Tuesday, July 8, 2014

NFL 2014 Turnover Margin Analysis – It’s Impact on the Outcomes of NFL Games: Part I



If you are a statistical analyzer, and break down numbers on NFL games with any sort of regularity, you know that TO margin is an incredibly important statistic in the outcome of all games.  Straight up, ATS – doesn’t matter.  Perform well in this area, be it taking care of the ball, or taking the ball away from your opponent on a regular basis, and the likelihood you are playing football come January greatly increases.  Most NFL fans, even the casual fan that doesn’t get too involved in breaking stats down is aware of this phenomenon.  But how good a tool can it be to predict future outcomes, whether it be on a week to week basis trying to pick straight up winners in your office pool, picking a few games to wager your hard earned cash on, or even trying to predict the chances your team has at making the playoffs?  It’s very useful, and below I will discuss various angles and ways to utilize this one stat in your handicapping efforts.

Initially let’s examine the relationship between turnovers, points scored vs. points against (which is a team’s points margin), and projected records based on those two stats.

TEAM
POINTS
TURNOVER
NORM
ACTUAL
NEW PROJ
WIN
2+


MARGIN
ADV/(DIS)
MARGIN
RECORD
RECORD
VARIANCE
TMS

ARI
3.44
0.00
3.44
10-6
10-6



ATL
(5.63)
(1.25)
(4.38)
4-12
5-11
(1)


BAL
(2.00)
(1.25)
(0.75)
8-8
8-8



BUF
(3.00)
0.75
(3.75)
6-10
6-10



CAR
7.81
2.75
5.06
12-4
11-5
1


CHI
(2.06)
1.25
(3.31)
8-8
6-10
2
CHI

CIN
7.81
0.25
7.56
11-5
12-4
(1)


CLE
(6.13)
(2.00)
(4.13)
4-12
6-10
(2)
CLE

DAL
0.44
2.00
(1.56)
8-8
7-9
1


DEN
12.94
0.00
12.94
13-3
14-2
(1)


DET
1.19
(3.00)
4.19
7-9
10-6
(3)
DET

GB
(0.69)
(0.75)
0.06
8-7-1
8-8



HOU
(9.50)
(5.00)
(4.50)
2-14
5-11
(3)
HOU

IND
3.44
3.25
0.19
11-5
8-8
3
IND

JAC
(12.63)
(1.50)
(11.13)
4-12
3-13
1


KC
7.81
4.50
3.31
11-5
10-6
1


MIA
(1.13)
(0.50)
(0.63)
8-8
8-8



MIN
(5.56)
(3.00)
(2.56)
5-10-1
6-10



NE
6.63
1.75
4.88
12-4
11-5
1


NO
6.88
0.00
6.88
11-5
12-4
(1)


NYG
(5.56)
(3.75)
(1.81)
7-9
7-9



NYJ
(6.06)
(3.50)
(2.56)
8-8
6-10
2
NYJ

OAK
(8.19)
(2.00)
(6.19)
4-12
5-11
(1)


PHI
3.75
3.00
0.75
10-6
8-8
2
PHI

PIT
0.56
(1.00)
1.56
8-8
9-7
(1)


SD
3.00
(1.00)
4.00
9-7
10-6
(1)


SF
8.38
3.00
5.38
12-4
11-5
1


SEA
11.63
5.00
6.63
13-3
11-5
2
SEA

STL
(1.00)
1.75
(2.75)
7-9
6-10
1


TB
(6.31)
2.50
(8.81)
4-12
4-12



TEN
(1.19)
0.00
(1.19)
7-9
7-9



WAS
(9.00)
(2.25)
(6.75)
3-13
5-11
(2)
WAS











LEGEND:

Points Margin: points scored – points allowed for 2013 regular season

Turnover Adv/(Dis): represents the per game impact each team’s turnovers had on their points margin.  This number is derived by taking the total TO Margin on the season, dividing by 16 to get a “per game” TO Margin, then multiplying by 4 (or whatever value you choose to place on turnovers – discussed below)

Norm Margin: represents normalized margin & is calculated by taking Points Margin MINUS Turnover Adv/(Dis).  This figure hypothetically represents what a team’s point’s margin would have been stripping out the impact turnovers had on it

Actual Record: straight forward, each team’s true SU record for 2013 regular season

New Proj Record: uses the Normalized Margin calculation, and fits each into the Points Margin Pythagorean Theory matrix.  It is generally assumed that teams who on average outscore their opponents by 1.5ppg will go 9-7, 3ppg 10-6, 5.5ppg 11-5 and so on increasing ppg by 2.5pts for each win – and using the reciprocal of each of those marks for losing records.  Note, since these figures are quoted in decimals & rounded, the sum of 255 wins does not equal 256, the amount we see in an entire NFL season if there are no ties.

Win Variance: calculated by taking Actual Record MINUS New Projected Record.  Teams highlighted in red may have played worse than their record indicated last season (this analysis could certainly be used tying into my first post about +/- 4 wins along with betting future season win totals) while green teams would be bullish targets as far as this analysis goes.

