Using Net Yards Per Play to Determine Success

[bla bla bla]

This is a popular method in trying to determine projected winners against the spread. There is another column that should be added (Current Market Spread) on a week to week basis but since the regular season is over here is what the final results look like for 2017.

*Column Key*

2 - Yards Per Play Gained on Offense

3 - Yards Per Play Given up on Defense

4 - Net Yards Per Play

5 - Spread Value (Net YPP/.2) - for every .2 net yards the team gains 1 point towards the spread

6 - Home Field Advantage on a scale from 1-3. 3 being the strongest HFA and 1 being the weakest HFA. This was populated based on Home games won vs lost this season. Haven't had enough time to figure out the last 5 years but that will likely be my next step to determine future outcomes. Teams that won 6 or more games received a 3, teams that won 4 or 5 games received a 2, teams that won 3 or less received a 1.

7 - Strength of Schedule

 

According to these numbers the playoffs should have looked like this:

The teams highlighted in orange represent teams that made it into the playoffs. The Seahawks and Chargers would be the teams that were left out this year but Seattle sustained heavy injuries and the Chargers lost a 3 way tie to the Bills. Obviously not a perfect system but I think it does provide some interesting data. 

Other Interesting Points

- Buffalo was lucky to make it into the playoffs. They were tied for 2nd worst offensive YPP and but were one of the best in defensive YPP. More consistent offensive play would make them a very dangerous team.

- Carolina is a good team that finds ways to win close games. (7 of 11 wins were within 1 score)

- Only 3 teams made the playoffs who faced a top 10 SoS - Patriots (10th) - Bills (5th) - Chiefs (2nd)

- Average SoS for playoff teams was 17th - If you remove the 3 teams that had SoS in the top 10 then that average drops to 21st

- If Dallas had Elliot for those 6 games they'd likely be one of the best overall teams.

- The Patriots were tied for the worst in defensive YPP but their offense was tied for 3rd in offensive YPP. 9 of their 13 wins were by more than 7 points so the defensive YPP could be skewed due to garbage time yards.

- 7 of the bottom 10 teams either don't have franchise QBs or sustained major injuries to their teams. The other 3 QBs are Newton, Winston, and Flacco.

- Cleveland likely looks better than they actually were, most of their yards likely came in garbage time since a lot of their losses were by a pretty wide margin. 

- The AFC East played the 2nd hardest SoS in terms of division - Patriots (10th) Jets (8th) Dolphins (6th) Bills (5th) only division to play a harder schedule was the AFC West - Raiders (4th) Chargers (3rd) Chiefs (2nd) Broncos (1st). AFC West's tough schedule is likely due to the strength of the overall teams in that division. Outside of the Patriots the AFC East is a pretty easy division that got waxed by a hard schedule this year.

 

Divisional Round Predicted Outcomes:

Formula: Team 1 SV - [Team 2 SV + HFA] = Net YPP Spread + Current Market Spread

AFC: Titans (.5) - [Patriots (1) + Patriots HFA (3)] = 3.5 points in favor of the Patriots + CMS (Patriots -13) = Advantage Patriots by 16.5 Points (Won by 21)

        Jaguars (3.5) - [Steelers (2.5) + Steelers HFA (3)] = 2 points in favor of the Steelers + CMS (Steelers -7.5) = Advantage Steelers by 9.5 points (Anomaly) 

 

NFC: Falcons (3) - [Eagles (2) + Eagles HFA (3)] = 2 points in favor of the Eagles + CMS (Falcons -2.5) = Advantage Falcons by .5 points (Too Close to Call)

         Saints (4.5) - [Vikings (3.5) + Vikings HFA (3)] = 2 points in favor of the Vikings + CMS (Vikings -4) = Advantage Vikings by 6 points (Won by 5 but knelt the extra point)

 

I prefer to bet Winner vs Loser as opposed to spreads in the playoffs because the teams are much more evenly matched compared to Vegas odds but that is sort of how the formula would work. The one anomaly being the Steelers game but given the way that matchup went during the regular season I don't think I would have touched that game.

 

Championship Weekend Predicted Outcomes

Formula: Team 1 SV - [Team 2 SV + HFA] = Net YPP Spread + Current Market Spread

AFC: Jaguars SV (3.5) - [Patriots SV (1) + Patriots HFA (3)] = .5 points in favor of the Patriots + CMS (Patriots -8) = Advantage Patriots by 8.5 points

NFC: Vikings SV (3.5) - [Eagles SV (2) + Eagles HFA (3)] = 1.5 points in favor of the Eagles + CMS (Vikings -3) = Advantage Vikings by 1.5 points

 

If this is to be believed the Patriots should soundly beat the Jaguars and the Vikings should beat the Eagles. The advantage for the Vikings should be higher because most of this season's data was recorded with Wentz as the starting QB. This works well during the regular season, usually after week 4 to establish a current baseline, to find mismatched spreads to gamble on. I would steer clear from betting on either of the games this weekend since there is not a large discrepancy.

