Nfl Betting Model Excel
While most MLB models make projections based on how a team's been hitting as a whole, our offensive projections are based on each and every player included in that particular team's lineup for the day. This means our model waits for each lineup to be posted (usually within a few hours before first pitch), then analyzes it on a player-by-player basis. This method is to ensure the highest accuracy in predicting a team's performance.
The pitching/hitting evaluation component of the model uses advanced MLB metrics that go way over the casual baseball fan's head. Exit velocity, batted ball profiles, splits, plate discipline metrics, park factors, performance with or against certain pitches/velocities (combined with pitch usage rates), BABIP, FIP/xFIP, SIERA, and wRC+ are among the many metrics incorporated in the model. The challenge of MLB is analyzing advanced data to determine which players have been lucky and unlucky in relation to their actual performance. This is something that public/square bettors are very poor at figuring out, leaving a lot of value on the table in the betting market. Much like a player projection system, our model identifies a 'true' performance level for players and projects games accordingly.
Our NFL model uses an advanced stat concept known as DVOA, founded by the Football Outsiders research/analysis group. DVOA measures a team's efficiency by comparing success on every single. A sports bet tracker is the simple, but essential tool to maximizing your potential on the betting market. This useful, preforumulated Excel spreadsheet, will help you manage your wagers, by keeping track of. Does anyone use a betting model by implementing certain MLB statistics into an excel spreadsheet to find value in a play? If so please outline the certain statistics used and how you use them. Building the 2019 SportsGrid NFL Betting Model There’s a well-established process to modeling out NFL games and our approach. To start by paraphrasing Richard Feynman, the key to science is you.
The SportsGrid engineering and data science teams have a very particular set of skills. A set of skills that make the guy most likely to be asked: “Can you model this?”. The answer is invariably yes. We’ve released our NFL Betting Model and will be updating each week with our predictions for that week’s NFL games. The betting model picks can be viewed against consensus picks or against individual sportsbooks.
2018 NFL Betting Model Results
We are here to talk about how to build a betting model and in particular our betting model for the NFL. We ran this same model last year over on our DFS site DailyRoto and had a good degree of success. We also entered and tracked our results at Betting Pros for the entire season where the model finished 6th place out of more than 140 different competitors.
56% Overall
53% Against the Spread
58% On Totals
Betting Pros 2018 Accuracy Competition Results
We also submitted confidence ratings each week based on the same information presented on our NFL Model Picks. The results were interesting in that generally our highest valued picks performed the best (winning 64% overall) but that didn’t always carry down through on individual bet types or lower star recommendations.
Betting Pros Confidence Rating Results
Overall, the 2018 NFL Betting Model picks performed at a high level.
Building the 2019 SportsGrid NFL Betting Model
There’s a well-established process to modeling out NFL games and our approach. To start by paraphrasing Richard Feynman, the key to science is you make a guess at a truth or law, you calculate the consequences of that guess and then you test it against nature. If it disagrees with experiment your guess is wrong. This is the crux of the matter. Your model is worthless if it isn’t reflective of reality.
Step 1 then is measuring reality or the reality of that particular sport. That means data. Specifically, results and events. How are the games scored? How do those scores move over time? What critical values can I capture from the data? Thankfully we are quite good at getting this data.
For example, Win totals distributions for season Win Totals bets:
Which we can use to determine the value of a half win on your win total line from Vegas.
Or Game totals for the last five seasons:
Or home team margins:
What this data lets us do is to characterize normal for the sport.
What are the usual outcomes and their frequencies? What is true for the NFL? The next step is to use the data to build the model right? Nope.
Step 2 is to go out and do research. Like all good engineers, we were taught very early to never re-invent the wheel. We are not the first or the last person to try to build a predictive model for NFL games. There’s a significant amount of research out there into this problem. Reading it put us on the path to the things that matter. The four largest inputs into our NFL Betting Model include:
What’s the relative team strength? This is a major driving factor. To come up with our relative team strength we have our own proprietary blend which includes advanced team statistics and performance, “wisdom of the crowds” knowledge which considers some of the top statistical-based ELO models and betting odds including pinnacle lines. All of these factors have been back-tested to come up with our own ELO model and team strength ratings which are adjusted weekly during the NFL season.
Where is the game being played? Our model takes into account where the game is being played, whether it is at home, on the road or in a neutral venue. We also consider the specific venue and home-field advantage for that team.
What’s the surface? Is the game being played outdoors, indoors, on grass or on turf?
Is it very windy or very cold? Different weather conditions can impact not just the totals, but the expected winning percentages and spread for each NFL game. We weight the temperature and wind conditions in our model.
The model weights these over different periods of time in a variety of different ways but ultimately these are things that we know for each game or can derive and a lot of analysis suggests matters.
Step 3 is the nitty-gritty. Building the actual model and testing it against actual lines. Then doing it again. Then again and like a dozen or more times until you’re sure that it’s effective. One important fact for any model is that your work is never done. The underlying factors move. The games evolve over time.
Step 4 then is to continue to revisit it every offseason to adjust your priors as needed.
For the NFL we ended up with three NFL betting models, a spread betting model, a win total betting model and an over/under betting model. Let’s talk about all three.
NFL Spreads are the most popular bet type for the NFL and the sharpest market. The NFL is a low number of possessions game and that means that games suffer a very wide range of possible outcomes. Throw in the fact that humans are decent at linear outcomes (i.e. straight lines) and the margins are very small.
Over/Unders are next. Here your edge is more significant. Game Situation is very important to this. There are massive swings to play calling and scoring depending on location. It’s very much an exercise in data collation.
Finally, we have the moneyline were you need to pick a winner. This is much trickier because you get uneven payouts. Your price varies and what you really care about is if that price is right. We ran a Kelly betting recommendation system here. It basically looks at the price of a side versus the odds and makes a buy recommendation. That ran at a 26% profit last year.
So we have a model that ran well versus historical and ran well against actual games and lines. Good right? Not quite. I did mention that the offseason is for review and improvement.
Best Online Nfl Betting Sites
Step 5 is always continuous improvement. I am looking at making any necessary adjustments both to the underlying model be it for additional data or for any changes to the NFL game. I’m also looking at optimal betting strategies for all the models
Nfl Betting Percentages
Because I know that like me you are looking for an edge and I can tell you that we have a very particular set of skills.