Core Portfolio
Tier A / B · structural line
Grade A
The only lane that receives full exposure. Reserved for fixtures where price, line and structural edge align simultaneously.
Hit rate—
ROI—
Sample—
SaiX is a research desk for Europe's Big Five leagues. Every fixture is graded by tiered conviction gates, then routed by a cascading diversion engine into the single channel where its structure actually holds value — and where it doesn't, it is published as a rejection, with the reason attached.
“A model that recommends every match recommends nothing. We would rather publish eighty-five rejections with reasons than eighty-five guesses with confidence.”
Most prediction products optimise for output volume. SaiX optimises for explainability under uncertainty: a stated probability, a stated price, a stated reason, and a stated reason to walk away. Four disciplines hold the whole thing together.
Elo differentials, rolling xG, rotation load and squad availability are fused into one structural view per fixture — no single-factor signals.
Every match runs a Monte Carlo pass plus a closed-form Poisson cross-check. Scorelines are ranked by joint probability, not vibes.
Selections are graded A / B / C with a hard exposure ceiling per tier. Grade A is never diluted to make the board look busier.
When the pricing gate blocks a fixture, the block code and the historical bleed behind it are shown. Rejections are a product feature.
A pure probability model only cares about maximum expected value. That leaves roughly 85% of fixtures as "no view" — mathematically clean, commercially useless.
Our cascading engine keeps the top gate untouched, then re-routes the remainder by tactical signature. One fixture, one active channel, all other channels logged with their bypass reason.
The only lane that receives full exposure. Reserved for fixtures where price, line and structural edge align simultaneously.
Slow-tempo leagues compress toward parity on flat lines. Priced inside a narrow consensus corridor, this is a structural — not a hunch.
Heavy favourites priced into thin value are the recurring source of late stalls. The radar flips the exposure instead of following the crowd.
For fixtures where the side is live but the draw is fully priced in. The highest hit rate on the desk, and the lowest variance.
Activated only when the summed simulation goal expectation clears the threshold and the price sits inside the accepted band.
Every fixture on the board carries three ranked scorelines. This is what guarantees 100% coverage without inventing a view.
Reporting window —. Settled fixtures are published in full — teams, line, consensus price, closing score and outcome. Unsettled fixtures are sealed: no teams, no selection, no price. Nothing on this page requires an account, and nothing locked is inferable.
| League | Fixture | Channel | Selection | Price | Close | Result |
|---|---|---|---|---|---|---|
| Loading the settlement sheet… | ||||||
The core pool is tested season by season, not aggregated into one flattering number.
Two diagnostic payloads are returned for every fixture. One explains why the desk took the position. One explains why it refused. Both cards below are the most recent settled examples, pulled live — including when the read was wrong.
The weekly settlement sheet, the routing audit and the three-season performance ledger are open to anyone. Sealed fixtures stay sealed until settlement — that is the boundary, and it does not move.