C. Wong vs A. Shepp — prediction

Wong's Elo rating (1792) sits 102 points above Shepp's (1690), the kind of gap that typically points to a clear favorite. His ATP ranking of 109 (with a positive trend of 15 spots) reinforces that he is the higher-level player on paper.

Yet the baseline model caps Wong's expected win rate at just 45%, well below what the Elo gap alone would suggest. That gap between Elo-implied dominance and the more conservative baseline figure is worth noting: the model is not fully convinced this is a lopsided match.

The recent form lines tell a story that cuts against the Elo edge. Shepp has won 8 of his last 10 matches, including a streak of seven straight wins, and enters on a current 1-match winning streak just like Wong.

Wong, meanwhile, has managed only 4 wins in his last 10, with a choppier pattern of alternating losses. This recent-form imbalance is one of the clearest signals in the data favoring Shepp's competitiveness in this matchup, regardless of the rating gap.

The data shows Wong winning 65% of service points and 32% of return points, numbers that describe a player who leans on his serve to control points. Without comparable serve or return figures for Shepp, this can only be read as a standalone marker of Wong's game rather than a direct comparison.

On a match where form and Elo tell different stories, Wong's serve numbers are a modest positive marker for his side, but they do not resolve the tension between the recent-form and rating signals.

At odds of 1.20, the market prices Wong's win probability at roughly 83%. The model's own estimate is 64%, translating into a -22.9% expected value on backing him at this price. That is a significant gap, and it means the market is considerably more confident in Wong than the model itself is.

Since this is an Elo-based Challenger estimate, the market here should be treated as a soft, less battle-tested one — the edge is unproven either way. Combined with Shepp's stronger recent form, this is not a spot to treat Wong's favorite status as automatically translating into value; the honest read is that backing him at 1.20 is a negative-EV proposition by the model's own numbers.