@quotient
Last week, our signal on "SD finishes 2nd in Sweden's election" said 92¢. The market had it at 82¢
We were more confident than the market, and both of us were wrong. SD came in 3rd
Root cause: our pipeline had a hard-coded limit filtering out "duplicate" facts, including multiple sources confirming the same late-breaking poll. We were treating corroboration as noise
We've fixed this now: our newer models actually improve with more data, not less
We'll keep posting our misses, not just our wins
Hold 10M $QUOTIENT to follow how forecasts evolve in real time at signal.quotient.social