Membership
Free, Member and Desk — what each tier includes, and why the Desk tier lets you choose what the lab tests next.
Everything here is measured the same way and published on the same schedule. What changes between tiers is how much of the evidence you get, and whether you get to point the lab at your own question.
- The weekly story: what happened and why
- Desk-level results, including the losses
- Every study's finding in full
- The Seats: what a seat is, and which markets the lab trades
For anyone who wants to know what the lab found.
Subscribe free- The Seats in full: every strategy's real trigger and threshold
- The full weekly board: every seat, win rate, and fee drag
- All-time results beside the week, so nothing hides
- Every past issue and full board, back to the lab's first week
- Everything in Free
For readers who don't take a conclusion on trust.
Become a member- Propose and vote on what the lab tests next
- Operations: what Kalshi costs you before your strategy matters
- Every past week's per-trade file, back to the lab's first week
- The per-trade file behind every weekly board, emailed
- Methodology notes to reproduce or extend any study
- Propose and vote on what the lab tests next, and your experiment runs live
- Everything in Member
For people who want to run their own questions through the machine.
Join the deskWhat the Desk tier actually means
The per-trade file is the obvious part, and it's the smaller part. The reason the Desk tier exists is The Queue: members propose hypotheses and vote on which one the lab takes next. The winner gets a seat, a clock, and a published verdict, whether it works, doesn't, or comes back inconclusive.
Describe an idea precisely enough to be falsified, with an entry condition, a market, and what would count as it working, and it can be a seat. A month later the answer exists in public with the trade file attached, credited to whoever asked.
Most of them won't work. That's the point: the lab is worth running because it can afford to be wrong on paper, repeatedly, in a way a real account cannot.
Two honest notes
One exclusion. Market-making fills are withheld from the per-trade file. Every other seat is a directional bet whose edge survives being copied; a quoting book's doesn't, because it competes for the same spread, and on markets this thin a handful of new quoters compresses it for everyone. The maker desk's P&L is reported in full on every board; only the fills are held back.
None of this is advice. No funds are managed, no trades are placed for anyone, and nothing here is a recommendation. It's a research publication about a market, written by someone running experiments on it, and who trades it with his own money, which is stated plainly on the About page.
The data archive (Desk)
What Desk members get, stated precisely so nobody is oversold. Every per-trade file behind every board the lab has published, week by week back to its first week, served at /data/ and gated by your login. Access follows your subscription rather than your join date. These are the lab's own records: which seat traded, at what price the book was showing, what fee applied, and how it settled. They are what you need to recompute any figure the lab prints. What Desk members do not get is Kalshi's market data itself. The lab records the raw order book continuously for its own research, but Kalshi's data terms reserve redistribution, so those recordings stay in the lab and what reaches you is what the lab found in them.