How to Predict the Future (Mathematically)
Credibility score: 88/100 — Highly Credible. This video is highly credible with well-supported claims.
Claims analyzed
Coin flip prediction game: $1 per correct guess over M rounds, max $M or $0. — Just Vibes (50/100)
Straightforward setup for a prediction game — classic thought experiment to hook you on probability math.
Designated coin flipper expert always predicts correctly because they know the result — Solid (85/100)
Logically sound setup — flipper knows the outcome, so 100% accuracy as long as they report truthfully.
Repeatedly eliminating wrong experts finds the designated one eventually — Solid (85/100)
This is straight out of online learning theory — the halving algorithm converges to the best expert. Solid math.
Best case: 1 round if others wrong immediately; worst case many rounds if experts rarely err — Verified (95/100)
Nailed the bounds perfectly — best case instant elimination, worst case logarithmic in errors. Math checks out.
Majority vote of remaining experts; eliminates at least half if wrong — Verified (100/100)
Classic majority voting — guarantees halving the pool on mistakes. This is textbook ensemble learning.
Halving algorithm makes at most log n mistakes — Verified (100/100)
Dead on — classic halving algorithm bound in online learning. Math checks out perfectly.
Algorithm works for any binary prediction like stocks or rain — Solid (85/100)
Correct generalization — applies to any binary sequential prediction. Examples spot on.
Algorithm needs perfect expert; rarely guaranteed in practice — Verified (100/100)
Nailed the key limitation — perfect expert assumption rarely holds IRL.
Halving fails if all experts wrong once; use weights instead — Verified (100/100)
Spot-on failure mode analysis — straight to Weighted Majority fix. Smart.
Trust starts at 1, stays same if correct, halved if wrong — Verified (95/100)
Standard weighted majority setup — checks out perfectly. Classic algo move.
Mistake when ≥ half total trust predicts wrong — Verified (92/100)
Spot-on definition of when weighted majority errs — no notes.
Halving keeps ≥1/4 correct trust, loses ≥1/4 wrong, total ≥3/4 prior — Solid (88/100)
Math holds tight — bounds are correct for β=1/2 halving parameter.
Total trust ≥ (3/4)^T; algorithm makes M mistakes, each multiplies trust by 3/4 — Solid (85/100)
Math setup holds — classic trust aggregation in prediction algorithms. Clean bounds.
After M mistakes, total trust ≤ n * (3/4)^M — Verified (95/100)
Inductive bound is textbook correct — trust decays geometrically with mistakes.
Best expert's final trust = (1/2)^{best} after best mistakes — Verified (100/100)
Best expert trust halves per mistake — pure math, checks out perfectly.
Algorithm mistakes ≤ 2.4 × best expert's + constant — Verified (95/100)
Spot on — that's the exact bound from the classic Weighted Majority Algorithm. No cap.
Experts rep outcomes; trust = prob; penalty = 1 - εL — Solid (85/100)
Standard extension to multi-outcome via MWU — penalty factor is textbook multiplicative update.
Advanced algo gets 1× best expert performance (in expectation) — Solid (80/100)
True in expectation for randomized MWU — regret vanishes relative to best, factor →1 asymptotically.
Technique lets non-experts predict almost as well as top experts — Solid (80/100)
Checks out — no-code AI tools and citizen data science make this real. But 'almost as good' needs the full context.
See the full analysis with sources and timestamps →