Thursday, August 28, 2014

2014 College Football: Week 1

Football is back. Football is today.


I'm bringing back my Watchability ratings this year. They're a way to try to identify which games are likely to be close and well-contested. There's a long post (with math even) located here with more detail on Watchability - how it works and how it's calculated. But in summary

  • Watchability is a measure scaled from 0 to 100
  • Its purpose is to identify which games might be fun to watch, based on how good the teams playing are, and how close the game is likely to be
  • Oregon @ UCLA (on October 11th) is the currently highest rated regular season game (91.3!)
  • Watchability doesn't predict upsets, those are just surprises to be enjoyed by us all
This week
Without further ado: this week's Watchability guide. The orange bars are Watchability. Teams are shaded in green by their chance to win. Darker = better.

Tuesday, August 26, 2014

Now is the Time for Pumpkin Ale

Pumpkin ale and FOOTBALL! Happy first week of Football everyone! I've expanded my Team Dashboard for this year. Please take a look!

Once you look through the team dashboard and get a feel for it, head over here and make your own!

Friday, August 15, 2014

2014 College Football

It's coming!!!

Model structure is coming together. Team ratings are still just random numbers (i.e. WSU in the Rose Bowl)

Friday, July 18, 2014

World Cup Wrap Up

Have some final thoughts to share after a few days to reflect on the World Cup.

What a month! Creating and uploading heat maps plus in-game probability charts for every World Cup game turned out to be WAY more work than I bargained for. I had to teach my wife to update graphs while I was at work, I made updates while riding in the car, from Starbucks, from bars, from the Oregon Coast, from the San Diego Airport, and from Pacific Beach! I gathered some numbers on Posting frequency and ridiculousness:

  • Greece vs. Costa Rica was that crazy first knockout game that featured a halftime goal, a red card, a 91st minute equalizer and a penalty shootout. 
  • England were featured in two games (vs. Italy and vs. Costa Rica) where I only made one post. I'm not sure what that says about English football. 
  • That last week of group play I was making and posting win probability graphs AND advancement probability graphs for every pair of games.
  • I have no specific knowledge Nate Silver was imitating my Heat Maps when he made this post. But his work looks really really similar to mine.

I also managed way more traffic writing about the World Cup than ever before. I've written about three major topics so far: The 2012 Presidential Election, the 2013 College Football season, and the 2014 World Cup. I summed up my total page-views for each of the 3:

What a huge difference! Thanks for reading! The 3 most viewed posts were:
  1. USA USA USA!!! vs. Germany and Ghana vs. Portugal
  2. World Cup SEMIFINAL: Germany vs. Brazil
So extra thanks to USA#1 and of course ze Germans.

I also watched way more World Cup than ever before. For the first time I could see the difference in level of play between the World Cup level and the MLS level (Go Sounders!). Mostly I noticed it in passing accuracy, first touch quality, and off-ball movement/coordination. Soccer played at this level (and even MLS level) is really fun to watch.

I'm ready for a nice break. I'd also like to start posting more soon about MLS, of course 2014 college football (Go Huskies!) and who knows what else! Stay tuned!

Sunday, July 13, 2014

World Cup Final: Germany vs. Argentina!

Are we ready? It's time for the biggest soccer match in the world, rumored to be watched by over a billion people! The game features probably the world's best team (Germany) vs. probably the world's best player (Messi) and it definitely does't feature Brazil.

Germany is the favorite but barely. Kind of like how they were the favorite but barely against Brazil (side note: there is a 0.057% chance of another 7-1 Germany victory!)

Heat Map - The top 4 most likely outcomes are 2-1, 1-0, 1-2, and 0-1 but even those are all under 15% chance. It's pretty wide open! Especially with Germany's vaunted offensive machine.

World Cup Wrap Up: Goal Differential

Before the beginning of the World Cup, I modeled expected goal differentials for every team. As the World Cup comes to an end I thought it would be fun to look at what the model expected from teams vs. what they did.

Goal differential is Goals Scored minus Goals Conceded. Let's take the USA#1 as an example:

USA scored 2 goals against Ghana and allowed 1, so for that game they had a goal differential of 1. For the tournament as a whole the USA scored 5 goals and conceded 6 for an overall goal differential of -1. Make sense?

For the whole tournament, the model predicted the USA would have a goal differential of -2.5. By coming in at -1, the USA out performed the model's expectations by 1.5 goals (WOOHOO USA#! etc).

How did the rest of the field do?

  • Netherlands and Germany have both been over 8 goals better than the model expected (and Germany still has one game to go)!
  • Some of the fun positive surprises (Costa Rica, Mexico) jump out at the top of the list, and of course the wild disappointments (Spain, Portugal) sit near the bottom
  • Brazil being in last is a function of them having a very high positive expected goal differential and (of course) them having given up 10 goals in their final two games
  • This isn't an absolute ranking of teams, but a ranking of how teams have performed relative to expectations

Saturday, July 12, 2014

World Cup 3rd Place Game: Netherlands vs. Brazil

Holy cow Brazil.

This post  has a little more on how the win probability graphs work
This post has a little more on how the pregame heat maps work