See what is actually training.
Candidate evaluations are counted separately from frozen models, so throughput stays visible without pretending every experiment is a new model.
CANDIDATE LEDGER4,096REGISTERED IN CURRENT EPOCH
QUANT RESEARCH / NO FAKE CERTAINTY
Trading models.
With receipts.
We search thousands of candidates, freeze one winner per registered round, then make it prove itself on market data it has never seen. Until the evidence is mature, it stays unranked. Simple.
RESEARCH TOOL. NOT INVESTMENT ADVICE.
4,096 REGISTERED CANDIDATES
2,956 UNIQUE ENGINE RUNS
8 FROZEN WINNERS
0 MATURE RANKINGS
CURRENT EPOCH SNAPSHOT / AUG 22, 2026.
NO FORMING BARS
NEXT-BAR OPEN FILLS
ONE EQUITY ENGINE
UNDEFINED MEANS NULL
WHAT YOU ACTUALLY GET
Open the model.
Check the work.
No magic score. No giant return number with the ugly parts hidden. You get the model, its signal chart, its costs, its sample size, and a plain answer about whether the evidence is ready.
Price, signals, stats. One page.
Open any frozen winner and see its market chart, LONG, SHORT and OUT signals, costs, drawdown, trade count, exposure and evidence status.
Experimental means experimental.
You can pick an immature model to watch it, but we make you confirm the warning first. A pick is not a validation stamp.
EVIDENCE STATUSCOLLECTINGNOT ELIGIBLE FOR RANKING
HOW IT WORKS
Search hard.
Trust slowly.
- 01
Search development data.
Each registered round searches development data only. The locked holdout stays untouched.
- 02
Freeze exactly one winner.
One artifact wins by preregistered return and drawdown rules. We hash it, lock it, and score that exact artifact once.
- 03
Start forward evidence from zero.
The nominee and benchmark see the same finalized bars. No model ranks until duration, periods, trades and exposure all mature.
NO VICTORY LAP YET
Zero ranks.
On purpose.
The old system produced impressive-looking numbers too easily. We kept the raw history, flagged the old model pages, and restarted promotion-grade evidence. If the answer is not ready, the site says "Insufficient evidence."
SEE THE CURRENT RESULTS STATUS ↗One selected artifact per registered round.
No immature model gets dressed up as a winner.
WHY THIS EXISTS
LETS F*ING
DEMOCRATIZE QUANT.
quant funds have kept the process behind closed doors for decades. we are opening it up. not just the winners. the search, the failures, the costs, the warnings, the weird nulls, all of it. if a model earns trust, you will see why. if it does not, you will see that too.
FREE ACCOUNT / NO CARD
See the research
before the rank.
Browse the frozen models, open their detail pages, inspect the signal charts, and watch forward evidence grow. You do not need to pretend the answer is ready to start paying attention.
Checking the founding offer...