Validation Platform · From $49/mo
A backtest checked against live fills, a live pilot on our capital, and a record built for an allocator's bar. No pedigree required.
Keep half the net profit your strategy makes in the pilot, and 30% at allocation. You keep your IP.
For independent quants and small teams whose strategy trades often enough that fills, latency and fees decide the result: HFT, market making, intraday, grid, anything that rests limit orders.
The problem
Every idea you test is scored by your backtest: every parameter sweep, every indicator, every model you tune. Weeks of compute go into a search whose objective function nobody has checked. If the simulator pays for fills you would never get, the search finds the strategy that collects them.
The only test of a backtest is parity with your own live execution: same venue, same period, fills and fill times inside a tolerance. That needs live fills, and nobody goes live before the backtest looks good. So the instrument is never questioned; the strategy goes live once, loses, and you blame the strategy or the model.
A backtest is only as good as its parity with the stack it is calibrated to. The one a large firm trusts is calibrated to that firm’s fees, latency and queue position. You cannot replicate their stack, so their assumptions do not transfer. Yours need parity with the execution you will actually get.
What sets us apart
Our simulator's fills are compared with our own live fills on the same venues and re-fit as the venue changes. It is a model; the difference is that it is checked.
Eligible pilots run on our institutional tier: low single-digit basis points round trip, against the 20–40 a solo account pays.
Tick trades and full order-book depth on our live venues. A candle-only backtest cannot resolve queue position or intrabar execution, so it cannot test a resting-order strategy.
Pilot records are read against an allocator's written bar once allocation opens; the quant and the allocator never have to find each other.
How it works
Your strategy on tick data, out-of-sample, with fills from a model checked against our own live executions on the venues we cover. Plus a capacity read.
The same strategy on market data it has never seen, with the same fills. Agreement with the backtest is not a live-fill test on its own; it is the second check.
Real orders on our capital at minimum size once the four-check gate passes. Your own executions, checked by us against exchange records. Pro tier.
Earned by pilot results, with a 30% share of net profit at allocation. Never purchasable.
Backtest replays history. Paper runs forward in real time on data nobody has seen. Both use the same calibrated fills. The pilot is the first time the fills are real.
$0 setup · cancel anytime · 14-day money-back.
How we compare
| Candle backtest tools own capital, candle fills | Custom-fill-model engines own capital, self-calibrated | Demo-based funding programmes simulated fills, then allocation | Martian Crypto-native | |
|---|---|---|---|---|
| How it works | Tools on your own capital | Engine on your own capital | Subscription → demo record → funding | Subscription → validation → live pilot → allocation |
| Fill realism | Typically fill at touch or trade-through | Fill model you calibrate yourself, if you have live fills | Demo fills | Fill model checked against our live fills on the same venues |
| Market data | Typically candles | Whatever you connect | Broker feed | Tick trades and full order-book depth on our live venues |
| Fees during live validation | Your retail tier | Your retail tier | None, demo | Our institutional tier (low single-digit bps round trip) |
| Capital to trader | None | None | Simulated account after a paid evaluation; payouts typically funded by evaluation fees | Our capital, after the gate |
| Your share of profit | n/a | n/a | Payout % on simulated profit | 50% in the pilot, 30% at allocation, on real P&L |
| Track record | Self-reported | Self-reported | Verified from demo | Recorded from exchange executions, checked by us |
Your edge stays yours
Your strategy and IP stay yours. Always.
Runner isolation and access restrictions, under a mutual NDA. We retain orders, fills, P&L and agreed risk information.
A validated strategy is yours to take anywhere.
Standard on both sides before anything connects.
Pricing
Everything in Base, plus:
$0 setup fee · cancel anytime · 14-day money-back. Graduation credit: pass the gate and every payment not already credited becomes credit against Pro fees, not cash, valid 12 months from the gate pass, each payment counting once (a fee credit, never a deposit — we do not hold or trade your money). Rev-share on live-funded P&L applies only to strategies we fund with live capital (upside-only, negotiable). A meaningful allocation is earned by merit — never purchasable.
