For each match I pull the bookmakers' match-winner (1·X·2) and over/under odds, strip out the
margin to recover the implied probabilities, and fit an independent Poisson model for each team's goals
(the expected goals shown above). That yields a probability for every possible scoreline.
I then score every candidate prediction against that full distribution under the Kicktipp rules
and report the three scores with the highest expected points. The table's top row (in green)
is the expected-points-maximising tip — not necessarily the single most likely score.
Kicktipp scoring: 4 = exact score · 3 = correct goal difference (non-draw) ·
2 = correct tendency (win/draw/loss) · 0 = otherwise. Because a draw tip only pays when the match is
actually drawn, narrow favourites usually win on expected points even when a draw is fairly likely.
Odds via The Odds API. For fun, not betting advice.