Pinned Tweet
1/ ๐ฏ "Polymarket or Kalshi โ which is better?"
Wrong question. โ
Reading it that way costs traders ~7ยข an event, forever. ๐ธ
The same event prices differently on the two venues. Not noise โ structure. Here's the part nobody explains ๐งตโคต๏ธ
1
3
756
1/ ๐ The Assymetrix MCP server is live.
Your AI agent can now see every prediction market at once.
An agent wired to one venue is blind to the other half of the market. This gives it Polymarket + Kalshi + Limitless โ one normalized schema, one connection. ๐งตโคต๏ธ
3
1
576
2/ ๐ Connect in one line. Works with Claude Desktop, Cursor, any MCP client:
npx -y @assymetrix/mcp-server
Set ASSYMETRIX_API_KEY, restart, done.
2
42
๐ real answer: depends on the market more than the venue:
โ Polymarket = depth. deepest resting books, best for size
โ Kalshi = flow. tight top-of-book + fast repricing, but thinner resting depth (watch slippage on size)
โ Limitless = fast-settling Base markets
the catch: top-of-book hides depth. same market can be deep on one venue and paper-thin on another โ so an agent shouldnโt marry one venue. thatโs the whole point of reading all of them normalized in one call ๐
1
19
3/ ๐ ๏ธ Three tools out of the box:
๐ search_markets โ any market, across venues
๐น get_market_prices โ current cross-venue prices
๐ get_market_history โ history for backtesting + context
All normalized to one schema โ your agent reads every venue the same way.
1
23
4/ โก Free tier, no card: 10,000 requests/mo. Sub-second latency.
More venues (HIP-4, Predict.fun) landing this quarter.
Give your agent the whole market, not one book:
data.assymetrix.com/mcp
@assymetrix_com
28
7/ ๐ณ๏ธ Which venue you should "pick" is the retail question.
๐ How to read the spread between them is the professional one.
Stop picking. Start synthesizing. ๐ฎ
1
17
8/ ๐ Full breakdown โ the complete rule-divergence matrix, all four forces, and what the spread means whether you're a quant ๐งฎ, a macro desk ๐๏ธ, or trading retail ๐ฑ:
assymetrix.com/blog/polymarkโฆ
We aggregate this across every major venue. One API. โก
24
1/ A thousand AI agents running the same foundation model are not a thousand minds. ๐
They are one prior with leverage.
And they're trading prediction markets right now.
The most important structural question in prediction markets that nobody is asking yet ๐งตโคต๏ธ
1
205
11/ ๐ฎ What gets built next: flow composition analysis.
โ Behavioral fingerprinting (machine flow has statistical signatures)
โ Cross-venue correlation detection (correlated agents hit ALL venues at once โ invisible from inside any single one)
โ Composition-adjusted signal scoring
The price alone can't say who's behind it.
1
20
12/ The wisdom of crowds was never a property of crowds.
It was a property of diversity and independence โ which humans provided for free.
AI ends the free ride.
The crowd is changing. The infrastructure that reads the crowd has to change first.
assymetrix.com/blog/ai-agentโฆ
@assymetrix_com
19



