State of the Lab — week ending Oct 4, 2026

This week the lab ran 1 seat across 1 desk and settled 34 trades, every one of them real money. Best seat: Fed decision · gap (new). Roughest seat: Live · natural gas 15-min.

The lab traded no real money after Tuesday. The last live lane, natural gas, hit the lifetime stop it was given before its first order, at minus four dollars and twenty seven cents after 185 of its 250 orders. The 34 real trades on this week's board, minus two dollars and eighty nine cents between them, are its final ones. October is paper only, by decision, and the two leftover live units were switched off and removed from the watchdog on Saturday so nothing can quietly restart them.

That does not mean the lab is idle. A desk with no live lane is the right moment to go back and check the evidence for the next one, and this week that check paid for itself. Our best crypto paper result, following the perpetual futures price at about one and eight tenths of a cent a trade over six thousand trades, turned out to be measured against a price that was about twenty seconds old. Rerun on Wednesday against the price that was actually on offer at the moment of each decision, the same idea loses about seven cents a trade. The full study is out this week, and it is the most useful thing we have written for anyone building a bot: it explains how a strategy can look good on paper for weeks for no reason at all.

The new paper lanes all have their pass marks written down before their first trade, and most are judged at the end of the month. The late favourite sniper was restarted on Friday after we found a bug that made its safety filter read a stale value, and every fill before the fix is void. A last minute Bitcoin rule, the tennis favourites and the movie score favourites are all collecting fills. The recurring political speech lane is the furthest along, at about two and a half cents a trade on honest fills, but one Trump speech is three quarters of that result, so by its own rule it has not passed.

We also published the whole real money record in one place: fifteen strategies across crypto, oil, natural gas, weather and esports, about six thousand eight hundred real trades, five of them finished ahead. The strategies page now lists every one, and so does a video going out on Friday. It is the most honest head start we can give anyone who wants to try this: what we tried, what it cost, and where the gap between paper and real money showed up.

For sports readers, the first reading of Kalshi against the sportsbooks is in. On 49 NFL games at the last price before kickoff, Kalshi with its fee was the cheaper price 54 times and DraftKings 44 times, close to a tie. In 44 of the 49 games each was cheaper on a different team. College football leans Kalshi, about two times in three. We will rerun it every week through the season.

And a second arena has opened. Alongside Kalshi, the lab is now testing rules on US exchange traded funds through a paper brokerage account. Four ideas were written down on the second of October before any of them was tested, with a bar each had to clear. Three failed and are closed. One, a monthly trend rotation, passed and is now running on paper with one thousand paper dollars. Its first verdict is not before the second of December, and it is judged on whether it can be run exactly as tested, not on two months of returns.

Each desk name links to The Seats, which gives every strategy's real trigger and threshold plus the Kalshi tickers it trades, enough to run any of them yourself. The one exception is market making: which markets it quotes is listed, how it quotes them is not, and the About page explains why.

All 160 seats behind these five lines, with win rates, fee drag, all-time results and the per-trade CSV, are in the members edition. 13 seats retired to date; their records stay on the board.

The market-making seats are simulated fills, and a simulation cannot model adverse selection: the tendency of a resting order to be taken exactly when someone better informed wants the other side. Where this lab has run the same strategy with real money, the live result has come in materially below the paper one. Read the maker desk as an upper bound on what the idea could earn, not as what it would have earned. The live desk below it is that check running in public: same ideas, one contract at a time, settled by the exchange.

Fill verification: 34 of 38 trades (89%) were checked against real order-book depth.

The rest are entered at the ask the exchange reported, but the lab does not record the book for every market. Challenger tennis, MLB, soccer, basketball and football have no depth archive, so those fills cannot be re-checked after the fact. Trades whose price level had no size behind it are discarded, not counted.

The archive. The lab records and permanently keeps the raw market data behind every number it publishes: trade tape from 2026-08-15 (49 days); order book depth from 2026-07-28 (68 days); top of book ticks from 2026-07-27 (69 days); settlement index from 2026-08-16 (49 days); 15-minute market quotes from 2026-07-21 (75 days). 8.9 GB and growing nightly. None of it can be recreated after the fact, which is why it gets recorded. It stays in the lab: the exchange reserves its own market data and this lab does not pass it on. What gets published is what the lab finds in it.

Every trade above was logged live, then sealed: each night the lab hashes its raw market archive and trade ledgers into an append-only chain. Rewriting any past trade would change every head after it, so check this one against last week's issue.

Chain head as of 2026-10-04T04:54Z: 93bcc534433ef60ffe20a369e44abc8bbb194848b818de2b00761a1ed838979e

The full board for this week, every seat with its win rate and fee drag, is in the full board for members.