The highest-scoring voices on record are those who challenged consensus at peak confidence: Burry on subprime (2005), Rajan at Jackson Hole (2005), Summers on ARP inflation (2021), Tufekci on COVID (Jan 2020). Being early and right is systematically undervalued in real time.

Central banks and official institutions score lowest on directional calls made in public. The Fed’s “transitory” consensus (2021) and Bernanke’s “contained” call (2007) illustrate the pattern: institutions protect consensus longer than individuals do.

Several high-profile voices (Roubini, Rosenberg) score poorly not because they are wrong directionally but because they never update. Being right once on a big call does not redeem years of incorrect follow-through. The record is permanent.

“Tesla will reach $3,000” with no timeframe is not a prediction — it is a statement that can never be proven wrong. We track these separately and exclude them from accuracy scores. A high unfalsifiable count is itself a signal: this voice is optimizing for appearing bold rather than being accountable.

Political actors’ economic predictions correlate strongly with their policy preferences, not with economic evidence. Track the rare exceptions — they signal genuine insight rather than motivated reasoning.