Calgarypuck Forums - The Unofficial Calgary Flames Fan Community - View Single Post - Rube's 2019 NFL Gambling Odyssey (and NFL Preview Thread)

rubecube · 08-28-2019, 02:21 PM

Hey gang, after last year's losing season and a general trend I'm noticing with my life in which I'm not quite watching as much football as I used to, I'm trying to take on a little bit more of an analytics approach to gambling this year. As such, I've built a pretty extensive database that I'm going to be relying on this year to make my bets. However, due to this being the first year that I'm employing this strategy, I'm not going to share exactly what I'm doing until I know whether it's successful. Being the man of science I am (LOL), I'm also going to be making bets based on both my old and new methods and tracking them to determine which ones are more successful throughout the year.

One of the methods I've been using for the last couple of years to predict overall W-L records is using the Pythagorean Expected Wins totals from previous seasons and then looking for value plays where I feel Vegas has set the O/U too far on one side. This has been fairly successful for me and I've been tinkering with it a bit more this offseason to try and get more accurate predictions.

With that said, I thought I'd use this thread to do a bit of an NFL preview series that's based entirely on data and not on my own personal opinions and biases.

First, a bit of a primer:

The main stats I've elected to include for analysis are aforementioned PEW and the PEW differential (PEW - Actual Wins). I use the "Adjusted Pythagorean Wins" model developed by football outsiders. Here's a link that provides a little more background on how model works.

https://www.footballoutsiders.com/st...thagorean-wins

PEW is incredibly consistent in being able to predict which teams are candidates for positive and negative regression in the upcoming season. In 2018, teams with a negative PEW differential of -0.5 or greater lost an average of 2.5 more games than the previous year. Teams with a positive differential of 0.5 or greater won an average of 2 more games the following season. Of course there are exceptions and outliers but the model is generally solid

The other stats I look at when predicting records for the upcoming season are fumble recovery rates, adjusted man games lost (injuries), winning percentage in one-score games, and net penalty yards vs. league average. Why? Because most of these factors are a result of "luck" and fluctuate from year to year, so if a team is above or below the league average in any of these areas, there's a possibility for regression. Fumble recovery rates and winning percentage in one-score games, for instance, average out as 50/50 propositions and are generally pretty good indicators of how lucky a team was in the previous season.

So now that we have that out of the way, I'll go ahead and post the previews.

08-28-2019, 02:21 PM	#1
rubecube Franchise Player Join Date: Mar 2006 Location: Victoria Exp:	Rube's 2019 NFL Gambling Odyssey (and NFL Preview Thread) Hey gang, after last year's losing season and a general trend I'm noticing with my life in which I'm not quite watching as much football as I used to, I'm trying to take on a little bit more of an analytics approach to gambling this year. As such, I've built a pretty extensive database that I'm going to be relying on this year to make my bets. However, due to this being the first year that I'm employing this strategy, I'm not going to share exactly what I'm doing until I know whether it's successful. Being the man of science I am (LOL), I'm also going to be making bets based on both my old and new methods and tracking them to determine which ones are more successful throughout the year. One of the methods I've been using for the last couple of years to predict overall W-L records is using the Pythagorean Expected Wins totals from previous seasons and then looking for value plays where I feel Vegas has set the O/U too far on one side. This has been fairly successful for me and I've been tinkering with it a bit more this offseason to try and get more accurate predictions. With that said, I thought I'd use this thread to do a bit of an NFL preview series that's based entirely on data and not on my own personal opinions and biases. First, a bit of a primer: The main stats I've elected to include for analysis are aforementioned PEW and the PEW differential (PEW - Actual Wins). I use the "Adjusted Pythagorean Wins" model developed by football outsiders. Here's a link that provides a little more background on how model works. https://www.footballoutsiders.com/st...thagorean-wins PEW is incredibly consistent in being able to predict which teams are candidates for positive and negative regression in the upcoming season. In 2018, teams with a negative PEW differential of -0.5 or greater lost an average of 2.5 more games than the previous year. Teams with a positive differential of 0.5 or greater won an average of 2 more games the following season. Of course there are exceptions and outliers but the model is generally solid The other stats I look at when predicting records for the upcoming season are fumble recovery rates, adjusted man games lost (injuries), winning percentage in one-score games, and net penalty yards vs. league average. Why? Because most of these factors are a result of "luck" and fluctuate from year to year, so if a team is above or below the league average in any of these areas, there's a possibility for regression. Fumble recovery rates and winning percentage in one-score games, for instance, average out as 50/50 propositions and are generally pretty good indicators of how lucky a team was in the previous season. So now that we have that out of the way, I'll go ahead and post the previews. Last edited by rubecube; 08-28-2019 at 03:17 PM.