Cross-Category Correlations
Does the real Fed-rate outlook move with real AI-market sentiment? Does Bitcoin move with either? A real Pearson correlation over the daily archive every one of these pages already keeps — no new data, no new fetch, just math over what's already been recorded.
Fed Funds Rate (contested threshold) vs Bitcoin Price
Not enough overlapping archived days yet to compute a real correlation — both series need at least 5 shared real calendar days, and this pair's daily archive is still accumulating. Check back as more days record.
Fed rate side tracks the "Above X%" contract closest to 50/50 each day — the threshold the market is actually contesting, since most rungs on Kalshi's rate ladder sit pinned near 0% or 100%.
AI Sentiment Index vs Bitcoin Price
Not enough overlapping archived days yet to compute a real correlation — both series need at least 5 shared real calendar days, and this pair's daily archive is still accumulating. Check back as more days record.
AI Sentiment Index is a volume-weighted average "Yes" price across open AI-tagged Polymarket markets (see snapshot-odds.mts).
AI Sentiment Index vs AI Model Leader (top lab confidence)
Pearson r over 6 real overlapping days
Low confidence — only 6 real days so far, close to the 5-day minimum this needs just to compute. One more day's move can swing this a lot; treat it as provisional.
Model Leader side is the current front-runner lab's own probability — which lab is in front can change day to day.
Fed Funds Rate (contested threshold) vs AI Sentiment Index
Pearson r over 14 real overlapping days
Low confidence — only 14 real days so far, close to the 5-day minimum this needs just to compute. One more day's move can swing this a lot; treat it as provisional.
Tests whether real Fed-rate expectations and real AI-market sentiment move together, purely from archived market prices.
What This Is (and Isn't)
Pearson's r measures how closely two real series move together on the same real days, from -1 (perfectly inverse) through 0 (no linear relationship) to +1 (perfectly together). It's computed only over calendar days both archives actually recorded a real point — never interpolated, resampled, or backfilled.
A correlation is not causation, and a short real history (these archives add one real point per day) makes any coefficient noisier the fewer overlapping days exist — the day count next to each result above is the honest gauge of how much to trust it.