What happens next
FAQ
Three separate pieces of evidence, and we keep them separate. First, the simulator is calibrated against our own live fills on the same venues, held out from the data it was fit on. Second, forward paper tests your strategy on data it has never seen, with those calibrated fills. Third, the live pilot is your own ground truth: real fills on our capital. Backtest agreeing with paper is not fill evidence on its own; both use the simulator. The pilot is. On public data a fill-at-touch backtest grants about three fills for every two a queue-aware model grants. The calibrated tolerance for your venue is handed over at onboarding, and parity is measured for your strategy in the pilot.
Base ($49) is tick-data backtest with fill-calibrated forward paper for up to three strategies, plus review against the gate. Pro ($149) raises that to ten and adds a live pilot on our capital (up to three strategies at once, each review-gated), a live record, and HF-grade data.
Four checks, the same for every strategy: out-of-sample performance holds; the forward-paper period agrees with the backtest within the applicable tolerance; capacity clears the pilot's minimum size; risk stays inside our limits (drawdown, exposure, order rate). What each check measures is published here. The numbers depend on venue and instrument, so you receive the ones that apply to your strategy at onboarding, and they stay fixed for that evaluation; if our model, the market or our risk policy changes materially, we tell you and agree a reassessment rather than move the bar. Pass all four and you move to the live pilot.
When any strategy on your account passes the gate, every subscription payment not already credited becomes credit against Pro fees. It is credit, not cash, valid for 12 months from the gate pass, and each payment counts once; refunds and charges paid with credit do not count. We never hold or trade your money. Base subscribers upgrade to Pro when a strategy passes and the credit applies there.
Your strategy runs on our infrastructure under access restrictions, runner isolation and a mutual NDA. We retain orders, fills, P&L and agreed risk information. Your strategy IP stays yours. These are technical and contractual protections, not a cryptographic guarantee against access by the infrastructure operator. The delivery and retention details are in the onboarding documentation.
You do, fully. There is no exclusivity — a validated strategy is yours to take anywhere. A mutual NDA is standard.
Promising strategies run a live pilot on our own capital (Pro, review-gated). You keep 50% of net profit in the pilot, and 30% at allocation. Net means after exchange fees and funding, on realised P&L, cumulative over the pilot; losses carry forward inside the pilot and there is no clawback. Full terms at onboarding. Allocation beyond the pilot is earned by those live results, merit-gated, never bought.
We do. It is recorded from exchange executions on our account during the pilot and checked by us against the exchange's own records. It is not attested by a third party, and we say so wherever the record is used.
The stack is not the edge, and the edge is not for sale. Our edge is the strategies and the calibration data behind the fill model; we sell checks against that data, never the curve, the counts or the parameters. The stack's cost is fixed and our own book's capacity is bounded by our strategies, not by the stack, so every graduate strategy that runs on it adds capacity we could not create alone. Every graduate is also a candidate with a reviewed record — a hiring funnel no job board can match.
The bar is the gate, and it is passed before the pilot: out-of-sample performance holds, forward paper agrees with the backtest within tolerance, capacity clears minimum size, risk stays inside limits. During the pilot there is no second performance target. The risk limits agreed at onboarding are hard stops — a breach pauses the pilot, not the subscription — and parity is measured: whether live fills and P&L per fill match the forward-paper period within the tolerance you were given. Allocation beyond minimum size is earned by pilot results through the 30% allocation share, never bought.
It adapts on a schedule, not on its own. The model is conditioned on venue, instrument, time of day and a volatility bucket, and it is re-fit against new live fills as the venue changes. A material change is disclosed to customers mid-evaluation and starts an agreed reassessment rather than silently moving the bar. Fee changes do not enter the fill model — they enter the fee tier applied to your run, which updates when the venue's does.
You get the honest read — realistic fills, out-of-sample results, and a capacity number. Most strategies do not clear; that is the point. You keep access to iterate and resubmit.
Yes — monthly, anytime. And there is a 14-day money-back guarantee.