ZIP model performance incl. minor leagues

Some days ago I extended my ZIP Poisson model by some additional leagues. These are: Championship, Seria B, La Liga 2, Eredivise, Liga Portugal. It’s always helpful to be able to select more possible bets. Playing more bets reduces the variance of your hit rate and provides a more stable average profit. So let’s have a look, how the ZIP Poisson model performs including the new leagues.

As for each other betting simulation I used a 1 unit flat-stack betting strategy. Each bet, which indicates value, is selected. The bets are placed against fair Bet365 odds. So the betting is free from any margin. Bet placed between the start of the season 2018/19 and now result in a loss of 908,8 units.

This corresponds to an average yield of -5.22% or a loss of 0.05 units per bet. That looks not really good.

But as I already recognized multiple times, my models do not predict the draw very well. And that’s again one of the main reason for the big loss. It seems like nearly every indicated draw bet lost, which results average loss of 73,44% for all settled draw bets. So, again just ignoring draw bets, will definitley increase the expected return.

Specific leagues are another week spot of my models. That’s something I recognized while analyzing my bets for Serie A. And it’s the same case for 2 of the new leagues: Serie B & Liga Portugal. In comparison the model performs way better for e.g. the Eredivise.

Taking these 2 facts in cosideration and exlcuding draw bets as well as bets for Serie A, Serie B and Liga Portugal provides a way better profit. Over the course of 4.5 season (2018/19 till now) all bets result in a profit of +555.1 units.

So the overall yield increased from -5.22% to +4.70%. Despite this good looking yield, it’s import to again take a look at the running profit. The main profit was created during the years 2018 and 2020. With the start of 2020 the profit curve becomes way flatter. I would guess looking at the profit of the last 1.5 – 2 years the yield might nearly equal to 0.

Taking into cosideration, that the betting backtest was done with fair odds, the yield indicates, at which bookies the model could have provided profit. If it’s a classical soft bookie with a typical margin of 5% – 6%, this model would definitely not have provided any profit.

In the next step, I will take a look, how the Vanilla Poisson model performed including the additional leagues.

If you have further questions, feel free to leave a comment or contact me @Mo_Nbg.