25.08.26 · Round of 32
Four Elite calls, one clean miss. All graded.
Five first-round matches from Monterrey — four Elite positions and three $4.99 leans — graded here against what actually happened on court. Includes the pick that lost outright, and the modelling bias we caught and refused to trade before a ball was hit.
Elite Filter
3 W · 1 L · +0.53uFour positions survived the full form, trend, fresh-collapse and opposition-quality review out of five matches on the card — the fifth, Osorio vs Li, was a dead-even PASS by both market and form and carried no position. All four Elite plays were set-handicap or games-handicap lines with a verified Pinnacle price edge before kickoff.
$4.99 Tier
1 W · 2 L · informationalForm-backed favourite/near-favourite leans that never cleared an independent Pinnacle price edge — disclosed as directional reads, not staked positions. No P&L impact either way, but tracked with the same honesty as the staked board.
Form is not the answer by itself. It is the first question.
These are not automatic picks.
Match by match
Combined Form Score (CFS) for both players, any fresh-collapse flag on file, and how it actually played out.
2.2-point gap — a wash by both form and market (51.3% devig). No flags either side. Every Pinnacle-verified market passed.
*Marcinko's number is a single-match sample artefact, not a real weakness — dominant on hard at Challenger level, excellent on clay/grass all season. Parks' flag blocked any lean in her favour outright.
~11-point gap plus a flag, both pointing at Tjen — the cleanest favourite case on the card. Lost anyway. The round's one genuine miss, and a reminder the flag describes a trend, not a guarantee.
*Opposition-quality discounted (some ITF-level wins in the sample) — still the largest genuine underdog form lead on the card. Hometown player at her country's own WTA 500. Class-gap kept the pick to the set line, not the moneyline — she won outright anyway.
~28-point gap, the round's largest — Vekic had explicitly lost her last four matches, two on hard, going in. No flag on either side; the cleanest, highest-conviction Elite read of the day.
What we caught before kickoff
A systematic model bias, flagged and excluded
Every one of the five matches independently threw a large, suspiciously uniform technical edge on Total Over 2.5 sets and on in-3 Correct Score lines — anywhere from +11% to +31% EV, regardless of how lopsided the match projected. That uniformity across matches of very different shape is the signature of a modelling artefact: the games engine treats each set as an independent trial, which systematically overstates how often a match goes the distance. Both market families were excluded from every board — Elite and $4.99 alike — before a ball was hit.
Worth saying plainly, in the same spirit as the rest of this page: four of the five matches actually did go three sets. Had that flagged market been traded anyway, it would have cashed four times out of five this particular round. That doesn't make the exclusion wrong — the formula behind it is still mathematically miscalibrated, and trading a known-broken edge because it happened to pay off once is how a model drifts into worse habits, not better ones. It's exactly the kind of result we report here rather than quietly skip.
What each tier is
Higher-conviction, higher-variance
The positions where the model signal and the full form review — trend, surface, fresh-collapse flags, opposition quality — line up together.
- Selected underdog +1.5-set and games-handicap positions
- Verified against Pinnacle pricing, never against the model alone
- Full risk disclosure and staking on every position, win or lose
Lower-variance, form-backed
Favourite and moneyline reads confirmed by current form, left unstaked where no independent Pinnacle edge exists.
- Form-supported favourites, disclosed without a stake
- Matches where the model's underdog lean was rejected by the review
- The same transparency, at entry-tier pricing
Read the next card before it posts.
Every match, every tier, graded openly against the result — including the ones that miss. That is the whole method.
Join on Patreon →