Category Archives: NFL

2016 NFL Playoffs Recap

After an otherwise forgettable first few weeks, this year’s NFL Playoffs were capped off by the greatest comeback in Super Bowl history, led by the league’s undisputed G.O.A.T, better known as Tom Brady. Facing a 25 point deficit part way through the third quarter, the legacy of the Brady-Belichick era was put on the line, and the light at the end of the tunnel was looking brighter and brighter for Roger Goodell. Little did he know, that light was a reflection beaming off of Brady’s fifth Super Bowl ring, and he’d soon be on the field coughing up the Lombardi trophy to the team he’s conspired against for the past several seasons. Although he dodged the bullet last year when the Broncos knocked the Patriots out in the AFC Championship Game, the four-game suspension he imposed on Brady to begin the year proved to only fuel the fire in the Patriots’ season of vengeance on the commissioner. Although our Models didn’t contain a variable of this nature, 5 of them correctly predicted the Patriots to beat the Falcons and our spread model had them winning by 9. For the playoffs as a whole, we had two models finish 10-1 (only missing the DAL/GB game) and our average Model finished 9-2. See the tables below for our predictions for all of this year’s games.
Continue reading 2016 NFL Playoffs Recap

2016 Super Bowl Predictions

We use team statistics to create models that predict outcomes of NFL games. Our models generate either a win probability for the home team or a point spread, with each model generating a unique prediction. See this article for a more detailed breakdown of each model. So far this postseason, our average probabilities have picked 8/10 games correctly, and Models 2 and 3 have been the most accurate individual models, both going 9 for 10.
Continue reading 2016 Super Bowl Predictions

2016 NFL Championship Round Recap

In the two conference championship games on Sunday, our average probabilities went 1-1, correctly picking the Falcons (57%), but missing the Steelers (55%), to bring the average probability record for the playoffs to 8-2. Models 2 and 3 remained the most accurate, as they each had the Falcons winning (78% and 75%, respectively) and the Patriots winning (61% and 59%, respectively) to bring their records to 9-1 for the playoffs as a whole. Model 5 also went 2-0 this weekend, and improved to 8-2 for the playoffs. Our spread missed the Falcons (had ATL winning by only 3 points), but correctly predicted that the Patriots would cover (had NE winning by 7).  Check back later this week for our Super Bowl predictions!
Continue reading 2016 NFL Championship Round Recap

2016 NFL Championship Game Predictions

We use team statistics to create models that predict outcomes of NFL games. Our models generate either a win probability for the home team or a point spread, with each model generating a unique prediction. Our models consider many different metrics, such as turnovers, first downs, points scored, pace of play, offensive and defensive efficiency, as well as a few metrics that we have created ourselves. Beyond these statistics, some models consider factors such as “did a team make the playoffs last year?” and “how far is the away team traveling?” See this article for a more detailed breakdown of each model. So far this postseason, our average probabilities have picked 7/8 winners correctly, and Models 2 and 3 have been the most accurate – also going 7 for 8.
Continue reading 2016 NFL Championship Game Predictions

2016 NFL Divisional Round Recap

In the second round, our average probabilities went 3-1, and all models individually were either 3-1 or 2-2. Model 1 was the only one to correctly pick the Packers over the Cowboys. For the playoffs as a whole, Models 2 and 3 are performing best, as they have correctly picked 7 of the 8 games, while Model 6 is performing worst, only picking 5 of 8 games correctly. The average probabilities are also 7-1, only missing the Packers/Cowboys game. Click here to see all of our predictions from the past weekend. Check back later this week for our Conference Championship game predictions!
Continue reading 2016 NFL Divisional Round Recap

2016 NFL Divisional Round Predictions

We use team statistics to create models that predict outcomes of NFL games. Our models generate either a win probability for the home team or a point spread, with each model generating a unique prediction. Our models consider many different metrics, such as turnovers, first downs, points scored, pace of play, offensive and defensive efficiency, as well as a few metrics that we have created ourselves. Beyond these statistics, some models consider factors such as “did a team make the playoffs last year?” and “how far is the away team traveling?” See this article for a more detailed breakdown of each model. The article below focuses on the four divisional round matchups for the coming weekend, but a breakdown of predictions for the entire 2016 NFL Playoffs can be found here, and our predictions from round 1 are here, where our average probabilities correctly predicted all four winners.
Continue reading 2016 NFL Divisional Round Predictions

The Methodology Behind our NFL Playoff Models

This article provides some background on the components of each NFL Playoff Model. By making use of multiple models that are comprised of different modeling techniques and variables, we can better assess each game. For example, if all the models are predicting the Pittsburgh Steelers to beat the Miami Dolphins, we can feel very confident in picking the Steelers to win. If the models are split, we can explore each model individually based on their predictors to identify areas where the model may be taking into account less than perfect information. For instance, if one model emphasizes defensive performance and Seattle is currently missing Earl Thomas, maybe that model does not represent Seattle as accurately as the others. Ultimately, by looking at multiple models we are able to reduce the noise inherent in all predictive models.

Continue reading The Methodology Behind our NFL Playoff Models

2016 NFL Playoff Simulation

Using probabilities from our five NFL Playoff Models, we have run a number of simulations for the upcoming 2016 NFL Playoffs. This allows us to estimate how often each team makes the division round, conference championship, Super Bowl, and ultimately win the Super Bowl. In order to do this, we took each game and generated a random number between 0 and 1. If the random number is less than our predicted probability of the home team winning, then we advance the home team as the winner of that game (and vice versa if the random number is greater than our probability). For example, Model 1 has a probability of 0.85 that the Houston Texans will beat the Oakland Raiders. If the random number is 0.95, we would pick the Raiders to advance, and if the random number is 0.55, we would pick the Texans to advance. This methodology was applied to the entire NFL Playoff bracket, running 10,000 Playoff simulations for each of our five models (so, each model generates 10,000 brackets and 10,000 super bowl champions).
Continue reading 2016 NFL Playoff Simulation

2016 NFL Wild Card Predictions

We use team statistics to create models that predict outcomes of NFL games. Our models generate win probabilities and point spreads, and we consider six different models, each of which generates a unique prediction for a given game. Each model considers metrics from a number of different areas, such as turnovers, first downs, points scored, pace of play, offensive and defensive efficiency, time of possession, drive statistics, point differential, as well as a few metrics that we have created ourselves. Beyond these statistics, the models also consider factors such as “did a team make the playoffs last year?” and “how far is the away team traveling?” The article below focuses on the four Wild Card matchups for this coming weekend, but a breakdown of predictions for the entire 2016 NFL Playoffs can be found here.
Continue reading 2016 NFL Wild Card Predictions