Backtest Results

Important: These results are based on simulated trading, not live trading. Historical mid-window Polymarket orderbook data is not publicly available, so market prices in this simulation are estimated. Actual results depend on real market liquidity, execution speed, slippage, and fees. Past performance does not guarantee future results. Never trade more than you can afford to lose.

We tested our pricing model against 7,815 resolved Polymarket Up/Down markets across all 6 assets and 4 timeframes. Here are the results.

Model Accuracy

Metric	Value	What it means
Brier Score	0.1847	26% better than a coin flip (0.25 = random)
AUC	0.794	Strong predictive power
Prediction improves over time	0.25 → 0.11	Gets more accurate as the window progresses

Accuracy by Asset (at 50% window elapsed)

Asset	5m AUC	15m AUC	5m Win Rate	15m Win Rate
BTC	0.754	0.824	70.7%	76.2%
ETH	0.770	0.859	73.0%	81.5%
SOL	0.797	0.813	75.0%	76.8%
XRP	0.776	0.849	72.7%	79.6%
DOGE	0.785	0.871	74.4%	81.9%
BNB	0.804	0.847	76.9%	78.2%

Simulated P&L

We simulated a trading bot that enters at the 50% mark of each window when it detects a 3%+ edge, betting $100 per trade.

Since historical mid-window Polymarket prices aren't publicly available, we tested two scenarios for how efficient the market might be:

Scenario	Assumption	Meaning
Half-efficient	Market moves halfway to fair value	Others trade, but slowly
Conservative	Market trails fair value by only 3 cents	Tight market — small edge only

In reality, market efficiency varies. Some windows will be closer to the conservative scenario, others may offer more edge. Treat these numbers as a range, not a guarantee.

Results (7-day period, $100/trade)

Scenario	Trades	Total P&L	ROI
Half-efficient	10,450	+$146,635	+22.5%
Conservative	11,414	+$52,252	+6.5%

Portfolio Performance (Half-Efficient, All Assets)

Metric	Value
Total trades	10,450
Win rate	76.3%
Total P&L	+$146,635
ROI	+22.5%
Avg profit per trade	+$14.03
Max drawdown	-$1,241

Daily Consistency

Date	Trades	P&L	Win Rate
Mar 9	222	+$3,068	76.6%
Mar 10	372	+$5,822	78.0%
Mar 11	1,508	+$19,187	75.2%
Mar 12	2,099	+$28,744	76.2%
Mar 13	2,085	+$33,459	78.4%
Mar 14	2,091	+$24,888	74.2%
Mar 15	2,073	+$31,468	77.1%

Top Opportunities by Risk-Adjusted Return

Rank	Asset	Timeframe	Win Rate	ROI	Avg $/trade
1	XRP	1h	87.2%	+38.1%	+$24.07
2	DOGE	1h	85.6%	+35.4%	+$22.39
3	ETH	4h	85.3%	+36.0%	+$22.58
4	ETH	1h	82.3%	+31.1%	+$19.53
5	ETH	15m	81.3%	+28.8%	+$18.17

Key Takeaways

1. Longer timeframes = higher edge. 1h and 4h markets consistently show the best win rates. These markets are less efficient and have more opportunity.

2. The model works across all 6 assets. Every asset shows positive ROI even in the most conservative scenario.

3. Consistent daily performance. No losing days across the 7-day test period.

4. ETH is the best all-rounder. Strong performance across all timeframes with the highest volume.

Important Disclaimer

This backtest uses simulated market prices — not real historical Polymarket orderbooks. Key limitations:

• Real orderbook depth and slippage are not modeled

• Execution latency is assumed to be zero

• Market efficiency may increase as more traders adopt similar strategies

• A 7-day test period is short — longer-term consistency is not proven

Past performance does not guarantee future results. Trade responsibly and only risk what you can afford to lose.

---

Backtest run: March 17, 2026 | Data: March 9-15, 2026