BALLAST.

Verklig, an Urverk service

The gate we hold our own fund to, pointed at your strategy.

We built a deflation gate so our own systematic fund could not fool itself. It has killed most of the strategies we ever pre-registered, in public, including one we had already approved. Before you trade real money on a backtest, let the same gate try to kill yours.

Get audited   Try the free detection floor calculator →

What the audit answers

Send a daily returns series, optionally a baseline to measure your uplift against, and the other variants you actually tried. You get a certificate covering the four questions a serious allocator will ask anyway:

  • What Sharpe survives once your trial count is corrected for selection luck?
  • Does the edge hold up in walk-forward, out of sample testing?
  • What is the probability of backtest overfitting across many in and out of sample splits?
  • What is the smallest edge your sample length could even detect in the first place?

The verdict applies the identical pre-registered significance floor we hold our own research to. It is the same code path that has killed almost every idea on our kill log, not a softened demo. Read the honest positioning plainly: this is the gate we run on ourselves, turned outward.

Who it is for

The buyer is always the person whose capital is at risk, because only that person profits from a kill:

  • Retail quants going live with savings. A drawdown on real capital plus months of misplaced conviction costs far more than an audit.
  • Prop shops and funded-trader candidates. Desk capital allocated to noise, and repeated challenge fees, add up fast.
  • Emerging managers pre-launch or pre-pitch. An allocator will ask about overfitting; walk in with the answer already computed.
  • Allocators and diligence teams. A seed into selection luck is basis points on real AUM plus the reputation of whoever signed it.
Kill it before it kills your capital. Most strategies we audit will not survive, and that is exactly the point. A kill costs you the price of an audit. Finding out with live money costs you a drawdown. We never price the verdict: a pass and a kill cost the same, so the certificate is never for sale. The rare pass is worth something precisely because the gate fails most of what it sees.

Two worked examples, rendered by the real certificate

These are the two demo inputs committed in our public repository, run through the exact certificate a customer receives. Nothing below is hand-written; every figure is computed by the gate.

The best of 40 pure noise variants on a 90 day backtest. The 39 losing variants are supplied so the gate can see the selection luck. This is the classic overfitting trap: the winner looks great in sample and falls apart once its luck is priced in. Verdict: Likely overfit.

Open this certificate full width →

A single pre-registered strategy with a genuine, modest edge over about 6 years. One idea tested once over 1500 daily returns, so there is no selection luck to deflate away and the edge is left to stand on its own. Verdict: Survives deflation.

Open this certificate full width →

Pricing

Standard audit
$299
per strategy

A full certificate plus a one-page written interpretation of what each gate means for your strategy. Verdict in 48 hours.

First ten
$199
per strategy, launch offer

The standard audit at a launch price for the first ten customers, in exchange for permission to publish an anonymized tally of the verdict.

Deep audit
$1,950
funds and allocators

Up to five related strategies, or one strategy with its full variant history, plus a methodology call and an allocator-facing letter that states scope and limits in diligence language.

A re-audit of a revised strategy is $149, because the intake is already done. The detection floor calculator and both worked examples above are free forever.

Get audited

Pay through the link below, then send your returns to the intake address. The intake note tells you exactly what to send: a returns CSV or JSON, the baseline if you have one, every sibling variant you tried or an honest count, the return frequency, and one paragraph on how the strategy was selected.

Pay and start an audit

Payment link: [PAYMENT_LINK_PLACEHOLDER]
Intake email: [INTAKE_EMAIL_PLACEHOLDER]

The payment link and intake address are wired in at launch. Until then these are placeholders, not a live checkout.

What a verdict is, and is not. This is statistical analysis of the return series you supply. It is not investment advice and not a recommendation to buy or sell anything. A passing verdict means we ruled out, at a 95% confidence floor, that your edge is indistinguishable from noise. It does not promise the strategy will make money going forward. Inputs are self-attested, and deflation cannot correct for variants you do not disclose. The gate behind every certificate is the one Ballast, a paper-traded fund, holds its own strategies to. Read that record before you trust ours.