R. Tokuda vs S. Rozin — prediction

The core of this model's lean is the Elo differential: Tokuda sits at 1659 against Rozin's 1552, a 107-point gap that in Challenger-level rating systems typically corresponds to a solid favorite. This translates into the model's 65% win probability for Tokuda, a meaningful edge but not an overwhelming one — Elo systems at this tier are built from thinner data and wider variance than tour-level ratings.

This is a rating-based edge, not a stylistic one: no surface, serve, or return numbers are available to explain the mechanism further. The gap alone suggests Tokuda has outperformed opponents of similar caliber more consistently than Rozin has, but it says nothing about how that plays out shot-for-shot in Granby.

Recent form adds a second, independent layer favoring Tokuda. His last 10 matches read WWLWLLLLWW — five wins — but he is currently on a 2-match win streak, meaning he arrives playing his best tennis of the stretch. Rozin's last 10 (WWLLWWWWLL) actually shows six wins, more than Tokuda's, but he is now on a 2-match losing streak, the opposite trajectory heading into this match.

Momentum near the point of competition often matters more than the full 10-match average, since it reflects current game feel and confidence rather than a stale sample. That recency favors Tokuda here.

Tokuda has had 24 days since his last match and zero matches in the past two weeks, while Rozin has played twice in the last 14 days with only 4 days of recovery before this one. All else equal, extra rest tends to mean fresher legs and a lower cumulative fatigue load, which can matter over the course of a Challenger match.

The flip side — that 24 days off can sometimes mean rust rather than freshness — is a reasonable general consideration, but nothing in the data confirms that effect for Tokuda specifically, so it is not factored in beyond a neutral caveat.

The market prices this match at even odds (2.02, implying a 50% win probability for Tokuda), while the Elo model assigns him 65%, producing a stated +31.1% expected value. That gap is real on paper, but it comes from a soft, less-liquid Challenger/ITF market where Elo-based edges are unproven in practice — this is a different animal from tour-level markets where large books have priced in far more information.

Being the favorite here is not the same as holding a proven edge. The model's lean toward Tokuda is grounded in a real 107-point rating gap plus favorable form and rest signals, which together build a coherent case — but the size of the implied value should be treated cautiously rather than as a guaranteed opportunity.