DualQuant is an independent check for trading strategies. Describe yours in plain English and get a verdict in under a minute: could every trade really have happened, and did it pay for the risk it took?
It made money, but not enough to justify the risk it took.
The best-known strategy in the world, rejected. A check is only worth something if it can say no.
Private preview, as of 6 November 2026. We built every check on our own crypto, gold, index and FX strategies first, and every headline number moved down. We are now taking the first submissions from outside.
Tell us what you are testing. Accounts are issued by hand during the private preview, and we reply with an API key.
Email us ->Write it the way you would explain it to a colleague, or send structured rules. Use the browser page at api.dualquant.ai/app, or give the key to an AI agent: the API explains itself.
The agent API contract ->We restate your strategy as rules and you confirm them. If anything is ambiguous, we ask rather than guess. Then the plan is frozen, the checks run, and the report comes back in plain words.
Neither mode tells you what to buy, hold or sell. Each one shows you what is there.
Is the record real, and was it worth the risk?
Five checks run in order and stop at the first failure. Every trade is rebuilt from raw bars by an auditor that shares no code with the simulator. You get one of four verdicts, the reason in plain words, and what trading costs would take from the return.
Failing a Proof check means the result is invalid. Context checks tell you whether it was worth having.
It says whether the record is real and worth its risk. It does not say what the strategy will do next.
What do this company's own filings say?
Ask for a ticker and see the company the way its own SEC filings describe it, ten years at a glance.
Each figure is dated by its filing, which is when the public could first have seen it. There are no prices on the view: it links to a public chart instead.
No score and no valuation model: no discounted cash flow, no cost of capital, no fair price. Stocks only, since a fund has no margins or insiders to read.
Each of these is a property of how it is built, not a promise.
One of four verdicts, the reason named, the comparison spelled out, every term defined on the page, and a list of what the verdict does not claim.
A pre-flight tells you whether your rules are supported and the data is ready before anything runs, and a refusal there costs nothing. Then the full check, start to finish, in one sitting.
The API explains itself, so an agent can read the contract, submit rules and read the result as JSON without being taught. People without one use the same API from a browser page.
The report names what is weak: concentrated profit, exits that decided nothing, costs, long stretches underwater. The changes are yours. The more you try on the same data, the higher the bar, so a pass still means something.
Every verdict rests on the same fingerprinted market data, so any two are comparable. Your plan is locked before the first bar, so the rules cannot drift toward the result.
One attempt is one complete verification of a frozen plan, or one company view, over the API.
Tell us what you are building or testing and we will issue an API key.
Most strategies do not clear this bar, and ours did not either. That is what you are paying for: a check rather than a compliment.
Attempts are a balance, not a discount. Paid for or not, every idea you test counts toward the bar.