Next big piece of this analysis is to explain how we valued the turnover.  Most analysts who work with turnovers in their models will value these at approximately being worth 4 points.  Of course this number is not set in stone, and can be debated & supported at various “point” impacts – but for this analysis I will be using 4 points.  In reality, any number you select within reason – the number has to be worth anywhere between a minimum 2pts and a maximum 5pts because a turnover either way leads to the possibility of scoring or allowing points – working with estimated % chances of scoring/allowing a FG/TD will allow you to derive your own worth of a turnover; so long as you have support & utilize a consistent value for all teams your analysis will be sound.  By using that method of “valuing” turnovers, we can calculate a new point’s margin based on a team’s pure play performance – stripping away the advantage/disadvantage turnovers had on their point’s margin.  This is a valuable way to place a barometer on how team’s truly performed, statistically speaking, in their games.  

Now that we have explained all the data, here is where it gets useful.  As mentioned, a “model” or any analysis is only good if you back-test it, and prove that it has worked in the past.  While any model may add value for a short period of time or even a year the ones that offer the best value & assistance in your handicapping efforts are those where you apply your theory for a minimum of 5 years back & check how significant its results are vs. actual results.  Especially in this day & age there are tons & tons of new statistics, analysts & bloggers publishing their work – but the biggest issue I see many have is information overload.  Sure most of the new statistics & theories can help you predict outcomes of sporting events but you should attempt at mastering a small data set & metrics, knowing how to utilize those the best you can to handicap games; you do not want to be a jack of all trades / master of none – too many times I read on Twitter handicappers using tons of different analytics & metrics every other night – there is a such thing as information overload, which is where many people go wrong.  Remember my old saying – give me either side of any game & I can give you a write-up supporting that play….

Back to this analysis, let’s first focus on the teams I have highlighted in red – teams that achieved a record in 2013 that was above and beyond their actual performance stripping the impact of turnovers.  These team’s we forecast to drop in wins from 2013 to 2014 because as we know, turnovers typically, but not always, revert back to the mean – so a team’s performance that was positively impacted by a strong TO margin the prior season often flips in the very next season.  Numerous articles have been posted on this topic to my blog over the years & this has proven to be a solid leading indicator for the following season, barring IMPACT signings or SIGNIFICANT free agency defections.  For the 2014 season here are some teams we would be bearish on, i.e. those that are likely to win less than they did last season:
  • CHI
  • IND
  • NYJ
  • PHI
  • SEA
Now let’s move onto the team’s that were negatively impacted by TO margin in 2013, which means we expect this group to have a stronger record in 2014 comparing to 2013:
  • CLE
  • DET
  • HOU
  • WAS
As I continually stress back-testing must be completed on any analysis to confirm its accuracy.  We have done that the last few years with this analysis, but for ease let’s examine below which teams we expected to slide in the 2013 season vs. the 2012 season (posted here last summer):
Team’s that were likely to see a drop in their 2013 record went 3-1-1 as I show games won in 2012 vs. 2013 in parenthesis:
  • ATL (13 to 4): CORRECT
  • HOU (12 to 2): CORRECT
  • IND (11 to 11): SAME
  • TEN (6 to 7): INCORRECT
  • WAS (10 to 3): CORRECT
Team’s that were likely to win at least one more game in 2013 vs. 2012 went 2-0-1 as change in wins is shown in parenthesis:
  • DET (4 to 7): CORRECT
  • KC (2 to 11): CORRECT
  • PIT (8 to 8): SAME
In summary, using TOM’s impact on NFL Pythagorean Theory, heading into the:
  • 2011 season there were 13 teams that were projected to slide up or down in wins – 10 moved the way we projected while 2 stayed the same; the only one that missed was SD who went from 9 to 8 wins & we forecasted a rise
  • 2012 season there were 12 teams that were projected to slide up or down in wins – 10 moved the way projected while 2 slid opposite (SD again & PHI both were projected to drop but increased their wins vs. 2011)
  • 2013 season there were 8 teams that were projected so slide up or down in wins – 5 moved the way projected while 2 stayed the same; the only one that missed was TEN who went from 6 to 7 & we forecasted a drop
  • SUMMARY: over last three seasons we have projected 33 teams to shift their wins either up or down – only FOUR of those THIRTY THREE moved the opposite way (just 12%).
Going into the 2014 season we have 9 teams projected to slide up or down in wins based on this initial look at TOM & Pythagorean Theory.  

That will conclude Part I of our Turnover Margin analysis.  Trust me when I say I have a lot more great analysis related to turnovers that will be discussed in the coming week’s right here on my blog.

Thanks for reading, just a few weeks till preseason football gets going!


Email me directly: boss@thesportsboss.com
Visit my website: www.thesportsboss.com
Follow me on Twitter: @SportsBoss
Follow me on Facebook: The SportsBoss
Leave comments here on the blog

COPYRIGHT: THE SPORTSBOSS, 2014

No comments:

Post a Comment