PSG vs Troyes

The model lands at 9.6 corners, exactly the Ligue 1 baseline. That alignment is itself the signal: the team-level inputs aren't pulling the prediction in either direction, and medium confidence follows directly from that absence of deviation. When the model can't separate a fixture from the population average, the case for high conviction isn't there.

Both sides carry a match-total average of 7.8 across their recent samples. PSG's last four home totals read 6, 11, 8, 6, and Troyes' last five away fixtures show 6, 6, 13, 7, 7. The dispersion across both sequences is notable, and Troyes' 13-corner result sits well above the rest of their profile, inflating the average without reflecting a consistent tendency. High variance in the underlying data is one of the cleaner explanations for why the model holds at baseline rather than committing to a directional read.

The PSG sample is also only four games deep, which limits confidence in the home-side input. Thin samples and volatile away data together produce a prediction that's firmly anchored to the league mean of 9.6 with no structural reason to move off it. That's not a deficiency in the model; it's the model working correctly.

Position sizing at medium confidence should reflect that limited edge over baseline. Treating this tier identically to high-confidence outputs is where accounts bleed across a season.

**Ligue 1 medium-confidence record: 45.6% from 68 settled picks.**