Edge Leaks

We scan every Polymarket weather market so you don't have to. 9 forecast models, error bars calibrated on 116k observations, real edge vs CLOB ask. Same event. Two prediction markets. Two prices. We catch that too 👀 Free: edgeleaks.com

@PolyDekos The crowd overcommits to one bucket constantly. Miami today: 79¢ on 92–93°F, but the ensemble mean is 90.5 → 90–91 is likelier and it's priced at 14¢. That gap is the whole edge.

This trader made $16,427 on Polymarket by buying weather contracts at 15-60¢ and waiting for the price to reach $1. Strategy: -> Identifies markets where "Yes" (a specific narrow temperature range) is trading at a discount -because the crowd considers this scenario unlikely. -> Buys positions at $130–$1,700. -> The market is about to close. He doesn't predict the weather any more accurately than meteorologists do. He looks for places where the market has already decided that something is impossible - and turns out to be wrong. Top 3 deals: $1,517 → $5,832 (+284%) - NYC, 72-73°F $1,731 → $3,298 (+91%) - LA, 70-71°F $499 → $2,946 (+491%) - Miami, 76-77°F Most traders are looking for certainty. He’s looking for the opposite - markets where uncertainty is so high that no one wants to bet against the “no.” That’s exactly where the edge lies. It's not really trading. It's a lottery turned on its head - and he's selling tickets.

@mopozeuX @Polymarket @PolymarketTrade Clean one today: market has Miami's high at 92–93°F, 79% sure. Every model I track says 90–91. The model's bucket is sitting at 14¢. Resolves tonight — we'll see who's right.

Polymarket trader earned $25,998 He joined Polymarket in January 2026. During this time, he made 3,505 predictions. They were all in the weather markets. Here are his best deals: > $1,124.98 —> $2,226.57 > $2.00 —> $970.96 > $810.78 —> $1,509.96

Polymarket is 79% sure Miami's high hits 92–93°F today. All 9 forecast models I track say 90–91°F. That bucket trades at 14¢. $25K riding on the hotter bucket. Resolves tonight 👇

@Autonomous_Chad Weather is the most mispriced corner of Polymarket imo - the crowd anchors on the wrong bucket constantly. Been tracking model-vs-market gaps daily, they resolve in the model's favor more often than not.

The Trillion dollar usecase for Prediction Makerts is not gambling. It's Parametric Insurance. gambling is a small market ($600B), investors did not value Polymarket and Kalshi at $20B each because they hope to capture the gambling market. The real big fish they are chasing is the insurance market ($8T). They are betting of a future where event contracts replace the traditional Indemnity Insurance model In the indemnity insurance model has 3 major flaws : 1 - Does not scale An adjuster must inspect every claim, assess the damages and if it is covered or not. This incurs additional costs that will in the end be paid by the consumer. If these costs are greater than the expected profit for a given category, there simply won't be an insurance market for it. 2 - Slow If your house burned to the ground, you need the money now, not in 6 months. Good luck explaining that to the adjuster that will want every receipt for every piece of furniture in your living room 3 - Adversarial The insurance provider who has every incentive to deny your claim since the payout comes out of their pocket. Unfortunately the judge (the adjuster) also works for them, and will find every reason to deny you a payout There are already a few categories where Parametric insurance (event contracts) is already the dominant option : - Fire insurance - Extreme weather events - Political risk (for example if a bill gets passed that would ruin your business) - War insurance (try to find a traditional company that will insure your house in Donetsk or Tehran) @Polymarket and @Kalshi investors are betting that this market will keep expanding and that they will be the ones to capture it It's the real play behind Prediction Markets. Of course there is still regulatory risk, but that was also the case for Bitcoin in 2013. Risk = Profit Take some risks and find a way to invest in Prediction Markets before everyone realizes what is up

Resolved: NYC 72–73°F. The 22¢ bucket paid $1. Market's favorite (68–69°F @ 33¢) went to zero. All 9 models had 72.3°F on the board 11 hours before close. This is what model-vs-market divergence looks like.

Polymarket's favorite bucket for NYC high today: 68–69°F (33¢). All 9 forecast models we track say 72.3°F. That bucket trades at 22¢. One of us is wrong. Resolves tonight 👇

Full bucket board → edgeleaks.com, search "New York" Model's closest bucket (72–73°F) is the cheapest of the mid-range at 22¢. Δ column = gap between each bucket and the 9-model ensemble.

@0xMarchyel @Polymarket @PolymarketTrade Nice call. Underrated detail: Windy is just a viewer — under the hood it's ECMWF/GFS/ICON, and the edge usually lives where those models disagree. We run 9 of them against Polymarket temp markets daily; the market is slowest to reprice exactly when the ensemble splits.

Yesterday I predicted the thunderstorms ahead of the Mexico–England match and anticipated a rescheduling. I relied on Windy for both observed lightning and forecasts. Weather can create some interesting opportunities in sports markets, it was super fun 😂

@armouredme @Polymarket Solid breakdown. One thing copy-traders miss: these setups exist across ~51 cities daily, not just the ones ColdMath trades. Our 9-model ensemble flags 5 for tomorrow alone — biggest is Seoul 30°C+, NO @ 95.5¢ with +3.5% model edge.

How to print +$53,000 on @Polymarket without guessing the news. I reverse-engineered trader ColdMath. While retail gambles on geopolitics and crypto, this account is running a pure algorithmic barbell strategy on prediction markets (specifically weather). Here is the exact breakdown of his math. 𝟏. 𝐓𝐡𝐞 𝐆𝐫𝐢𝐧𝐝 (𝐋𝐢𝐪𝐮𝐢𝐝𝐢𝐭𝐲 𝐏𝐫𝐨𝐯𝐢𝐬𝐢𝐨𝐧) He parks massive volume ($10k-$12k per bet) into ultra-high probability outcomes (98¢-99¢). He acts as a market maker, scooping up 1-2% yields. Yes, occasionally a tail-event happens and he takes a brutal -$6k wipeout on a "safe" bet. But the sheer volume and win rate absorb it over time. 𝟐. 𝐓𝐡𝐞 𝐀𝐬𝐲𝐦𝐦𝐞𝐭𝐫𝐢𝐜 𝐒𝐧𝐢𝐩𝐞𝐬 This is where the real alpha is. He leaves microscopic limit bids at 0.1¢ across dozens of illiquid markets. When a desperate trader market-sells or an order book gaps, his bot catches the fill. The result? He risks $10.83 to win $5,391 (+49,697% ROI). He risks $10.00 to win $5,000 (+49,900% ROI). He balances "picking up pennies in front of a steamroller" with buying mathematically free lottery tickets. Stop trying to outsmart the news. Start exploiting order book inefficiencies. Math > Miracles.

Live scanner (free, no signup) → edgeleaks.com The 5 setups from the screenshot are under the "Tail ops" filter, Jul 7 tab. Updates 3x a day.

ColdMath printed $135K farming 90-99¢ NO on Polymarket temp markets. Everyone's copy-trading the wallet blindly. Our scanner screens all 49 cities for the same tail setups — before they hit 99¢. For tomorrow alone it flags 5, up to +3.5% edge 👇