2021 NFL Midseason Predictions

Part I: Using Variance to target the best NFL Team Futures Bets

NFL Football seasons are 16-17 weeks long. With so few games (samples), along with the natural chaos of a contest between 22 people within 19,200 square yards, outcomes don’t always correlate to team performance. In other words, the “best team doesn’t always win”. Rather, variance/luck (or lack thereof) come into play and tend to prop up players/teams, making them seem more productive than they are, when some/much of their success can be attributed to (unstable) luck, and vice-versa. Knowing this, we can take a select group of variables we know to be “not-sticky” and/or highly volatile throughout the season to identify teams that:

  1. Have gotten lucky in said variables falling their way and are likely to see a (negative) regression towards the mean from those same variables
  2. Have gotten unlucky in said variables falling their way and are likely to see a (positive) regression towards the mean from those same variables

Armed with this information we can make sound predictions likely to “go up or down”, and lever to make profitable NFL Team Win Futures Bets. Below you will find the individual metrics, along with how each team ranked, followed by an aggregate board that we will use in conjunction with current win totals to target opportunities.

*All data up to data as of 10/28/21
**All charts include NFL teams identified with their appropriate betting slip abbv.
***Rank (Rk) is based on team that has been least lucky (#1) to most lucky (#32), and expected to regress towards the mean

Change in Strength of Schedule

TeamPastScheduleFutureScheduleNetRk
ARI160.10%84.30%823
ATL31-11.70%94.30%2231
BAL21-2.30%26.50%1929
BUF32-12.10%32-11.80%018
CAR27-6.60%44.70%2332
CHI111.70%35.50%-217
CIN25-4.50%170.60%824
CLE19-1.90%54.50%1426
DAL18-1.00%24-3.30%-611
DEN28-7.70%19-0.70%925
DET29.90%141.50%-127
GB20-2.00%19.20%1930
HOU67.30%27-4.80%-212
IND150.20%23-2.90%-89
JAX24-4.20%20-1.10%421
KC76.90%161.10%-98
LAC57.30%26-4.50%-211
LAR104.30%122.20%-215
LV26-5.90%112.30%1527
MIA121.80%30-8.00%-183
MIN85.40%64.40%219
NE29-9.50%22-1.40%722
NO23-3.80%25-4.00%-216
NYG17-0.50%21-1.10%-414
NYJ30-9.80%131.90%1728
PHI140.50%31-8.50%-175
PIT39.80%74.30%-413
SEA49.70%180.40%-146
SF94.80%151.30%-612
TB22-2.80%29-6.90%-710
TEN112.20%28-5.40%-174
WAS131.80%103.10%320

*The 2nd column, as a percentage is Football Outsiders’ Opponent DVOA
**Net reflects the change in previous games played strength of schedule vs upcoming strength of schedule

Table Above Explained

Using Football Outsiders flagship-stat-based metric DVOA to determine the true value of an opponent, then ranked and compared between previous and upcoming schedule, we find teams that have benefited most/will face the toughest challenges moving forward.

Fumble Recovery Luck

Team2021202020 to 21 change2021 Rk
ARI66.67%52.63%14.04%26
ATL25.00%42.86%-17.86%9
BAL40.00%35.29%4.71%16
BUF60.00%41.38%18.62%24
CAR18.18%68.18%-50.00%3
CHI66.67%50.00%16.67%27
CIN20.00%42.86%-22.86%5
CLE22.22%45.83%-23.61%6
DAL75.00%61.90%13.10%29
DEN66.67%37.50%29.17%28
DET50.00%29.41%20.59%20
GB40.00%46.67%-6.67%15
HOU18.18%54.55%-36.37%4
IND75.00%50.00%25.00%30
JAX0.00%29.41%-29.41%1
KC28.57%38.89%-10.32%11
LAC27.27%46.67%-19.40%10
LAR33.33%50.00%-16.67%14
LV41.67%50.00%-8.33%17
MIA54.55%55.00%-0.45%21
MIN75.00%41.18%33.82%31
NE75.00%36.36%38.64%32
NO25.00%44.44%-19.44%8
NYG60.00%42.31%17.69%23
NYJ57.14%40.91%16.23%22
PHI16.67%52.38%-35.71%2
PIT30.00%50.00%-20.00%12
SEA62.50%44.44%18.06%25
SF23.08%36.36%-13.28%7
TB44.44%50.00%-5.56%18
TEN30.00%38.10%-8.10%13
WAS50.00%34.78%15.22%19

*2021 is the current team fumble recovery rate
**Included 2020 as context to see how easily this metric fluctuates

Table Above Explained

Forcing a fumble, and holding on to the ball takes skill. However, once the ball is on the ground there is almost no skill involved, and significant recovery rates on both sides of the ball tend to move towards 50% overtime. Hence teams that have a high recovery rate in year 1, will likely not have the same luck.

