New User? Sign Up / Forgot Your Password?
All Live
Projections
    All Live Projections

    Introducing NFL Expected Points (12/20/16)

    By Frank Brank @realfbra
    Today, we're introducing a new metric, Expected Points, for NFL games. Using simple box score statistics, we've developed an algorithm using regression analysis to predict the final score and win percentages for each team in any given game.

    We intend to publish more on the statistic going forward as the building of the model is still fluid; however, we wanted to publish initial results before the end of the season. As one would assume, turnovers are extremely important to the outcome of NFL games. For that reason, we will be removing end of half turnovers (i.e. interceptions on a Hail Mary or fumble on lateral plays at the end of halves) since desperation-type plays do not provide any value to the outcome of the game.

    A few things we learned in the creation of this Expected Points model are:
    - Turnovers are even more important than we had previously perceived.
    - Penalty yards matter, but they only have worthwhile, measurable effects when there are large disparities within a game.
    - The location of turnovers are mostly random, but largely change the scoring and outcome of the game.

    Here is a quick look each team's predicted score and win percentage for Week 15 in the NFL:


    One immediate standout in the table above is the Bengals, who only had an expected score of 6.7 points and a 4.2% chance to win in a game that they led throughout. The results are largely in part due to the Steelers kicking six field goals and the Bengals gaining only 38 yards in the second half, after scoring on their first four drives to start the game (one of which was only 22 yards). Also, about 39% of their yardage gained was "lost" due to penalty yardage.

    Another standout is the Jaguars-Texans game. The Texans would have been expected to win by a final score of approximately 18-to-10. In actuality, they need a last minute drive by back-up quarterback Tom Savage to win the game. This expectation mostly came from the result of the Jaguars gaining only 150 yards of total offense, committing 91 yards worth of penalties, and being gifted a few scores by Brock Osweiler.

    The Packers stood a 79.7% chance to win based on our model; however, they needed a bomb to Jordy Nelson to pick up enough yardage for the game-winning field goal in Chicago. Our model expected a larger margin of victory mostly due to the Bears committing four turnovers (three of which were accounted for). The only reason the Bears were able to mount a large comeback was due to Davante Adams dropping two easy touchdowns.

    As stated above, this is still a fluid method and we expect to make constant improvements going forward. A few potential avenues where we will look to improve the model are:
    - Implementing special teams,
    - Including starting field positions after kicks and/or turnovers,
    - Score-adjusted yardage,
    - Testing predictability which may include regressing turnover rates, especially for quarterback interception rates.

    Needless to say, we are excited about the model's implemenations going forwards.
    Print This Article
    PREVIOUS ARTICLE
    NFL Week 16 Daily Fantasy Targets (12/22/16)
    NEXT ARTICLE
    NFL DFS Value Plays - Week 15 (12/15/16)
    Want More Tips and Information? Click To Register!

    COMMENTS

    SEARCH BLOGS

    ARCHIVE

    The Predictalator

    The Predictalator is the most advanced sports forecasting software available today because it has the ability to account for all of the relevant statistical interactions of the players, coaches, officials, home field advantage and weather in each game.

    Picks for every game, including our Play Value Key and Calculator recommendations for optimal bankroll management and access to the Play Analyzer can be purchased in the Shop or by game here.

    03/27/2017 Highlight: With the start of MLB right around the corner, we can use our ResultsFinder tool to uncover some past success in baseball. April has historically been a great month for the Predictalator, as evidenced by last season's outstanding 41-22 (65.1%) record on all "normal" or better ML, RL, and O/U plays. A $50 bettor would have profited a whopping $1,002 using our recommended wager sizes in April. MLB season win totals have also been extremely fruitful in the history of this site, with the Predictalator producing an outstanding 54-19 (74.0%) record on all playable win totals.

    March Madness is winding down and it was another solid month for the Predictalator. There were a total of 69 "normal" or better college hoops positions (sides and totals) in the month of March, and the Predictalator posted a strong 42-27 (60.9%) record, generating $704 in total profits. NBA has also followed suit with "normal" or better positions winning at a 58.1% clip since the beginning of February, and at a 57.9% clip for March.

    NHL profits continue to roll right along as well, as the Predictalator has cashed in on nine of 12 "normal" or better positions in March. The Predictalator has now amassed an incredible 45-28 (61.6%) on all "normal" or better money lines, puck lines, and totals this season, totaling $499 in total profits for a ?$50 bettor.

    Check out the Shop now to learn more.

    Recommended Features

    LOGIN

    LATEST ARTICLES

    MLB Win Totals - Over/Under Picks
    We take a look at the 2017 MLB over/under team season win totals. Highlighted, "normal" or better MLB Win Total Picks are 54-19 all-time. See free picks on the Cubs, Dodgers, Phillies and more.

    Updated Bracket Odds - Final Four
    The Predictalator simulates the Final Four 50,000 times to see how likely each remaining team is to win the NCAA Tournament

    FIND US ONLINE


    SportsRadar


    All registered accounts are subject to our General Terms Of Service. Content is intended for informational purposes only. This is not a gambling website.

    FSTA Member 2015
    Home | The Predictalator | Blog | Privacy |  Media Contact |  B2B Solutions |  Sitemap |  Customer Support | 
    © 2010-2015 PredictionMachine.com, LLC. All rights reserved. | Site Designed by Randombell.com