 

SB LII: Vikings vs Patriots

[SenorGato]

It’s a joke that The Rivers and Bosa aren’t in the playoffs this year. They’d blow it but still....

Anyone here use Jonathan Stone or Sam Katz? 

[nycdan]

I love this stuff.  Thanks for sharing it.   Few thoughts...

You alluded to the issue with your point about CLE but to make this more statistically accurate, I would look to split the first 3 quarters from the 4th quarter.  4th quarter play-calling is so situational that these types of metrics can get very skewed.  Teams with the lead run the ball to eat up clock, usually generating a lower YPP.  Meanwhile in the last two minutes those same teams are giving up more yardage in the middle of the field than usual.  You could argue that it averages out over time, but with only 16 games to work with, there are going to be some anomalies.  The Q1-Q3 split would be a true indicator of general performance.

Another piece here is penalties.  If a team takes a 10 yard penalty on first down and then gains 5 on the next play (still first down) is their gain calculated as +5 or -5?  That can be a huge determinant and I would argue that penalties matter.

Also to another of your points.  No doubt the performance in net YPP correlates very highly to playoff inclusion but how predictive is it early in the season.  It would be great to see these numbers in weeks 4, 8 and 12 for example to see how much they shifted over the course of the season.  

 

[bla bla bla]

53 minutes ago, nycdan said:

I love this stuff.  Thanks for sharing it.   Few thoughts...

You alluded to the issue with your point about CLE but to make this more statistically accurate, I would look to split the first 3 quarters from the 4th quarter.  4th quarter play-calling is so situational that these types of metrics can get very skewed.  Teams with the lead run the ball to eat up clock, usually generating a lower YPP.  Meanwhile in the last two minutes those same teams are giving up more yardage in the middle of the field than usual.  You could argue that it averages out over time, but with only 16 games to work with, there are going to be some anomalies.  The Q1-Q3 split would be a true indicator of general performance.

Yes I actually did think of that but since I did this retro actively I didn't feel like going through each individual quarter haha for the sake of time I was okay with take 75% of the seasons worth of stats and hoped that the 25% was a combo of close games and blow outs. I think the teams on either end of the strength spectrum would be affected the most (Patriots and Cleveland) because both teams are higher and lower than I expected.

57 minutes ago, nycdan said:

Another piece here is penalties.  If a team takes a 10 yard penalty on first down and then gains 5 on the next play (still first down) is their gain calculated as +5 or -5?  That can be a huge determinant and I would argue that penalties matter.

Penalties were not included in Net Yards for offense or defense because they did not effect the "per play" portion. I think that is absolutely a valuable stat that I could add in. You'd have to imagine that teams higher on the list would likely have less penalties than the teams lower down. I love the idea!

1 hour ago, nycdan said:

Also to another of your points.  No doubt the performance in net YPP correlates very highly to playoff inclusion but how predictive is it early in the season.  It would be great to see these numbers in weeks 4, 8 and 12 for example to see how much they shifted over the course of the season.  

Again a little too much time for me to go back and do it retro actively haha but yes I'd love to see a quarterly breakdown. This isn't really used to determin playoff teams, it's really meant to find discrepancies in Vegas Spreads vs Statistical Spreads. If a matchup is heavily lopsided then that would be a game you should look into betting on. This is really just the first step because you also need to take into account injuries, is the team coming off a bye or are they on short rest, etc.

My guess just looking at the chart is that Seattle started pretty hot and the Chargers started out cold. If either team had been consistent they likely make it into the playoffs. The Panthers are the one piece that has me slightly confused and I'm not sure how to interprete them.

[freestater]

Soooo....the pig is NOT ripe?

[Jets Voice of Reason]

4 hours ago, nycdan said:

I love this stuff.  Thanks for sharing it.   Few thoughts...

You alluded to the issue with your point about CLE but to make this more statistically accurate, I would look to split the first 3 quarters from the 4th quarter.  4th quarter play-calling is so situational that these types of metrics can get very skewed.  Teams with the lead run the ball to eat up clock, usually generating a lower YPP.  Meanwhile in the last two minutes those same teams are giving up more yardage in the middle of the field than usual.  You could argue that it averages out over time, but with only 16 games to work with, there are going to be some anomalies.  The Q1-Q3 split would be a true indicator of general performance.

