Common Human Biases in Betting

A thread with Claude (Opus 4.5).

1. Overconfidence

“I’ve researched this. I’m 90% confident.”

Studies show that when people say “90% confident,” they’re right about 70% of the time. The gap is consistent:

         Said     Actually
         ─────    ────────
         99%      ~85%
         90%      ~70%
         75%      ~60%

Your edge is smaller than you think. Full Kelly based on inflated confidence = ruin.

2. Favorite-Longshot Bias

“100:1 odds! If I’m right, I’m rich!”

People overbet longshots and underbet favorites. Why? $1 to win $100 is exciting. $95 to win $100 is boring.

But longshots are usually overpriced (too many dreamers), and favorites are often underpriced (not sexy enough). The “boring favorite” is often where the edge lives.

3. Narrative Bias

“Two tech billionaires will get into an ego war over this card! It’s going to $20M!”

Good stories feel probable. But stack the actual probabilities:

P(two billionaires exist who want this card) × P(both show up to this auction) × P(neither has budget ceiling) × P(ego overcomes financial sense) = small number.

The story is vivid. The probability is not.

4. Ignoring Base Rates

“This is THE rarest card! Logan Paul wore it to a boxing match! It’s special!”

The instinct is to start from the narrative. The correct move is to start from the numbers: What do PSA 10 ultra-rare cards sell for? What’s the base rate of $10M+ card sales? What % of hyped auctions meet estimates?

Base rate of $15M+ card sales in history: ~0. Your prior should be low, then update.

5. The Certainty Effect

People treat 95% as “basically certain” and 5% as “won’t happen.”

But 5% happens 1 in 20 times. Make 100 bets treating 5% as impossible, and you’ll be catastrophically surprised about 5 times.

         ┌─────────────────────────────┐
         │  ████████████████████  95%  │
         │  █                     5%   │ ← This happens
         └─────────────────────────────┘

The 5-12% loss scenario is not zero. Size accordingly.

6. Correlation Blindness

“I’ll diversify! Bet on >$7M AND >$8M AND >$10M. Three bets is safer than one!”

If the sale fails, all your YES bets lose together.

         Sale fails:
         
         YES >$7M    ████ LOSE
         YES >$8M    ████ LOSE  
         YES >$10M   ████ LOSE
                     ▲
                     └── "Diversification" was fake

You wanted three independent bets. You got one big bet wearing three hats.

7. Anchoring on Fake Data

“A PSA 9 sold for $4M! So PSA 10 is worth $7M+”

That sale never completed—buyer didn’t pay. Real comparable: $625K at Heritage (Aug 2025).

TRUE ANCHOR          FAKE ANCHOR
    │                    │
    ▼                    ▼
  $625K               $4,000K
    │                    │
    └────── 6x gap ──────┘

The fake number is in your head now. Your estimate is anchored too high.

The Meta-Bias

The biggest mistake is treating betting as “picking winners” instead of “pricing probabilities.”

Wrong frame: “Will it sell for >$15M? …Probably not. Pass.”

Right frame: “Market says 12% chance. I think it’s 5%. That’s a 7% edge on NO. Size accordingly.”

You’re not predicting the future. You’re finding prices that are wrong.