Quick Blog Update
All of the work I’ve been doing coding these predictions for the NFL and College Bowl games is leading up to my one true passion, college basketball. I now have the code and the pipeline to make predictions on basketball games (plus my original model from last year’s tournament), but getting the data for the 2018-2019 season was the limiting factor (keeping in mind that Kaggle has all the juicy data through the end of the 2018 regular season from their March Madness competition)…until now. Wednesday night I worked until the wee hours of the morning and made a break through in gathering the data I wanted for this season. I need to do a little bit of clean-up, but I now have a way to grab results from the previous night’s games and apply predictions to the next night’s games. I’ve not decided yet how I want to report that out here. Maybe just a weekly blog on Mondays with predictions on Top 25 teams and teams of interest to me (like Texas A&M) for the upcoming week or something like that. The only problem with that is that, most of the time, teams play two games per week so maybe a Monday update and a Friday update. Will see how it goes. Starting February 1st I’ll be driving back and forth (or flying) to Plano from Austin for 24 straight weeks on Fridays.
Another update I wanted to do was with the West Texas Intermediate Crude post I made back in October. I’ve made some improvements (parameter changes) moving to a weekly average model rather than daily prices that I will report out sometime soon with a forecast for the next six months that (spoiler!) is a little more realistic than my last rather bullish forecast.
Wild Card Recap
Well, I certainly didn’t do too well in my Wildcard predictions. I had three of the home teams winning with one home team losing, and the exact opposite happened with all four games. Who saw that coming? Apparently not me! As far as scoring the four models I had plus the average of the four, the best model ended up being the modified Elo model I use (Model #2) with a 0.2033. Not a great score overall.
That brings us to this upcoming weekend’s Divisional Playoff games.
Divisional Predictions
| # | Game | Model 1 (Original) | ELO Model 2 (HFA=52.5) | Model 3 (FO+ELO) | Model 4 (Bayesian w/ 10,000 simulations) |
Avg of All Models | Vegas | NFL’s Cynthia Frelund Picks |
Poisson Predicted Final Score |
Expected Tot. Points Scored |
Vegas Over/Under |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | IND | 17.52% | 35.84% | 53.82% | 19.03% | 31.55% | 33.33% | 43.40% | 33 | ||
| KC | 79.11% | 64.16% | 41.28% | 80.97% | 66.38% | 69.16% | 56.60% | 34 | 67 | 57 | |
| 2 | DAL | 10.76% | 35.52% | 32.08% | 8.46% | 21.71% | 26.53% | 44.70% | 24 | ||
| LAR | 86.42% | 64.48% | 62.49% | 91.54% | 76.23% | 75.93% | 55.30% | 28 | 52 | 50 | |
| 3 | LAC | 45.99% | 43.41% | 31.24% | 32.53% | 38.29% | 36.63% | 41.10% | 25 | ||
| NE | 48.16% | 56.59% | 63.45% | 67.47% | 58.92% | 65.88% | 58.90% | 22 | 47 | 47 | |
| 4 | PHI | 8.46% | 37.30% | 7.72% | 23.41% | 19.22% | 23.31% | 39.90% | 22 | ||
| NO | 89.09% | 62.70% | 90.03% | 76.59% | 79.60% | 80.91% | 60.10% | 27 | 49 | 51 |
Once again, going by the average of all models highlighted in green, I have all four home teams winning. But so does Vegas and the NFL’s Cynthia Frelund as well. I think the only thing there, though, is, if we use 60% as a “threshold” for how confident we are in our picks, Cynthia is only confident that New Orleans beats Philadelphia, Vegas is confident that all four home teams will win, and I’m confident in all but the New England-Los Angeles Chargers game.
Let’s quickly break down my various models game-by-game:
Indianapolis at Kansas City (3:35 PM CST Saturday on NBC):
Indy is one of the playoff darlings here. They made pretty quick work of eliminating Houston last week and have quite a bit of momentum going into this game. Three of my four models, however, are fairly confident that Kansas City will win. The one that does not is the model that combines Elo scores, my original model, and stats taken from Football Outsiders. However, that model only gives Indy a 54% chance of winning so that pick isn’t quite as confident as the others. It should definitely be a fun game to watch and hopefully as high scoring as predicted.
Dallas at Los Angeles (7:15 PM CST Saturday on Fox):
As mentioned last week, I was at AT&T Stadium for the Cowboys’ win over the Seahawks. It was such a beautiful night, I wondered why Jerry Jones didn’t open up the roof or the side doors, but then realized it was so that the stadium would be loud. It was certainly loud and electric. I’m glad the Cowboys won, though, for certain.
Unfortunately, all four of my models are confident that the Cowboys will lose in Los Angeles including Model #4 which basically simulates 10,000 games between the two teams (I’d love to try and simulate more — Cynthia did 20,000 games — but I get memory errors in Google Colab trying to do more). And that last model gives the Rams nearly a 92% chance of winning! Yikes! As with last week, I just hope I’m wrong.
Los Angeles at New England (12:05 PM CST Sunday on CBS):
Is this Philip Rivers’ year? Only two of my models are confident that New England will win. Of course, keep in mind I only factor home field advantage and not weather effects. With snow in the forecast for New England, it’d be hard to consider the Chargers coming even close to winning. But that’s why they play the game, right? I feel that if there’s any game this weekend that will feature an upset, it’s definitely this one.
Philadelphia at New Orleans (3:40 PM CST Sunday on Fox):
This game is fascinating for many people that live in and around the Austin area. Both quarterbacks, Nick Foles of the Eagles and Drew Brees of the Saints, went to school at Westlake High School in West Austin (although I’ve often argued that Westlake is not Austin so I don’t understand why it’s considered Austin Westlake). As you might imagine, the coach and team of both QB’s have been interviewed extensively locally.
The Eagles are the other playoff darling here, though, having won on a double donk from the Chicago kicker last week (oh, it was brutal). However, the Eagles winning does keep my hopes alive that with a win this weekend as well as a Cowboys win, then I will be able to go to my first Conference Championship game in my life in Dallas (even if my credit card does not want that to happen)!
Unfortunately, for the Eagles anyway, this is the only game that has all four of my models, Vegas, and Cynthia confident that the Saints will win. Ah, well, we tried our best!
Enjoy this weekend’s games.