A week ago I deposited $17 on @Polymarket
Didn't trade elections. Not crypto. Not geopolitics
Found a strategy on weather markets - and just let it run.
Checked my account 1-2 times a day. Nothing else
Today's balance: ~$33-34.
$17 → $34 in one week. Deposit doubled
Weather markets are the least competitive niche on Polymarket. Almost nobody looks there
Account in comments
Day 1 with weatherscan_bot
First signal: Tokyo, YES, 45% odds → $30.00 win on $13.50
I didn't take it. Decided not to rush on day one
$16.50 left on the table. Today
Day one
https://t.co/zXdgaiaAcd
Kim just revealed North Korea's third uranium enrichment site
Weapons-grade production more than doubled in 5 years
New factory opened June 3. "Exponential" expansion ordered
Now look at Polymarket: 3+ missile tests at 46%, zero tests at 44%
The market is basically saying: we have no idea
But here's my read: Kim doesn't reveal a new nuclear factory without demonstrating it
3+ tests is where I'd put my money
Institutions just used Polymarket to hedge Nvidia H100 GPU costs
Not elections. Not politics. AI compute infrastructure
A six-figure block trade between FalconX and AneraLabs settled on-chain via Polygon
GPU rental costs are now a tradeable asset class on a prediction market
That's not the future. That's today
We're excited to announce Polymarket has facilitated the first ever on-chain, institutional block trade in the prediction market space.
Institutions using Polymarket to hedge GPU compute exposure at scale give a glimpse into both the future & the promise of prediction markets.
The best way to know what a tool needs - build it yourself
I did. One month, real bugs, real losses
Now I'm using "weatherscan_bot" Because I know exactly what it solves
https://t.co/zXdgaiaAcd
My bot was consistently 1-2°C off across all cities
Shenzhen, Seoul, Bangkok, Miami - same pattern everywhere
Finally figured it out: Polymarket uses specific weather stations, not city coordinates. But my bot was fetching forecasts for city centers
Bao'an Airport vs Shenzhen center = 25 km = different microclimate
Same problem in every market I was trading
Now the bot automatically reads the market description, finds the station code, pulls the exact forecast
Lesson: One small wrong assumption compounded across all positions
This is what systematic errors look like
Found a critical bug in my Polymarket weather bot while reviewing signals
The bot flagged a Seoul temperature market at 1.6% as a "great opportunity" with 75%+ edge.
But here's what actually happened:
> Market was at ~80% probability all day
> At 12:00 PM the temperature was already measured
> Market crashed to 1.6% because the outcome was known
> Bot saw 1.6% and thought it found a mispriced bet
In reality the market was perfectly priced - it just already knew the answer.
Fix: bot now only analyzes markets for tomorrow and beyond. No more trading on already-measured temperatures.
Always review your signals manually before placing bets
I predicted 10-20 ships through Hormuz by May 31
Market resolved 0-10 at 97.4%
I was wrong. Here's my actual miscalculation:
> Knew Project Freedom got paused May 4
> But thought US would find another way to guide shipments through
> Thought even "dark sailing" (AIS off) would add up to 10-20 vessels
> Actual reality: 4 vessels on May 24. Blockade stayed locked
I underestimated Iran's control and overestimated US commitment to reopening
The edge I thought I had was based on incomplete geopolitical reading
Average ships transiting Hormuz by end of May?
Market says 0-10 at 64%
But look closer:
0-10: 64% ↑12%
10-20: 22% ↓32%
20-40: 6% ↓47%
40-60: 4% ↓39%
60+: 3%
The market is pricing complete blockade
Yet we know:
> Normal pre-crisis Hormuz traffic was 138 ships daily (United Against Nuclear Iran)
> May 17 saw only 2 vessels - 2% of normal (Straits)
> Negotiations in Doha closing on a deal
> Blockade is weakening
Even getting to 10-20 ships by May 31 means partial normalization - which the latest headlines suggest
My read: 10-20 ships at 22¢ is mispriced
The spread tells you the market doesn't believe the deal is happening. But the deal is already in motion
The real story: Trump's "weaponization" fund was a backdoor way to pay Jan 6 rioters with taxpayer money
Once that became clear, even allies abandoned ship
You can get away with a lot in politics. But not that.
Market saw it coming. Politicians didn't