third-party bridging protocol. The $42 was then debited off-chain, in HyperPNL’s internal ledger. For the one purchase we can follow from start to finish there is no Payment Router transaction, no allocation to a Payout Treasury, no protocol revenue split and no referral commission — and the three contracts the docs name (Payment Router, Payout Treasury, Data Recorder) still have no published address, not on the site, not in the GitBook, not in the Terms, not in the app’s JavaScript bundle. The single field that looks like an on-chain record, labelled “Tier” in the failed-challenge post-mortem, is neither an Arbitrum transaction nor an Arbitrum block. We are not saying those contracts do not exist. We are saying that for the part of the chain we control there is nothing to verify — while the homepage sells “Everyone can see everything on-chain”. The contrast is not theoretical: at Hypernova the payout reserve and the 80% split reconcile to the cent on Arbitrum, and Propr’s Payout Distributor is a verified contract on Ethereum with published source code. At HyperPNL there is no address at all.

HyperPNL scores 64.5/100 on our published grid, its own line revised 9 September 2026 — up 2.7 points from 61.8 the day before, because most of what the money established is in the firm’s favour and two of our own earlier criticisms fell. The 85/100 cap that applied while we had bought nothing no longer applies to it: we bought, funded and traded an evaluation there. The payouts criterion did not move at all, and that is deliberate: we have never requested a withdrawal there, so there is no dated payout fact to score. What stays untested is the part that matters most — no payout. The account is in Phase 1 and negative, so the 80% split and the “instant” wording have never been put to the test. For context, Propr sits at 86.7 and Hypernova at 85.7 on the grid revised 10 September 2026.

Breakout is the odd one out: a crypto prop firm owned by Kraken (acquired Sept 2025), with an 80–90% split (90/10 as a paid upgrade, read 5 Sep 2026) and up to $200K — but it settles on a centralised browser terminal, not on-chain (source). If on-chain transparency is your reason for going decentralized, Breakout doesn’t qualify — but it’s the most battle-tested option for pure crypto trading.

How we tested them

Every ranking on this site comes from a published grid revised on 10 September 2026: 7 firms scored out of 100, 10 tracked in our status registry, 1 already gone (GT Funded, June 2026). The two firms that have actually paid us — Propr (86.7) and Hypernova (85.7) — finish one point apart. HyperPNL, whose evaluation we bought and traded on 9 September 2026, sits at 64.5 on the grid revised that day, with no payout observed.

We don’t rank on marketing. We open a real evaluation account, run our strategy on it, and track every rule, payout and edge case — the full criteria are in how we test prop firms, and the short-form version of this ranking lives on the prop firms we recommend. The weighting itself is written up in how the Trust Score works, and the evidence behind the payouts criterion is the list of payouts we verified on-chain. For a deep dive on our #1 pick, read the full Propr.xyz review. (Disclosure: we’re a Propr affiliate — which is exactly why we hold our recommendations to a public methodology and cite our sources.)

How to choose a decentralized prop firm

On the grid revised 10 September 2026 the practical shortlist is two firms one point apart: Propr (86.7/100) — from $25, flat 80% split, $300,000 aggregate cap, and 48 payouts to our own account — 7,670.94 USDC, every hash reconciled on Ethereum, read 19 September 2026 — versus Hypernova (85.7/100), whose smart contract settles payouts in 6 seconds on average (the firm's own homepage figure, read 21 September 2026).

• Profit split & payout speed — higher split and faster on-chain payouts win.
• Self-custody & transparency — verifiable on-chain rules/payouts beat a black box.
• Rules realism — a sane daily-loss limit and no time pressure matter more than a big number.
• Automatable — an open API means you can run a bot and remove emotion (Propr’s big edge).
• Credibility — funding, public payout proof, and a real team.

One shortcut for comparing offers that look nothing alike: divide the fee by the maximum drawdown to get your cost per dollar of risk capital. It’s the number that exposes what an evaluation-free account really charges you, and I walk through it on a $10K account in instant funding vs a prop firm challenge.

Two rules on that list decide whether you can even attempt an evaluation around a day job: whether the firm imposes a time limit or minimum trading days, and whether its rulebook permits automation. Propr has neither deadline and allows bots, copy trading and API access — the part-time playbook built on those two facts is in passing a prop firm challenge with a full-time job.

The credibility item on that list deserves its own answer, because “is the money real?” is the question every newcomer asks and almost nobody answers straight. The short version: your evaluation ledger is simulated everywhere, the price feed and the payout rail are where firms genuinely differ, and only two of those three layers are verifiable. Full breakdown in prop firm accounts: real money or demo?

And if your real question sits one level higher — “is this whole industry even legitimate?” — I broke down the entire business model, from the fee-vs-payout arithmetic to the nine red flags of firms that never intend to pay, plus the 20-minute on-chain vetting routine, in are crypto prop firms legit?

One rule on that list quietly doubles as trading advice: the leverage tier table. Every on-chain firm caps BTC and ETH higher than everything else because it has already ranked which books survive a liquidation cascade — which is why a passing allocation looks a lot more boring than most people expect. The full reasoning, with the sizing arithmetic, is in the best crypto pairs to trade in a prop firm challenge.

Are decentralized prop firms legit?

The proof is checkable rather than promised: Propr made 48 payouts to our own account — 7,670.94 USDC, every hash reconciled on Ethereum, read 19 September 2026; Hypernova made 6 withdrawals on our own funded account — 1,932.09 USDC reconciled on Arbitrum at the firm's 80% split, read 19 September 2026; and Hypernova’s payout reserve read $497,973.33 on Arbitrum on 1 October 2026. The same category also buried GT Funded in June 2026 — verify before you pay.

On-chain settlement is a genuine trust upgrade — you can verify payouts and rules yourself. But these firms are new and unregulated, and leveraged trading carries a high risk of loss. Check each firm’s payout proof and track record, start small, and only trade what you can afford to lose.

FAQ

Affiliate disclosure: we may earn a commission if you sign up to Propr.xyz via our links, at no cost to you. This article is informational and not investment advice. Leveraged trading carries a high risk of loss — only trade what you can afford to lose.