We Backtested Polymarket's Most Popular Scalp Strategy on 428 Trades — It Loses Money

Quick glossary (for readers who don't trade daily):

• ROI — Return on Investment. How much profit you make for every dollar you risk. +10% ROI means you turn $100 into $110.
• Win rate (WR) — percentage of your bets that win. If you make 100 bets and 92 win, your WR is 92%.
• Breakeven WR — the win rate you need just to not lose money, given the price you paid. If you buy a share at 94¢, your breakeven WR is 94%. Below that, you lose on average.
• Expected value (EV) — your average profit or loss per bet if you ran the bet thousands of times. Positive EV is profitable; negative EV is a slow bleed.
• Slippage — when you actually fill at a worse price than you wanted (0.95 instead of 0.94 because someone else got there first). Every cent of slippage raises your breakeven WR.
• Gas — the small network fee you pay to execute a trade on the blockchain. On Polygon (Polymarket's network), typically a few cents.

There's a strategy every new Polymarket trader discovers within about a week of paying attention. You scan the ultra-short-horizon markets — "Will Bitcoin be higher in 15 minutes?" — and you wait. When one side hits 0.94, you buy $100. You win roughly 19 times out of 20. Free money, right?

We ran the math over 428 historical entries across every BTC, ETH, and SOL "updown" 5-minute, 15-minute, and hourly Polymarket market in our database. Here is exactly what happens when you run that strategy at scale.

The strategy, stated precisely

• On any active btc/eth/sol-updown market, watch both outcome prices.
• When one outcome reaches ≥0.94 (and we assume we can execute at that price, zero slippage, zero gas — the generous assumption), buy $100 worth.
• Hold to resolution. If your outcome settles at 1.0, you collect $100/0.94 = $106.38 (profit $6.38). If your outcome loses, you lose $100.
• Filter: we only count markets that actually reached 0.94+ on at least one side and then resolved.

The backtest result

Asset	Timeframe	Entries	WR	Avg entry	Net P&L	ROI
BTC	5m	341	96.2%	0.956	+$196	+0.57%
BTC	15m	57	87.7%	0.958	−$484	−8.49%
ETH	5m	14	64.3%	0.955	−$458	−32.69%
SOL	5m	8	50.0%	0.960	−$381	−47.63%
ETH	15m	1	100.0%	0.959	+$4	+4.23%
BTC/other	mixed	2	50.0%	0.942	−$94	−46.81%
ETH/other	mixed	2	0.0%	0.943	−$200	−100.00%
SOL/other	mixed	3	33.3%	0.953	−$196	−65.28%
Total	—	428	92.1%	0.957	−$1,612	−3.77%

The first row is the only configuration that doesn't lose money. BTC 5-minute, +0.57% ROI. Half a percent.

Every other configuration — BTC 15-minute, any ETH, any SOL — loses decisively.

Try it yourself — the math is ruthless

Before we explain the math with words, try the simulator below. Drag the entry-price slider to 0.94, set your win rate to 92%, pick a bet size, and hit simulate. Run it again. Again. Watch your $10,000 bankroll erode. This is the trap — it feels like you're winning (you are, 92% of the time!), but the expected value is negative and the bankroll grinds down.

Now drag the win rate up to 95%, or drop the entry price to 0.70. Watch the "Expected P&L" flip green. That is what positive edge looks like. The scalp strategy doesn't have it; momentum and contrarian strategies on longer-horizon markets do.

Why 92.1% win rate loses money

The math is brutal once you stare at it.

Breakeven win rate at price p is simply p. If you buy at 0.94, you need to win 94% of the time just to avoid losing. 92.1% WR across the full cohort is below that threshold. The 5.5-point margin the unsophisticated trader thinks they're capturing is worth ~1.9 points per trade at best — and on the 5-minute micro-markets, that's before you account for anything that eats the margin.

The BTC 5-minute cohort clears breakeven by 2.2 percentage points (96.2% WR vs 94% breakeven). That's the +0.57% ROI. Now subtract:

• Gas on Polygon: $0.01–$0.05 per transaction. On a $100 bet, that's 0.01%–0.05%. Negligible, but real.
• Slippage: at thinner moments you'll fill at 0.95 instead of 0.94. Raises your breakeven to 95%. The 2.2-point margin collapses to 1.2.
• Opportunity cost: 5-minute holds at 100% of capital works out to astronomical APY in theory. In practice, capital utilization is low because you can't always find a qualifying market.
• Sample size: 341 trades is a lot but not enough to claim +0.57% is statistically distinct from breakeven. The confidence interval on a 96.2% win rate at N=341 runs from roughly 93.5% to 98.2%. The low end is below breakeven.

