Insightix - The same event, priced differently across markets
byā¢
Insightix tracks the live probability of the same event across prediction markets and derivatives/sports data ā and alerts you the moment they disagree. Crypto, sports & esports. De-vigged, normalized, apples-to-apples. Information only, not advice.
Replies
The discordance alerts on esports matchups caught me off guard ā saw two platforms pricing the same game with a 12% gap that closed within minutes. Clean, no-noise feed and the de-vigged math actually lines up.
@bidavkubra60771Ā Love hearing this ā those fast-closing esports gaps are some of the sharpest, because two platforms disagree and one just hasn't repriced yet. Glad the de-vigged math held up for you. If there's a line or vertical you'd want us to cover next, tell us ā we're adding them
@bidavkubra60771Ā This is exactly what you asked for ā so we built it, and it's live
Two things:
ā¢ The aggregate: for every alert we've sent, how often the gap actually closed. The wider the gap, the more reliably ā 10pp+ divergences narrowed 91% of the time (median 34% of the gap closed). insightix.io/how-it-works#backtest
ā¢ The replay you described: real past alerts with the gap trajectory over the days that followed. insightix.io/proof
Cases are picked by the gap size at the alert (not by outcome), so it's not cherry-picked ā a 40% case sits right next to the 96% ones. Convergence only, not a profit claim. Would love your read
Really useful concept since mispricings across markets are where the edge lives. One thing that would make it even better is a backtested alert feed showing each divergence's eventual resolution, so I can see how often a given spread actually meant anything before acting on it.
@birsensfl7Ā 100% ā knowing how often a spread actually resolved is the line between signal and noise. Good news: we already log every divergence and its eventual resolution internally (it's how we validate the engine), so a per-signal "resolution history / hit-rate" is very buildable and near the top of our roadmap. Really appreciate you framing it so clearly
@birsensfl7Ā 100% ā and it's live now.
We backtested every alert: 5ā10pp gaps narrowed 72% of the time, 10pp+ gaps 91% (median 34% of the gap closed). insightix.io/how-it-works#backtest
And per-signal, with the actual gap trajectory: insightix.io/proof ā picked by how wide the gap was, not how it turned out, so the honest mix shows (there's a 40% right next to the 95s). Thanks for pushing us on this one
Update from the makers š ļø
A few of you asked the same great question: "does a flagged divergence actually mean anything?" So we shipped the answer:
ā¢ Backtest ā for every alert, how often the gap closed (10pp+ gaps: 91%). insightix.io/how-it-works#backtest
ā¢ Track record ā real cases with the gap trajectory. insightix.io/proof
Convergence only (not profit claims), selected by gap size not outcome. š 50% off with code Y5MTG3MA (card) through Jul 31. Keep the feedback coming