Goalimpact: Bundesliga and Serie A 2023/24 prediction preview

Goalimpact: Bundesliga and Serie A 2023/24 prediction preview

In our last article, we examined how Goalimpact can help bettors find edges in Pinnacle's markets for specific games. In this article, we will examine not just the probabilities that emerge for a single game, but also how Goalimpact rates the entire season, specifically for Serie A and the Bundesliga.

The model: How Goalimpact predicts soccer leagues

Since Goalimpact predicts game outcomes based on the starting XIs of both teams, there is, at least in principle, the possibility to predict entire leagues as well.

However, some additional challenges emerge when you want to predict a league's final table.

The starting XIs of the teams change frequently from game to game due to various reasons. Players get injured or rotated out of the starting XI due to fatigue, tactics might change and coaches might switch. All these aspects make league predictions more complex, at least based on objective Player Ratings.

Hence, in the past, Elo ratings were often used to cope with the dynamic nature of starting XIs during league seasons. This is, by the way, also how Goalimpact started initially.

Anyway, we also do league predictions – the question is, how do we tame the complexity of the game? Or more precisely – how do we estimate the future starting XIs and game probabilities over a long time?

If you think about soccer in the mid to long run, the quality of the starting XIs is insufficient to estimate a club's strength. So, what we do instead is not just rate the best 11 players and assume they will play every game, but based on historical data, we include how playing time is distributed in the squad. Overall, the majority of playing time will be given to the Top 16-18 players with the highest Goalimpact.

However, not all of them will play for the same amount of time. Our model assumes that the players with the highest Goalimpact will play almost 100% of all possible minutes. With every lower position in the Goalimpact Team Hierarchy, 5% less playing time is given to the players. Consequently, Player 19 and 20 will have no playing time anymore in our model.

Some important objections: 

Of course, it is not always the case that the good players (according to Goalimpact) will play, because coaches and staff may not see the same quality of a player, but it is a practical rule of thumb. Furthermore, based on the empirical data we have, we know the rough distribution of playing time for the players down to player 16. This obviously can differ from reality as well. So, the Team Goalimpact we give is not just a simple average of the Top squad players, instead it emerges from the historical data. In addition, if a player is injured, we do not take them into account until they will be available again. 

The Team Goalimpact is thus not a static value but a dynamic one. 

Another key factor that Goalimpact corrects for is the game order. Let’s say, as an example, that you play against the best team early on, but your best players will miss the game due to injury. In this case, your winning probability in this game will be lower than if you competed later into the season after your best players have recovered. So, we model the available squad for the future. Additionally, we also take ageing into account.

A young player will tremendously improve within a season, but older players will decline. This explains why teams with a lower Goalimpact can have a higher expected ranking in our predictions than a team with a higher Goalimpact. Lastly, Goalimpact also takes home field advantage into account, which also emerges from empirical data.

Once that model is established, it’s a mere simulation of outcomes. Since randomness is such a crucial part of the game, you have to run a tremendous amount of simulations to achieve probabilities for rankings. We currently do 50,000 simulations – so the league is played out virtually 50,000 times with the current conditions (available squad and match plan).

But enough theory – let’s have a look at the predicted champions for Serie A and the Bundesliga and compare it to Pinnacle's odds.  

Goalimpact: Bundesliga champion prediction

Bayern Munich won the Bundesliga 11 years in a row. Despite the dangers such a monopoly poses to soccer in general, Bayern will remain the favourites next year as well. In particular, taking Raphael Guerreiro as a free transfer hurts the competition in the Bundesliga, since he was one of the best players of Bayern's biggest rivals - Borussia Dortmund.

So, how do the title predictions of Pinnacle and Goalimpact compare?

Both models see Bayern [-500] 1.200 clearly in the pole position. However, Goalimpact sees Bayern’s dominance as more pronounced, giving them a winning probability of 79.6%. Pinnacle, in contrast, estimates Bayern's odds at 1.2, equaling 83.3% – a difference of 3.7%. This is not a significant difference we would take a bet on.

This implies that the chances for Dortmund, Leipzig, and Leverkusen, the biggest rivals in both models, are higher from Goalimpact's perspective than from Pinnacle's. While we see Dortmund's chance at 8.6% and Leipzig at 5.9%, Pinnacle sees Leipzig on a similar level (7.1%) but Dortmund significantly better (18.1%) Here, Edin Terzic’s impact could account for the difference of 9.5% between Pinnacle and Goalimpact. Goalimpact does not (yet) take Coachimpact into account.

However, his 1.5 seasons were impressive and seem to indicate that he makes players better, demonstrated by the improvements of Donyell Malen and Emre Can. In contrast, Goalimpact gives Leverkusen higher chances (4.1%) to win the league than Pinnacle (2.1%). Nevertheless, both models agree: Bayern Munich are very likely to win the league again. 

Undoubtedly, the squads are constantly changing during the ongoing transfer period: Sadio Mané, a player whom Goalimpact rates among the best Bayern players, just left Bayern. Even though Mané did not play a lot, Goalimpact calculated the odds with him in the squad. Consequently, his absence now decreases Bayern’s strength and winning probability. With a possible Harry Kane transfer, Bayern's Team-Goalimpact would rise approximately five points and improve their title chances to almost 90%.

This huge difference of 27.6% is hard to pin down on soft factors. The different ratings of the club might be rooted in Inter's strong performance in the Champions League, giving them a strong rise in their Elo rating.

*Odds subject to change

 

Do you want to be informed about bookmakers' latest promotions?

Click  and subscribe!