Honest crypto quant research

Most edges are statistical mirages. Ours are statistical outliers.

We take your favourite trading strategy, try everything we can to kill it: fee-realistic, look-ahead-free, intra-candle honest, tested across multiple market regimes, and we publish the results. Both the graveyard and the rare all-weather strategies that survive. Read the refutations for free.

Education & data-journalism only. Never signals. Never personalised advice.

The Refutation mark — a 3-D parameter-sweep cube: refuted configurations in terracotta, the surviving optimum climbing to green
The running tally · strategies through the guillotine last updated: 7 August 2026 · and counting
7,305
distinct strategies tested
5
refutations in the graveyard
3
survivors running live on real money
4
member letters
105 million backtests run · and counting

Bridging the gap between institutional and retail trading strategies.

The method

Falsification, not hype.

This is for anyone serious about markets, from a first trade to a professional book. You deserve research that shows its work and owns what failed, and being able to put that in front of you is a privilege. So every strategy gets the same honest approach: build its best version, try hard to break it, and publish all of it.

01 · SWEEP THE PARAMETERS

Optimise the strategy

We tune the parameters to build a strategy's strongest, best-case version — real fees, no look-ahead, enough data that one lucky run can't hide. Give it every advantage, then try to break it.

02 · TRY TO KILL IT

Run the chamber

The best-case version goes into our proprietary backtest chamber: scored out-of-sample on held-out market regimes it was never tuned on, with one-independent-trade-at-a-time checks that expose the mirages a single tidy curve is hiding.

03 · COMPLETE THE AUTOPSY

Publish everything

We publish all of it — the rare survivors and the graveyard alike: what looked promising, why it turned out to be a mirage, and the exact test that caught it.

Reproducible method notes

The statistics, in plain English.

You don't need a maths degree to tell a real edge from a lucky one: four core ideas do most of the work, shown on real charts instead of formulas. We'll teach you the tools. What we build with them is the letter.

+2σ +1σ mean 2σ move — rare
Note 01 · the ruler

Standard deviation

How far is “far”? A move only means something once you measure it against normal noise. One standard deviation is ordinary; two is rare. It turns “that looks like a big candle” into a number you can actually test.

plateau = robust spike = luck
Note 02 · the map

The parameter surface

Trust the plateau, not the peak. Sweep two settings and every backtest becomes a landscape. A real edge is a broad mountain — nudge the dials and it still works. A lone bright spike is just a setting that happened to fit the past. We only keep the plateaus.

LEAVE-ONE-REGIME-OUT repeat ×6 trained on 5 regimes UNSEEN crash survive the era it never saw — or it was overfit.
Note 03 · the proof

Leave-one-regime-out

Prove it on markets it's never seen. Hold out a whole market era — a crash, a long grind — build the edge on everything else, then test it on the part it was never shown. Survive the unseen regime, or it was just memorising the past. This is the test almost nothing passes.

t = SIGNAL / NOISE your edge signal blindfold darts = noise Land clear of the scatter, or it's just luck.
Note 04 · the verdict

The t-statistic

Signal over noise. Before trusting a result, shuffle the same trades into random order thousands of times: blindfolded darts at the same board. That random mix is the noise floor, what pure luck alone would score. The t-statistic is simply your edge divided by that noise. Below about 2, the result sits inside the scatter and could be chance. Well above it, the signal stands clear of the noise.

The survivors

The graveyard is free. The survivors aren't.

The refutations are open to everyone, free. The survivors are emailed in the paid monthly letter. Fund a quant researcher and get the exclusive strategies that survived, for the cost of brokerage fees each month.

4 member letters published so far — the full archive ships with every plan.

Free
$0

The Refutations

  • Public teardowns of edges that didn't survive — with the data and the exact test that killed them.
  • The statistics-mirage checklist: the ten ways a backtest lies to you.
  • The reproducible method notes — standard deviation, parameter surfaces, out-of-sample testing — taught with real charts, not formulas.
Get the free teardowns

Every paid plan gets the same letters. Cancel anytime; access runs to the end of the period you've paid for.

The house speciality

Ten ways a backtest lies.

Every survivor has to clear all of these. Most “edges” you'll read about elsewhere fail at least one — quietly.

MIRAGE 01

The in-sample glow

Gorgeous on the data it was tuned on; a coin-flip on data it's never seen.

MIRAGE 02

The fee wall

A tidy edge before costs. Then the arithmetic of spread, slippage, and funding erases it before you ever profit.

MIRAGE 03

The regime mirage

Prints money in one bull run, quietly bleeds everywhere else. Validation has to span regimes.

MIRAGE 04

The tail carry

Ninety percent of the “profit” is five lucky trades. Take them out and the edge is gone.

…and six more. The full checklist ships free to every subscriber.

Inside the letter

A survivor teardown.

The refutations are free. The survivors — which all-weather strategies lived, and exactly how they held up under the method — are for premium members. Here's the shape of one.

Case file · subject

🔒

🔒

Survivor teardowns are for premium members only.

The refutations are free. The tactics that survived — and exactly how — live in the letter.

Unlock with a subscription · US$29.90/mo
R

“Most strategies only work in one kind of market. I look for the all-weather few you can run all year, every year. It’s an honour to put a full year of that research in your hands for less than a single small trading loss.”

The operator behind The Refutation
Who writes it

A decade in markets, now deploying the survivors.

No social-media guru vibes. No rented Lamborghini. The Refutation is written by a mathematics head teacher with bachelor's degrees in mathematics and chemistry. He has spent a decade trading and investing his own money, got tired of backtests that were fiction with a P&L attached, and applied statistics and the scientific method to build a robust backtesting chamber. He now runs that research live and publishes it here. You don't have to trust the author. The graveyard of everything that failed is the work it took to find each survivor.

That's the whole moat: honesty you can verify. A free graveyard of scrutinised strategies, and the survivors that pass, for the cost of brokerage fees.

Start free

Get the refutations. Then decide.

Free teardowns and the statistics-mirage checklist, straight to your inbox. Upgrade to the paid letter (US$29.90/mo, or US$299/yr) only once the process has earned it.

No spam, no signals, no “urgent alerts.” Unsubscribe anytime.