Another piece here is penalties.  If a team takes a 10 yard penalty on first down and then gains 5 on the next play (still first down) is their gain calculated as +5 or -5?  That can be a huge determinant and I would argue that penalties matter.

Also to another of your points.  No doubt the performance in net YPP correlates very highly to playoff inclusion but how predictive is it early in the season.  It would be great to see these numbers in weeks 4, 8 and 12 for example to see how much they shifted over the course of the season.  

 

Very good points regarding the sample size problem with 4th quarter in 16 game samples and the penalties.

Special teams has been so marginalized in importance due to the rule changes that it puts such an importance on offensive and defensive performances.

I haven't calculated it, but it seems to me like the greater standard deviation in ypp offense or ypp defensive is in offense for the league as a whole, which would indicate to me that it's become way easier to be a good offense or bad offense than it is to break the mold and become a great defensive based team. 

I.e defenses are more or less performing about the same as each other but teams are focusing more on the offensive side of the ball for good reason. 

 

[bla bla bla]

3 hours ago, Jets Voice of Reason said:

Very good points regarding the sample size problem with 4th quarter in 16 game samples and the penalties.

Special teams has been so marginalized in importance due to the rule changes that it puts such an importance on offensive and defensive performances.

I haven't calculated it, but it seems to me like the greater standard deviation in ypp offense or ypp defensive is in offense for the league as a whole, which would indicate to me that it's become way easier to be a good offense or bad offense than it is to break the mold and become a great defensive based team. 

I.e defenses are more or less performing about the same as each other but teams are focusing more on the offensive side of the ball for good reason. 

Do you think that puts a premium on defensive players? Since the rules cater to the offensive side of the ball it would likely mean more offensive players become options because they in theory don't have to be as good. 

[JetFreak89]

Wow dude, did you put this together or take it from a site? This is seriously interesting stuff. I would love to see this done on a weekly basis of the Jets vs their opponent. 

[#27TheDominator]

Interesting stuff.  The one thing that I think the ypp portion doesn't take into account is chunk or big plays.  I am pretty sure that the Bills lead the league in big plays in 2016, I'm not sure about 2017.  Teams with a higher percentage of big plays may have more scoring per yard. An example is that a team that hits on a 50 yard pass on 3rd down and has 3/3 and outs may have only 50 yards/12 plays (4.16). While another team might seem significantly better in ypp but only have two long FG drives.  YPP is an easier metric of dominance, but the big plays can put that out the window somewhat.  That is where I think these QBs that can move may be more likely to be  outliers.

[rangerous]

18 hours ago, bla bla bla said:

This is a popular method in trying to determine projected winners against the spread. There is another column that should be added (Current Market Spread) on a week to week basis but since the regular season is over here is what the final results look like for 2017.

*Column Key*

2 - Yards Per Play Gained on Offense

3 - Yards Per Play Given up on Defense

4 - Net Yards Per Play

5 - Spread Value (Net YPP/.2) - for every .2 net yards the team gains 1 point towards the spread

6 - Home Field Advantage on a scale from 1-3. 3 being the strongest HFA and 1 being the weakest HFA. This was populated based on Home games won vs lost this season. Haven't had enough time to figure out the last 5 years but that will likely be my next step to determine future outcomes. Teams that won 6 or more games received a 3, teams that won 4 or 5 games received a 2, teams that won 3 or less received a 1.

7 - Strength of Schedule

 

According to these numbers the playoffs should have looked like this:

The teams highlighted in orange represent teams that made it into the playoffs. The Seahawks and Chargers would be the teams that were left out this year but Seattle sustained heavy injuries and the Chargers lost a 3 way tie to the Bills. Obviously not a perfect system but I think it does provide some interesting data. 

Other Interesting Points

- Buffalo was lucky to make it into the playoffs. They were tied for 2nd worst offensive YPP and but were one of the best in defensive YPP. More consistent offensive play would make them a very dangerous team.

- Carolina is a good team that finds ways to win close games. (7 of 11 wins were within 1 score)

- Only 3 teams made the playoffs who faced a top 10 SoS - Patriots (10th) - Bills (5th) - Chiefs (2nd)

- Average SoS for playoff teams was 17th - If you remove the 3 teams that had SoS in the top 10 then that average drops to 21st

- If Dallas had Elliot for those 6 games they'd likely be one of the best overall teams.

- The Patriots were tied for the worst in defensive YPP but their offense was tied for 3rd in offensive YPP. 9 of their 13 wins were by more than 7 points so the defensive YPP could be skewed due to garbage time yards.

