R. Tokuda vs J. Sheehy — prediction

The 105-point Elo gap (1643 for Tokuda vs 1538 for Sheehy) is the clearest structural signal in this match, translating into a 65% model probability for the favorite. That figure sits almost exactly where the market has it (66% implied), meaning the rating gap is already priced in rather than offering a hidden edge.

This is a Challenger-level Elo estimate, which is inherently a softer, less-scrutinized market than the ATP tour. The gap is real but should be treated as a rough guide rather than a precise measure of superiority.

Sheehy actually holds the stronger raw serve number here, winning 61% of service points compared to Tokuda's 57% — a 4-point edge that suggests his service games should be the more reliable weapon on paper. Return numbers, by contrast, are close to a wash: Tokuda wins 36% of return points against Sheehy's 35%, so neither player projects as a clear returning threat against the other's serve.

Because the return metrics nearly cancel out, the serve gap becomes the deciding thread in this specific dimension, tilting the pure serve/return picture toward Sheehy even though he is the lower-rated player overall.

Tokuda's last 10 matches (4-6, WLLLLWWLWL) show more variance but also more recent wins, including two victories in his last four outings before a 1-match dip. Sheehy's form is starker: 2-8 over his last 10 (LLLLWLLWLL) with a live 2-match losing streak, a tougher recent trend to build momentum from.

Neither player shows a quality win in the data, so this comparison is about baseline trend rather than marquee results. Still, the form gap adds a modest tailwind to Tokuda beyond what the Elo number alone captures.

At odds of 1.51, the market implies a 66% chance for Tokuda, just one point above the model's own 65% estimate. The expected value comes out slightly negative at -2.4%, meaning the price does not compensate for the model's own uncertainty about this outcome.

Being the favorite here does not equate to a betting edge. With Elo-based Challenger models, the market has historically kept pace with or exceeded pure rating estimates, so this should be read as a fair-to-slightly-unfavorable price rather than an opportunity.