K. Quevedo vs G. Ruse — prediction

The most concrete number in this match is the Elo differential: Ruse sits at 1736 against Quevedo's 1568, a 168-point gap that historically translates into a meaningful edge in surface-neutral matchups. This is reinforced by the ranking spread — Quevedo at #126 versus Ruse's #69 — meaning the player the model tags as favorite is, on paper, the lower-ranked and lower-Elo competitor.

This is an unusual setup: the model's 71% probability for Quevedo runs directly against the two hardest, most objective signals available (Elo and ranking). That doesn't make the model wrong, but it does mean the case for Quevedo has to rest on the softer factors — serve/return shape, recent form, and rest — rather than on pedigree.

Ruse's 63% serve-points-won is a real advantage over Quevedo's 56%, a 7-point gap that would typically let her control more service games. Quevedo's return game is better (46% vs 42%), but a 4-point return edge is smaller than Ruse's 7-point serve edge, so on raw serve/return math Ruse's service dominance is not fully offset by Quevedo's return skill.

Both players enter with identical 7-3 records over their last 10 matches, so recent win-loss form is essentially a wash. The one asymmetry is quality: Ruse's win over L. Noskova (Elo 1943) is a notable result that Quevedo's form log does not match, suggesting Ruse has shown she can beat significantly higher-rated opposition recently.

Rest tilts slightly toward Ruse as well — she played 13 days ago and has a match in the last 14 days, while Quevedo has been idle for 19 days with no recent match play, which can mean either freshness or rust depending on how it's read. Quevedo's one positive is a 14-spot rise in ranking trend, a modest sign of upward momentum that cuts against the otherwise unfavorable picture.

The gap between the model's 71% and the market's implied 32% is large, producing a stated EV of 124%. That is a striking number, but it needs to be read against the fact that Ruse leads clearly on Elo, ranking, serve points won, and has the more notable recent win — the harder data points in this file. A soft, WTA-tuned factor model with ~64% out-of-sample accuracy can find real signal, but this particular case has the model and the market disagreeing sharply while the raw performance numbers lean toward the opponent.

Being tagged the model's favorite is not the same as being the more probable winner by the data shown here. Given the conflicting signals — favorable model output against unfavorable Elo, ranking, and serve numbers — this is a case for caution rather than confidence in the price, and any decision should treat the edge as unproven rather than assured.