- 7 of the bottom 10 teams either don't have franchise QBs or sustained major injuries to their teams. The other 3 QBs are Newton, Winston, and Flacco.

- Cleveland likely looks better than they actually were, most of their yards likely came in garbage time since a lot of their losses were by a pretty wide margin. 

- The AFC East played the 2nd hardest SoS in terms of division - Patriots (10th) Jets (8th) Dolphins (6th) Bills (5th) only division to play a harder schedule was the AFC West - Raiders (4th) Chargers (3rd) Chiefs (2nd) Broncos (1st). AFC West's tough schedule is likely due to the strength of the overall teams in that division. Outside of the Patriots the AFC East is a pretty easy division that got waxed by a hard schedule this year.

 

Divisional Round Predicted Outcomes:

Formula: Team 1 SV - [Team 2 SV + HFA] = Net YPP Spread + Current Market Spread

AFC: Titans (.5) - [Patriots (1) + Patriots HFA (3)] = 3.5 points in favor of the Patriots + CMS (Patriots -13) = Advantage Patriots by 16.5 Points (Won by 21)

        Jaguars (3.5) - [Steelers (2.5) + Steelers HFA (3)] = 2 points in favor of the Steelers + CMS (Steelers -7.5) = Advantage Steelers by 9.5 points (Anomaly) 

 

NFC: Falcons (3) - [Eagles (2) + Eagles HFA (3)] = 2 points in favor of the Eagles + CMS (Falcons -2.5) = Advantage Falcons by .5 points (Too Close to Call)

         Saints (4.5) - [Vikings (3.5) + Vikings HFA (3)] = 2 points in favor of the Vikings + CMS (Vikings -4) = Advantage Vikings by 6 points (Won by 5 but knelt the extra point)

 

I prefer to bet Winner vs Loser as opposed to spreads in the playoffs because the teams are much more evenly matched compared to Vegas odds but that is sort of how the formula would work. The one anomaly being the Steelers game but given the way that matchup went during the regular season I don't think I would have touched that game.

 

Championship Weekend Predicted Outcomes

Formula: Team 1 SV - [Team 2 SV + HFA] = Net YPP Spread + Current Market Spread

AFC: Jaguars SV (3.5) - [Patriots SV (1) + Patriots HFA (3)] = .5 points in favor of the Patriots + CMS (Patriots -8) = Advantage Patriots by 8.5 points

NFC: Vikings SV (3.5) - [Eagles SV (2) + Eagles HFA (3)] = 1.5 points in favor of the Eagles + CMS (Vikings -3) = Advantage Vikings by 1.5 points

 

If this is to be believed the Patriots should soundly beat the Jaguars and the Vikings should beat the Eagles. The advantage for the Vikings should be higher because most of this season's data was recorded with Wentz as the starting QB. This works well during the regular season, usually after week 4 to establish a current baseline, to find mismatched spreads to gamble on. I would steer clear from betting on either of the games this weekend since there is not a large discrepancy.

 

SB LII: Vikings vs Patriots

it's got merits but i also think the turnover ratio has to be factored in.  the jets were -4 on the season.  plus 4 and maybe they win 2 more games.  and then look at the patsies with their measily 0.2.

[bla bla bla]

32 minutes ago, rangerous said:

it's got merits but i also think the turnover ratio has to be factored in.  the jets were -4 on the season.  plus 4 and maybe they win 2 more games.  and then look at the patsies with their measily 0.2.

Completely agree, however TOs are largely luck. A bounce or tipped ball is harder to predict than the average yeads on every play. When trying to predict if a team will cover the spread you can't really bank on them getting a turnover prior to the game consistently.

That would be a great way to see which teams had success overall and maybe that differentiates a team winning with skill vs a team winning with luck.

[Jets Voice of Reason]

4 hours ago, bla bla bla said:

Do you think that puts a premium on defensive players? Since the rules cater to the offensive side of the ball it would likely mean more offensive players become options because they in theory don't have to be as good. 

I think certainly it does and the market tends to dictate so as well. How often do you see shutdown corners or elite pass rushers hit the market since the post tv deal cap explosion. it's both high demand and low supply there, Which results in overpays for mediocre defensive talent in the FA market as well. Especially when cap space was easier to generate since contracts hadn't yet adjusted to the new levels. 

It's really hard finding a defensive scheme that works that has enough good players to execute it. An elite passrusher and a lockdown corner can still be beaten. You need lots of good defensive players who can excel in these rules. They simply don't exist en masse. It's really not a surprise that teams like the patriots and Steelers went from defensive juggernauts to high powered offenses with questionable defenses. They understand how to build teams.