An independent, adversarial red-team of a trading strategy, backtest, or investment thesis. You get a signed report you can verify yourself — not an opinion you have to trust.
Paper research. Not investment advice. Not a signal service. Not a fund.
A backtest is the easiest thing in the world to fool yourself with. A great-looking equity curve, a t-stat north of 3, a confidence interval that never touches zero — and every one of those can be an artifact of a data gap, a look-ahead you didn't notice, a threshold you tuned without counting, or plain market beta wearing a costume. The usual checks (year splits, bootstraps, randomized twins) all run on the data that already survived, so they're structurally blind to the bias that deleted the evidence in the first place. The only real defense is an adversary — someone whose job is to find the knife, not to sell you the strategy.
We take your strategy, backtest, or thesis and try, genuinely, to break it — through the same gauntlet we built over months to kill our own edges, including our own former flagship. We report exactly what we find, whether the answer is "this is noise," "this is beta you're paying a fee to rent," or "this is real, but not what you were sold." Then we hand you a signed, tamper-evident report, issued under our formal name, Null Result Research — its integrity checkable with about 40 lines of standard-library Python, so a "this is clean" from us is something you can verify, not something you take on faith.
We are not an investment adviser. We do not tell you what to buy or sell, we do not manage money, and we do not sell signals. We stress-test the method. A passing report means "we could not break it with our tools" — never a promise of performance.