.svg)
.svg)
.svg)
The Polymarket Weather Leaderboard: Who's Winning and How
Polymarket publishes a public leaderboard sorted by profit, filtered by category. The weather leaderboard is one of the few places in prediction markets where you can see real on-chain profit figures — not simulated returns, not vendor claims, but actual USDC settled from real trades. The numbers are real. The question is: what do they tell us about strategy?
The Leaderboard: What the Data Shows
As of early 2026, the Polymarket weather leaderboard shows cumulative profit from weather-category trades across all time. The top positions reflect hundreds of thousands of dollars in net realized P&L.
| Handle | Approx. Net Profit (Weather) |
|---|---|
| gopfan2 | $343,000+ |
| aenews2 | $277,000+ |
| ColdMath | $120,000+ |
| gopfan | $118,000+ |
| bama124 | $87,000+ |
| Hans323 | $81,000+ |
| Handsanitizer23 | $71,000+ |
| automatedAItradingbot | $65,000+ |
| WeatherTraderBot | $57,000+ |
Below these, dozens more traders sit in the $10,000–$50,000 range. The total weather market profit extracted by the top 20 wallets runs into the millions of dollars — all of which is counterpart loss paid by the less-skilled traders on the other side. These are verifiable on-chain figures. What's not public is the exact strategy behind each wallet — which requires inference.
ColdMath: The Most-Analyzed Weather Wallet
ColdMath is the most-discussed weather trader on Polymarket, partly because of the magnitude of the returns and partly because an X user named @armouredme published a detailed reverse-engineering thread analyzing the wallet's on-chain trade history.
What the on-chain data showed:
- Concentrated activity in Buenos Aires, Cape Town, Dallas, and Atlanta — cities with relatively lower bot saturation and wider spreads at the time.
- A barbell structure: many small positions on underpriced tails (5–15¢ buckets) combined with occasional high-conviction central-bucket trades.
- Very weak performance in non-weather categories (politics, crypto, Venezuela markets all showed negative P&L).
- Consistent sizing discipline — bets rarely exceeded $100–$150 on a single bucket, despite significant total capital.
The pattern is consistent with an algorithmic strategy: the wallet was running a systematic program that scanned for buckets where model probability exceeded market price by a defined threshold and sized bets uniformly.
gopfan2: Simple Rules, Extreme Consistency
The gopfan2 wallet has attracted attention for what appears to be an almost absurdly simple stated strategy:
- Buy YES if the price is below $0.15.
- Buy NO if the price is above $0.45.
- Risk approximately $1 per position.
- Repeat thousands of times.
This is essentially a tail-buying strategy. At sub-$0.15, buckets are long shots the market has priced at less than 15% probability. The bet is that a meaningful fraction of these long-shot buckets are underpriced — that the true probability is 18%, say, when the market says 12%.
The profitability relies on the favorite-longshot bias in reverse: in temperature markets, where retail traders concentrate on the modal bucket and neglect the tails, cheap buckets can be underpriced rather than overpriced.
The key takeaway from the gopfan2 approach:Simple systematic rules applied consistently across thousands of trades can generate substantial alpha even without a sophisticated forecast model. The edge isn't the accuracy of any individual bet — it's the consistent identification of mispriced tails at scale.
Hans323: The Latency-Arbitrage Story
Hans323 was the subject of an Insurance Journal profile in April 2026 — reportedly a 23-year-old German law student who ranked among the top six all-time earners on Polymarket weather markets with approximately $81,000 in net profit.
According to that profile, Hans323's primary strategy was latency arbitrage around model release windows: entering positions in the minutes immediately after a new ECMWF or GFS run was published but before the market had repriced.
This requires:
- Knowing exactly when each model run becomes available (ECMWF 12 UTC primary run is available approximately 6 hours after cycle start — around 18:00–18:30 UTC).
- Parsing the new model output automatically (or very quickly manually) to identify which city/date markets have shifted materially.
- Placing orders faster than competing participants.
Hans323's approach (per the profile) was more systematic than truly automated — semi-manual monitoring of key model releases, with pre-staged orders ready to execute. The London and Paris markets reportedly showed particularly strong correlation between ECMWF model shifts and price moves.
Three Distinct Strategies Compared
Looking at these three wallets side by side reveals that there is no single “right” way to profit on Polymarket weather markets.
| Trader | Strategy Type | Cities | Avg. Bet Size | Edge Source |
|---|---|---|---|---|
| gopfan2 | Tail-buying, high volume | Multiple | ~$1 | Longshot mispricing |
| ColdMath | Barbell, algorithmic | Secondary markets | ~$50–$150 | Model vs. market gap |
| Hans323 | Latency arbitrage | London, Paris | Variable | Model update speed |
Jua: The Proprietary Model Angle
Perhaps the most interesting actor in the Polymarket weather ecosystem is Jua, a Swiss weather-AI startup whose CEO has confirmed they operate a trading vehicle that bets on max-temperature Polymarket contracts using Jua's proprietary AI weather forecast.
This represents a fundamentally different information advantage: not better probability estimation from public models, but a genuinely better forecast. The CEO noted that liquidity remains “too low for a well-sized fund” — which suggests that even with institutionally superior forecasts, Polymarket weather markets have a practical capacity limit of perhaps $500K–$2M in annual profit before liquidity constraints bind.
What Distinguishes the Winners From the Losers
Looking across the documented strategies, a consistent set of characteristics emerges:
- Systematic, not intuitive. Every documented top earner is operating some form of rule-based or model-driven approach. None appear to be trading on intuition.
- High volume. Weather trading is a law-of-large-numbers game. The edge only shows up reliably at hundreds or thousands of trials. gopfan2 has executed well over 10,000 individual positions.
- City-specific focus. Rather than trying to trade every city equally, top earners specialize: ColdMath in secondary markets, Hans323 in London/Paris. Specialization allows for deeper calibration.
- Strict sizing discipline. None of the top wallets show evidence of “going big” on a single trade. The winning approach is a large number of appropriately-sized bets with a consistent edge.
- Airport-level precision. Top traders use the resolution station's data directly — not city-center weather apps.
Copy-Trading: Following the Leaderboard Passively
For traders who don't want to build their own models, several tools have emerged to follow top weather wallets — including Stand.Trade (in partnership with Polymarket's official “COPYCAT” feature) and PolyCop, a Telegram-native bot that tracks and mirrors selected wallets in real time.
The limitation of copy-trading is that it's inherently backward-looking. By the time a ColdMath transaction appears on-chain and gets copied, the market has often already moved. The copy-trade fills at a worse price, capturing less (or none) of the original edge.
The Leaderboard Is a Floor, Not a Ceiling
The public Polymarket leaderboard shows only activity accumulated on-chain under specific wallet addresses. It almost certainly understates the full scope of weather trading, because sophisticated operators may use multiple wallets, and some of the best-performing traders may not appear under a recognizable handle.
The leaderboard is best read as evidence that systematic, model-driven weather trading is producing real, sustained profits at meaningful scale — and as a public catalogue of strategies worth studying for anyone building their own approach.