The data-driven case for betting NO
People overpay for YES. We sell it to them.
Math says we win 7 out of 10 times.
Every 10 minutes, the system scans Polymarket — a site where people bet YES or NO on questions like:
"Will Bitcoin be above $64,000 on March 1st?"
"Will Sean Penn win Best Supporting Actor?"
Each question has a YES price and a NO price. They always add up to $1. So if YES is 55¢, NO is 45¢.
Core insight from the Becker dataset (408,000 real resolved markets):
Most things people bet YES on... don't happen.
People are optimistic. They bid up YES on exciting outcomes. The crowd overpays for YES. Result: NO wins 59-82% of the time depending on market type.
| Duration | NO Win Rate |
|---|---|
| Same day | 53.3% |
| 1-7 days | 59.8% |
| 1-4 weeks | 63.0% |
| 1-3 months | 77.6% |
| 3-12 months | 82.6% |
| Category | NO Win Rate | Sample |
|---|---|---|
| Sports | 69.7% | 6,267 |
| Geopolitical | 68.8% | 6,715 |
| Politics | 68.2% | 12,201 |
| AI | 67.6% | 1,687 |
| Other | 60.6% | 232,308 |
| Crypto/Price | 54.4% | 117,723 |
If all pass → bet NO.
We use the Kelly Criterion — a formula that answers: "Given my win rate and the payout, what % of my bankroll should I bet?"
We use 1/6 Kelly — bet 1/6th of the mathematically optimal amount. Safe enough to survive bad streaks, aggressive enough to grow.
| Fraction | Risk Level | Bet on $10K |
|---|---|---|
| Full | Reckless | $6,400 |
| 1/4 | Aggressive | $1,600 |
| 1/6 | Current | $1,000 |
| 1/8 | Conservative | $800 |
The market closes on its end date. Either:
NO wins → We paid 40¢, get $1 back = 60¢ profit per share ✅
YES wins → We paid 40¢, get $0 back = 40¢ loss per share ❌
Since NO wins ~70-80% of the time on our filtered markets, we profit over many bets.
All backed by the Becker dataset — 408,000 real Polymarket markets with known outcomes.
| Filter | Markets | NO Win Rate | EV / $1 |
|---|---|---|---|
| Everything | 76,781 | 58.8% | $0.24 |
| 7-365d, $50K+ vol | 24,040 | 67.2% | $0.32 |
| 30-365d, $100K+ vol | 6,493 | 79.0% | $0.44 |
The tighter the filter, the higher the win rate — but fewer bets. Our settings balance edge with opportunity.
Once enough resolved trades confirm the edge holds live, the system connects to real exchange accounts.