MARKET DATA / ANALYTICS / CASE STUDY
ArbDetector.
Finding equivalent events across prediction markets, then checking the numbers after fees.
01 / The problem
Equivalent prediction-market events can have different titles and prices. A useful comparison needs both event matching and fee-aware return calculations.
02 / Architecture
An event-matching step identifies candidate pairs before the calculation checks complementary positions. A dashboard presents scan history and calculator output.
03 / Engineering decisions
Separate similarity matching from profitability checks. A similar title alone is not enough to establish an opportunity.
04 / My contribution
- Matched Kalshi and Polymarket events with Fuse.js and weighted text similarity.
- Checked both cross-market YES/NO combinations for positive returns after platform fees.
- Built timestamped scan history and a profit calculator.
05 / Outcomes & status
A dashboard prototype for comparing candidate events and exploring the calculations.
06 / Scope & limitations
Prototype only, not trading advice. Live market-feed reliability and any mock-data fallback have not been independently audited. Displayed opportunities must not be treated as executable returns.
InstaLead↗