World Cup - Quarterfinal Preview

 Hey kids

So today is the first day in 18 days without a World Cup match to watch.  Hope you're finally finding time to talk to your significant other or go to work or whatever stuff you're into.

Thought I'd take some time to throw together a summary of the model's predictions, and look at some of the tournament so far!

Overview

Here's where our remaining 8 teams stand for the remainder of the tournament.  Brazil continues to be the favorite, but Argentina is #2, as the next highest elo team.  Unfortunately (from a balance standpoint) the 3 largest elo ratings remaining are all on the same side of the bracket (and are still the highest 3 chances to win).  The model predicts that there's a 73.3% chance that the World Cup winner comes from the top side of the bracket (sorry Croatia, have fun!!!)

For some context to this, let's look at the starting odds for each of our Quarterfinalists:

No one will be blown away at the fact that Morocco has been the surprise of the tournament.  7/8 quarterfinalists are teams that were within the top 12 of odds to win the cup pre-tournament (5 of them in the initial top 8).  Morocco were +1300 to make the Quarterfinals pre-tournament, would be a nice little payday now!

Let's have a look at the path Morocco took to get here!

At the time, Morocco's win over Belgium was by far the most unlikely result (12.9% win chance) of their path so far.  Given what we know about these teams now, if there were to be a 2nd matchup, the model would have that matchup at about 58/42 win odds for Belgium.  This all goes to show an obvious drawback to elo as a rating method, given the relatively low number of games we have to draw from.  It can't adjust on the run for things like Hakim Ziyech coming back into the Moroccan side, or Romelu Lukaku's complete lack of ability to see a goal.  It takes actual results to update, and so it has!  Morocco are by far the biggest elo gainers this tournament:

Brazil, on the other hand, are the biggest gainers with regards to win likelihood.  The model essentially already had them as the strongest team, favored in any matchup.  And since "worse" teams than initially expected have made it to the quarters, they are the odds beneficiaries.  That's even taking into account that their rating is a little artificially depressed, due to losing to Cameroon in a dead rubber game 3 when they rotated heavily.  Elo, of course, treats this as a legit loss (massive upset!!)

Bracket Projection

This has now sort of just become a different format for the model overview, as teams' paths are very clear from here on out.  Portugal's dominating victory over the Swiss was enough to vault their rating high enough to become the favored team to escape the bottom half of the bracket.  This, however, is banking on a 70% chance to overcome the underdog Moroccans... and we've seen how teams underestimating the North Africans have done.

Game Forecasts

The model sees Brazil and Portugal as moderately large favorites (> 70%), while the other games are much closer to toss-ups.  Whatever happens, we're in for some outstanding football.  Hope y'all enjoy!

dope