Selling longshots

People overpay for lottery tickets. On a prediction market, that means low-probability “Yes” shares trade above their true odds. The question: is this bias collectable by being the seller — posting resting orders and letting retail buyers take them — across the whole platform?

1. Retail overpaysBuyers pay more than the odds justify for unlikely “Yes” outcomes.
2. We are the sellerResting orders wait; buyers come to them. No chasing, no speed needed.
3. The gap is the payAveraged over many fills, the overpayment is the return.

Discovery: April–September 2026

Positive in all eight categories, strongest in tech, culture, finance and crypto. Two refinements came out of the details:

• It is a “Yes” bias, not a general longshot bias. Selling cheap “No” earns almost nothing. People overpay for “the unlikely thing happens”.
• The edge lives where retail keeps buying. Counting each trade equally: +5% to +6% at every trade size. Counting each market equally: about zero. That is exactly where a resting order gets filled.

Holdout: 2025, rules fixed in advance

A sister idea had just failed its holdout after being measured on the same calm months, so the rules were written down and committed, then run once on 2025:

Positive in every category again — but the category ranking changed completely. The bias is platform-wide; picking categories is noise. Still unknown, and only measurable live: how often a new maker actually gets filled.

Related ideas tested here