Accuracy how the risk forecasts did

Can this be checked? Every live card is also a line of a public ledger: each line carries the hash of the line before it, so a card added, removed or changed later breaks every hash after it. The first line is the first live card; the cards issued before the ledger existed were chained once, in the order they were issued. Once a day the channel gets the last hash and the card count of the day; /ledger/head shows the present ones, and the rugyomi.ledger verify script of the public repository recomputes the chain and checks those posts. The outcomes are published beside it (outcomes), not inside it: they are not known at launch.

What is counted. Only our live forecasts: a launch is counted if it received a risk card when it launched, and only once its 24-hour outcome is final. Launches from before the cards existed, and history we loaded later, are never counted. A rug is defined on the methodology page: the insiders sold into the other buyers within 24 hours and those buyers lost.

The live score is not calibrated yet. The weights are provisional (v0), a backtest on history is further down: a score of 70 does not mean a 70% chance. Compare a band's rug rate with the base rate, not with the score. Most launches fade anyway, so a raw hit rate means nothing without the base rate. Counts are small, so every rate comes with an interval; rows with fewer than 10 launches are greyed. Dev exit and Buyers' loss are the share of launches with those secondary facts (defined on the methodology page).

What the score is worth. Walk-forward check of 2026-11-11: each of 6 weeks (2026-W41 to 2026-W46, 1676 launches) was scored by models fitted only on launches whose outcome was already known when the week began, and compared with one fixed yardstick, the plain rug rate of the previous 2 weeks. A model counts as better only if it beat that rate in more than half of the compared weeks and on average. The best model, 2w/v1, beat that rate in 3 of 5 weeks but lost on average; we do not count that as better. The AUC of the launch-time score ranged from 0.47 to 0.58 by week (0.50 = a coin flip). The weekly rug rate in those weeks ran from 6 % to 47 %. The launch-time score does not yet beat the recent rug rate of the market.

Market context. Rug rate, last 14 days: 19 % (262 launches). Launches of the last 14 days whose 24-hour outcome was already known; compare any rate on this page with it.