Week 11

 

Hi all, the playoffs are well underway and have been full of surprises this year. There have been a few big upsets this year including both the number one and number two seeds in the Eastern Conference being knocked out in the first round of playoffs. These upsets did not bode well for our predictions as our models favored both of these teams to win their matches. Our model only accurately predicted the outcome for 3 out of the 8 matches in the first round. Ouch! Thankfully our model did a lot in the second round of the playoffs.

 

Our very own Phoenix Rising have pulled off a couple of big upsets in the playoffs, knocking out both the two and three seeds in the Western Conferences. They are going up against the number one-seeded Sacramento Republic this weekend. With another upset, they will move on to the Championship. Good luck Rising!

 

I was assigned the Eastern Conference, so I created a model for the Easter Conference final between the Charleston Battery and Louisville City Football Club. The model is predicting a win for Charleston 41% of the time, a victory for Louisville 31% of the time and a draw 28% of the time. Interestingly, there are no draws in the playoffs (the game would go onto extra time and eventually penalty kicks if needed) but it would likely be very difficult to account for this in a model.

 

While this has been a fun experiment, our research is ultimately looking at how travel (particularly distance and time zone differences) affects teams and trying to find a way to incorporate that into a model for our predictions. So as we continue to follow the playoffs, we are also wrapping up our data collection to move onto the next step in our research.  

 

 


Comments

Post a Comment

Popular posts from this blog

Fall 23: Week 1

Image
      Coming into this semester, I really wasn’t sure what I would be doing for my authentic experience. I ended up with a few different exciting prospects including a coding project with Professor Manning and a Calculus project revolving around the tabular method for integration. However, at the last minute I came across a Sports Analytic Research Project being led by Dr. Posta here at PC. As a huge sports fan, I could not have dreamed up a better research project than one marrying math and sports. I was fortunate enough to be chosen to participate in this internship and I am beyond excited to be a part of this project!
Read more

Week 4

Image
      This past week, Dr. Posta tasked me with again building a simple model to predict an upcoming match between the Phoenix Rising and Detroit City FC. For this week, there was a focus on using Excel formulas while building this model. Once again, using stats from the past 10 games, I calculated expected goals for both teams heading into this match. Using that xG stat, I calculated the probabilities of several different possible outcomes of the game. Ultimately, the model expected the Phoenix Rising to win 29% of the time, lose 26% of the time, and draw the remaining 46 percent of the time, with the single most probable outcome being a 1-1 draw 12% of the time. The actual outcome of the match was a 5-0 blowout by the Phoenix Rising. The model only expects a 5-0 outcome 0.00006 percent of the time, so the result of this match was very much a statistical anomaly. This is what brings us to the next stage of our research project, data collection. For this project, we are going to
Read more

Week 2

Image
  A bit late with my blog post for this week. We got hit pretty hard by the storm on Thursday and lost both power and internet. Thankfully power was restored early Friday morning but four days later, we are still without internet service. Have a technician coming out tomorrow, so hopefully that gets resolved. Not having internet access while taking online classes is problematic to say the least. But I digress. Earlier this week I had the pleasure of meeting with Dr. Posta. We discussed what the internship will entail in greater detail he got me started with some further readings on the model we will be using for this project. Soccer is generally a low-scoring sport. This can be inherently problematic when attempting to create a model that can accurately predict match outcomes. Interestingly, the model doesn’t place as much weight on match scores as one might initially think. Take for example a match where team A defeats team B 2-0. In this match, team A has 2 shots on goal and sc
Read more