The only dedicated prediction-market intelligence company

See the logic connecting related markets.

GraphMarkets is built exclusively to sell prediction-market intelligence. Across Kalshi, Polymarket, and IBKR ForecastEx, we map verified relationships — equivalence, implication, mutual exclusion, multi-leg structures, and more — then provide the context and price condition needed to evaluate a potential opportunity.

Millions of past relationships discovered 3 connected platforms 2+ legs per relationship 8 relationship types
Continuously updated — 1K new markets ingested and 10K+ relationships discovered daily
800K
Intra-event relationships
20K
Inter-event relationships
10K
Cross-platform relationships
3M+
Past relationships discovered
What you do with it

Three ways to put it to work

The relationships are the raw edge. How you turn them into returns is up to you — here's how traders use the feed.

🔒

Identify a structural spread

When a verified relationship's price condition is executable, its legs can create a structurally hedged position. Assess available size, fees, settlement rules, and fill risk before acting.

⚖️

Hedge across platforms

Holding a position on one platform? The graph shows you the equivalent or implied contract elsewhere, so you can offload risk on the side with better depth or pricing.

📈

Run a market-neutral book

Use independent relationships to build a more market-neutral research or trading book, while retaining control of sizing, execution, platform exposure, and settlement risk.

What you get

Eight kinds of relationships — each with an evaluable price condition

Each relationship identifies two or more markets, the platform on every leg, its logical structure, and the price condition to evaluate. We break down the conditions under which each market resolves, attach the verification sources used, and surface nuanced differences in wording, timeframe, geography, thresholds, resolution rules, and more.

Intra-event
Within one event

Brackets, thresholds and outcomes of the same event that must add up — the mechanical relationships.

Inter-event
Across different events

One event's outcome logically constrains another's — connections no one thinks to look for.

Cross-platform
Across platforms

The same or related outcome priced across Kalshi, Polymarket, and IBKR ForecastEx.

One connected graph

Relationships within and across platforms

We map intra-platform relationships as well as every cross-platform pairing, including multi-leg structures that span the graph. We also capture conditional dependencies: for example, a market on a government shutdown lasting 30+ days depends on a shutdown occurring, while a market on a company completing an IPO depends on that company not being acquired first.

Virtually 100% precision on published relationships
Kalshi Polymarket IBKR ForecastEx

Cross-platform equivalence

The same real-world outcome listed on both Kalshi and Polymarket. Buy YES on the cheaper side, NO on the other.

ExampleThe same referendum outcome listed with matching resolution rules on two platforms.
yes_A + no_B < 100

Implication (A ⇒ B)

Two distinct questions where A being true forces B true. Buy A's NO and B's YES.

Example“Dems win House and Senate” ⇒ “Dems win the House.”
a_no + b_yes < 100

Mutual exclusion (A XOR B)

Two outcomes that can't both happen — rival candidates in a two-way race. Buy NO on both.

Example“Mamdani wins NYC mayor” vs “Cuomo wins NYC mayor” — at most one.
a_no + b_no < 100
¬

Complement (A = NOT B)

One market is the logical negation of another. Compare both YES legs or both NO legs, accounting for the exact resolution rules.

Example“Candidate A wins the election” = NOT “Candidate A does not win the election.”
yes_A + yes_B < 100

Exhaustive partition

A set of brackets that covers every outcome exactly once. The YES legs must sum to a dollar.

ExampleGDP brackets [<2%, 2–3%, 3%+] — exactly one resolves YES.
Σ yes_i < 100

Composite & threshold

A threshold contract priced against a basket of brackets across platforms — synthetic replication.

Example“Inflation above 3%” = sum of every “3%+” bracket on the other platform.
Σ brackets vs threshold

Threshold/Time monotonicity

The same metric at two cutoffs — clearing the higher bar guarantees the lower. Ladder the chain.

Example“GDP above 3%” ⇒ “GDP above 2%” — same number, ordered thresholds.
a_no + b_yes < 100

Aggregation

A specific case rolls up into a broader category — one outcome inside a larger set.

Example“Fed cuts 25bps” ⇒ “Fed cuts rates this meeting” (any amount).
a_no + b_yes < 100
Verification context, not a black box

Full context for every relationship

We continuously discover candidates and validate their event scope, timing, and resolution context before delivery. Your feed makes the reviewable evidence visible, so you can independently assess each relationship before trading.

Ingest

We normalize event and market data from Kalshi, Polymarket, and IBKR ForecastEx into a common relationship model.

Match

Candidate relationships are screened for mismatched geography, timeframe, event scope, and contract type.

Verify

We compare the market language, resolution timing, and available rules; identify material differences between otherwise related markets; and attach the verification sources and context to the record.

Deliver

Delivered records include every leg, relationship type, resolution conditions, verification sources, material market differences, and the observed prices used to calculate the current condition—so you can independently assess liquidity, fees, and whether it is worth acting on.

Questions

Frequently asked

What exactly do I get?

A continuously updated feed of relationships across Kalshi, Polymarket, and IBKR ForecastEx. Each record includes the relationship logic, platform and leg details, resolution conditions, verification sources, nuanced market differences, and observed prices for your own evaluation. You decide whether and how to trade.

Do you place trades for me?

No. GraphMarkets is a data and intelligence product, not a broker or fund. We surface the relationships and the conditions; execution, capital, and risk stay entirely with you.

How do you avoid false matches?

Every candidate is checked against its geography, timeframe, event scope, resolution date, and available resolution rules. Records without sufficient support are held back for additional review rather than published.

How fresh is the data?

Discovery and monitoring run continuously. Records carry an observed time and status, and price conditions should always be rechecked against available executable liquidity before trading.

How do I access the feed?

Through a web dashboard with CSV/JSON export, plus a REST API and real-time push for programmatic consumers.

Which platforms and categories do you cover?

We cover Kalshi, Polymarket, and IBKR ForecastEx across their available categories, including Elections, Economics, Crypto, Sports, Entertainment, and Science & Tech. Coverage depth varies by platform and market availability.

Does an arbitrage condition guarantee I'll make money?

No. A flagged condition means the displayed prices satisfy the inequality at the moment of observation. Real profit depends on fillable size, fees, slippage, the time until both legs resolve, and capital being locked until then. You should verify every leg before trading.

Are there geographic restrictions?

The data feed itself has no restriction. However, Polymarket blocks API and trading access from some regions (including US IPs), and Kalshi availability varies by jurisdiction. Whether you can act on a given relationship depends on your own location and the platforms' terms — that's on you to confirm.

Can I see a sample first?

Yes — leave your email below and we'll send a sample slice of the relationship feed so you can judge quality for yourself.

Want a taste first?

Leave your email and we'll send an annotated sample relationship record — no commitment. Ready for the full feed? Request access →