Testing the Score Predictor

Is this article I described how using Poisson distribution can be used to predict the most likely outcomes in a football match. It’s a well known, and a well documented method. My thinking however is that at the start of a season there isn’t enough data to make results significant but thought what the hell, it’s worth a try anyway. So the table shows the four most likely scorelines according to the data. The numbers in square brackets are the possin values, the higher the number the more likely score.

FixtureScore Prediction 1Score Prediction 2Score Prediction 3Score Prediction 4
West Ham vs Man United[12.27] 0-1 [11.924] 0-2[10.049] 1-1[9.765] 1-2
Arsenal vs Watford[9.495] 3-0 [8.791] 2-0[8.064] 3-1[7.692] 4-0
Everton vs Fulham[10.936] 1-1[9.337] 1-2[8.331] 0-1[7.179] 2-1
Huddersfield vs Tottenham [15.053] 0-2[14.179] 0-1[10.653] 0-3[8.777] 1-2
Man City vs Brighton [15.684] 3-0[14.583] 2-0[12.652] 4-0[9.039] 1-0
Chelsea vs Liverpool[10.953] 1-1[8.469] 1-2[8.32] 2-1[7.209] 0-1
Newcastle vs Leicester[12.023] 1-1[9.172] 1-0[8.712] 2-1[8.296] 0-1
Wolves vs Southampton [14.431] 2-0 [13.941] 1-0[9.959] 3-0[9.059] 2-1