← Leo Labs

38,637 Markets, 11 Categories, 164 Subcategories — I Built a Taxonomy Map of Polymarket

Data as of 2026-03-29 · Source: Polymarket site navigation + Gamma API

---

I've been trading live on Polymarket for a few months, mostly staying in crypto price predictions, weather, and Elon Musk-related markets. That's only 14% of the platform.

What's the other 86%? Box office, music charts, GDP forecasts, esports, economic indicators. Entire categories I never knew existed. So I pulled all the markets and built a classification map.

What does the map look like?

Starting from Polymarket's site navigation and breaking it down by category gives 11 major categories and 164 subcategories:

Note: These 7,880 are navigation-visible markets. PM has countless more unclassified or silent markets. The API total of 38,637 is much larger than what the navigation shows.

What do profitable traders in each category look like?

Numbers alone don't tell the story—you need to follow the money. I extracted on-chain trades, classified addresses by their primary market types, then looked at PnL levels for the top 10 profitable addresses in each category:

Methodology: Extracted all profitable addresses (PnL > 0) from Polymarket on-chain trades. Tagged each address by primary category across 34,564 markets. Ranked by PnL, took top 10 per category. Data through 2026-03-20. "Other" = addresses spanning multiple categories, unclassifiable.

Honestly, before running this I thought politics would win—the election cycle action was crazy. It wasn't.

Crypto crushes everything. Top 1 made $5.73M, nearly 3x second-place sports. But crypto has 22,000 profitable addresses. Most competitive field too.

Sports is interesting. Only 5% of navigation-visible markets, yet Top 1 hits nearly $2M and Top 10 threshold is $145K. Small pool, big money at the top.

Politics has lots of participants but terrible top-tier profitability. 5,528 profitable addresses, Top 10 threshold only $16K—1/14 of crypto. Tons of addresses playing, but few making serious money.

Biggest surprise: science. Only 53 profitable addresses total, Top 1 under $800. Almost no competition, but almost zero liquidity either.

Official classification doesn't cut it for strategy

Navigation categories work fine for browsing. But for strategy selection they're too coarse—Sports mixes NHL season championships with CS2 single-round kill counts. Completely different data sources, settlement frequencies, pricing logic.

For me, using official categories to pick a strategy direction is like using "food" as a label to decide what to eat today. So I did a second pass on top of the official taxonomy.

Three iterations, each one flipped the last

Round 1: Direct official categories
Sports / Politics / Crypto / Culture / Economics...
Problem: Too much variance within categories. Doesn't guide strategy selection.

Round 2: Regroup by data source
Binance API / Weather APIs / ESPN feeds / News events / On-chain data...
Better—same data source = reusable strategy. But blind spots: tons of categories I didn't even know had APIs.

Round 3: Layer by data accessibility
Open APIs (direct calls) / Semi-structured (scraping required) / Black box (inside info needed)
This round actually worked. Answers one question: Which categories can I start now?

The standard for deciding if a category is quantifiable

Box office. Music charts. GDP. Seems random. But once you classify, you realize if something meets three criteria, your weather strategy plugs right in:

1. Structured data source (API or database, not headlines)
2. Data frequency ≥ daily (supports daily market settlement verification)
3. ≥6 months history (enough for backtesting)

Running this filter, several categories have solid data but I never noticed:

Box office and music charts are interesting—public data, clear settlement logic, but orderbook depth shows liquidity way below crypto or politics.

What I'm trying next

Based on this map, here's my research roadmap:

1. Box office market pricing (Box Office Mojo data ready to go)
2. Spotify chart prediction
3. GDP Nowcast (Atlanta Fed real-time baseline)
4. Esports event pricing (LoL/CS2, already working on R8)
5. Weather market making (taker → maker transition)
6. Economic indicator markets (CPI/unemployment/retail sales)
7. Entertainment markets (Oscars/Grammys)

Logic: data accessibility × liquidity × my current capabilities. Box office ranks first because the data is cleanest and settlement most transparent.

---