Tottenham Hotspur vs Leeds United

The model's predicted total of 10.55 sits barely above the Premier League baseline of 10.5. Barely. That near-identical read is the first number to sit with: when the model's output and the league norm converge this tightly, the structural edge is thin, and the medium-confidence tier on this fixture is the correct classification, not a soft hedge against an outcome the model can't see.

Tottenham's home corners average across their recent run is 11.4, which nudges the combined picture above baseline. Their last five home totals read 16, 12, 16, 9, and 7. That range, from 7 to 16, produces the 11.4 average but also shows how readily a single disengaged performance collapses the total. Leeds carry an away average of 9.2, and their last five away totals, 16, 13, 15, 11, and 9, carry a similar character: three results well above their own average bracketed by two that pull it down.

Those two distributions, one elevated, one subdued, sit either side of the 10.5 league baseline and together yield a predicted total of 10.55 that's barely distinguishable from it. Medium confidence is precise here. It reflects a model with enough signal to produce a directional output but insufficient separation from the league baseline to treat that output as settled, which means stake discipline is not a footnote but the position itself.

Spectral's medium-confidence corners record in the Premier League: 97 settled, 45.4% win rate.