Last week I did present to you the first prediction of the Lars. The Lars is my Premier League Prediction model. In contrast to the Revevend, which is based on a Naive Based Classifier, this model is based on the Poisson statiscal model.

Just like the Nyls, it does predict the 1X2 market, Both team to score, the over/under 2.5 goals and correct score prediction for each game from the English Premier League. I will keep record for all predictions. Both for the % calculation and the correct score predictions.

Let's see how the model did perform the first week. Attention, I didn't predict scores for 3 games because not enough data was available for the model.

1X2 market based on %

Well an excellent start with predicting the winner or draw of a game. 5 out of 7 games were correct, this was good for a profit of 1,77 units and a ROI of 25,29%!

BTTS market based on %

Only 4 out of 7 correct, but still good enough for a small profit. The profit was 0,04 units, which is better than a loss. The ROI was an amazing 0,57% :)

Over/Under 2.5 goals market based on %

For the over and under 2.5 goals market, it was a bad week. Also here only 3 out of 7 correct but here a loss is recorded of -2.48 units.

This reflects of course a negative ROI of -35,34%. Better luck next week I guess!

Correct Score predictions

Very simple 0 out 7 correct, a loss of 7 units and a negative ROI of -100%.

1X2 market based on the correct score prediction

One week also a correct score prediction is done for each game. What if you would have followed this for the 1X2 market. Meaning that a 2-0 prediction would have resulted in a home win bet, a 1-1 prediction in a draw and so on.

In contrast to the % predictions, we do have a small loss here. Only 4 games predicted correctly, resulting in a loss of 0.12 units. Reflecting a negative ROI of -1.71%.

BTTS market based on the correct score prediction

Just like the % prediction 4 out of 7 correct. Also here a small profit of 0.04 units.

Over/Under 2.5 market based on the correct score prediction

In contrast to the % prediction a profit was generated here. A whopping 0.09 units.


Next week the results will be more meaningful. The next Lars will be posted tomorrow!


Cheers,

Peter