Put those together and you get a strategy that is, at best, a rounding error.

Why does the rest of the market behave worse?

The cross-asset split is interesting. BTC 5m is marginally positive; everything else is negative.

Our hypothesis: Bitcoin price in 5-minute windows has the lowest volatility of the crypto set, so its 0.94 signals stick. By the time Polymarket's BTC-updown market prices 0.94 on "up", the actual spot BTC is moving very slowly relative to the window.

ETH and SOL have more violent short-horizon moves. A 0.94 "up" price on the SOL-updown 5-minute can reverse in under 60 seconds if a big block trade hits. The signal is noisier. WR collapses from 96% on BTC to 50–64% on ETH/SOL — enough to turn the trade into pure negative EV.

This isn't novel finance insight. It's just that Polymarket traders assume all crypto markets behave like BTC, price the strategy as universal, and then get blown up on the assets with higher vol.

The cognitive trap

High win rates are psychologically satisfying. You watch the wins stack up and feel like a genius. The losses, when they come, feel like bad luck — "this never happens."

But at 0.94 entry, a loss costs you 15.7x what a win makes ($100 vs $6.38). A single bad streak — 4 losses in a row at $500 size — wipes out a month of wins. We've seen it happen.

This is the same mathematical structure as selling insurance: you collect small premiums almost every time, you pay large claims rarely, and your edge (if any) lives in the 2nd decimal of the frequency of claim events. Professional insurance companies employ actuarial teams, reinsurance, and decades of catastrophe data. A Polymarket trader running a 0.94-entry strategy with $10k has none of those.

What we did with this finding

• Polyloly's Speed Traders leaderboard explicitly filters OUT endgame scalpers by restricting entry-price range to 0.10–0.70. The traders who show up are ones whose edge comes from information, not pricing structure.
• Our paper-trade bot deliberately skips the updown/hourly categories. We tail insider wallets on political, geopolitical, and sports markets instead — where payoff ratios are 2:1 or 5:1 and even a 60% WR generates strong ROI.
• The cohort ROI analysis we published alongside this shows the full picture: scalp is the WORST cohort on Polymarket by ROI, behind every non-degen alternative.

Caveats because we're honest

• 428 trades across three assets and four timeframes is a decent sample but not overwhelming. If we had 4,280 we could say stronger things.
• Our zero-slippage assumption is charitable. The real performance is worse than −3.77%.
• One specific wallet in our data (Zany-Imbalance, in our deeper dive) ran 44 BTC-updown trades at 95.5% WR for +1.9% ROI over 3 days. Tiny sample — consistent with 428-trade aggregate but not individually proof of anything. You cannot extract the edge by picking the lucky 3-day window and extrapolating.
• ETH-updown and SOL-updown have smaller samples (14 and 8 respectively). Those ROIs are noisier. But the sign of the effect — negative — is very consistent with the payoff-ratio math.

What to do instead

If your goal is positive EV on Polymarket, don't scalp crypto updown markets. The math does not support it and the empirical data confirms the math.

• Geopolitical contrarian medium-size bets print +81.6% ROI in our analysis. Low WR (35%), high payoff per hit.
• Insider-tail (copy wallets with proven long-horizon edge) paper-traded to +46.7% ROI over 334 positions.
• Momentum on political markets with small tickets ran 87% WR at +16.7% ROI — the closest thing to a lazy profitable strategy.

All three beat scalp by orders of magnitude. The scalp trade exists because it's psychologically comfortable, not because it works.

---

About the author

Poly Loly — Prediction Markets Expert

Lead analyst behind Polyloly, a real-time analytics platform tracking whale positions across $1B+ in monthly Polymarket volume. Focus areas: on-chain data aggregation, insider-detection heuristics (80%+ win-rate flags on resolved markets), and market microstructure across political, sports, crypto, and esports prediction markets. Published daily trading-terminal intel, trader leaderboards, and automated alerts via @PolylolyHi.

🌐 polyloly.com · 𝕏 @PolylolyHi · ✉ hi@polyloly.com

---

This article is for informational purposes only and does not constitute investment advice. Prediction markets carry a risk of capital loss.

This article contains affiliate links. If you sign up through our links, Polyloly.com may earn a commission, which helps us produce free analytics. It does not influence our analysis.