Pythagorean Wins

TeamWLPFPAEXP WINSEXP – Real WinsE – RW Rk
ARI702251144.62.432
ATL331351762.60.420
BAL521871643.71.326
BUF42203984.00.017
CAR341461463.5-0.514
CHI341011622.70.319
CIN521891284.20.823
CLE431731653.60.421
DAL512051463.51.528
DEN341401273.7-0.713
DET071282002.7-2.71
GB611681463.72.331
HOU16972032.3-1.35
IND341691493.7-0.712
JAX151161722.4-1.43
KC341882033.4-0.415
LAC421481503.01.024
LAR612071464.11.930
LV521801663.61.427
MIA161272072.7-1.72
MIN331471373.1-0.116
NE341791403.9-0.98
NO421401013.50.522
NYG251391803.1-1.17
NYJ15801751.9-0.99
PHI251591853.2-1.26
PIT331171322.80.218
SEA251501623.4-1.44
SF241351492.9-0.911
TB612331474.31.729
TEN521931643.81.225
WAS251462102.9-0.910

*PF – Points scored
PA – Points allowed
EXP WINS – (PF / (PF + PA) * Games Played). Based on how many points a team “owns” relative to the total in a given game
EXP – Real Wins – Net Value of Pythagorean wins to actual wins

Table Above Explained

Pythagorean Wins is really just a way of looking at one team, how many total points were scored in games they played (their points AND their opponents’ points) and what the ratio of those points are “owned” by said team to get an idea of how many games they should have won. Admittedly, there is a bit of circular logic between this variable and some others, like YPP for example. That is, the amount of yards it takes for a team to get its points is, in itself, highly correlated to Pythagorean Wins. Hence there is some circular logic to this one, yet since it’s one of the “original variance predictors” so we had to include it. This is also the reason we did NOT include “Net Close Games Won”, which is all but fully addressed by this metric.

Yards Per Point (YPP)

OffeneDefense
TEAMYDSPTSoYPPRkYDSPTSdYPPRk
ARI2,81522512.5312,21711419.42
ATL2,12513515.7132,17617612.432
BAL2,92318715.6161,6219816.59
BUF2,46920312.2322,15214614.719
CAR2,29614615.7142,39316214.818
CHI1,78810117.762,37912818.63
CIN2,58718913.7262,06916512.531
CLE2,75417315.9122,28714615.716
DAL2,76520513.5272,26412717.86
DEN2,43314017.472,68120013.429
DET2,34112818.342,32014615.913
GB2,36516814.1232,74420313.528
HOU1,9269719.922,47914916.68
IND2,50016914.8222,47317214.421
JAX2,13611618.432,83220314.024
KC2,93518815.6172,18615014.620
LAC2,26514815.3192,61714617.94
LAR2,78020713.4282,47816614.917
LV2,75318015.3202,90420714.023
MIA2,15312717.082,15013715.715
MIN2,48514716.992,45414017.57
NE2,48517913.9241,98910119.71
NO1,78014012.7302,58118014.322
NYG2,46613917.752,41517513.825
NYJ1,6348020.412,52518513.626
PHI2,43415915.3182,11413216.012
PIT1,94311716.6112,90316217.95
SEA2,28315015.2211,94414913.030
SF2,12113515.7152,32514715.814
TB2,96423312.7292,63816416.111
TEN2,67719313.9252,67616416.310
WAS2,44714616.8102,84221013.527

YDS – Yards

PTS – Points

oYPP – Offensive Yards / Points
dYPP – Defensive Yards / Points

Table Above Explained

YPP is simply a look into how well yards, which correlate to wins from one year to the next better than wins themselves, were turned into points. Offenses with a lot of yards, but little points to show for it tend to score more the next year (relative to their yards). Defenses that gave up a lot of points, but few yards trend upward as well (and the opposite for both).

“3rd Down Rebound”

Offensive DownsDefensive Downs
Team1st/2nd3rdNETRk1st/2nd3rdNETRk
ARI171163110.519.528
ATL2716112626.529-2.513
BAL431-2712822632
BUF11.512-0.5151.54-2.514
CAR2729-21316.597.526
CHI19.532-12.5314.577.527
CIN19.5712.5287.521-13.54
CLE718-1164.531-26.51
DAL415-11716.5106.523
DEN18108241530-153
DET2130-9918.526-7.57
GB1028252114725
HOU31.51417.53212.525-12.55
IND181172220.5317.530
JAX16.526-9.5829.5272.520
KC713-61129.5236.524
LAC16511272161529
LAR532189.581.518
LV1521-6121518-312
MIA27.5207.5231728-116
MIN1495201413117
NE15.517-1.51415.5123.521
NO2161529511-610
NYG27.5225.5211917219
NYJ28.5253.51921.524-2.515
PHI20191172122-116
PIT23815301219-79
SEA1027-17226.5521.531
SF1423-91012.520-7.58
TB4.540.51610.515-4.511
TEN1224-12421.5165.522
WAS1628-1251432-182

1st/2nd – Team DVOA ranking on 1st and 2nd down
3rd – Team DVOA ranking on 3rd down 

Net – Difference between 1st and 2nd down DVOA and 3rd down DVOA

Table Above Explained

“3rd Down Rebound” is a Football Outsiders original, and highly predictive from year 1 to year 2. You can learn all about it in their article Stat of the day: Third-Down Rebound Effect. Essentially, we can glean some tendency towards the mean when teams are rather lucky/unlucky on 3rd down relative to their performance on 1st/2nd downs.

“RedZone Rebound”

Offensive Field Zone SplitsDefensive Field Zone Splits
TeamNon RZRed ZoneNet FZRkNon RZRed ZoneNet FZRk
ARI1219-7.5949-512
ATL241113.25272331-7.759
BAL1468.252223121.532
BUF1517-2.251635-2.516
CAR261214301526-11.256
CHI271413.252817610.527
CIN16511261215-3.2514
CLE1123-12.2551420-6.2510
DAL527-21.7511221-9.57
DEN1425-11.562024-4.2513
DET1829-10.7571932-134
GB915-61219181.2519
HOU29280.751818107.7524
IND1231-18.7521324-11.55
JAX2226-4.5132928118
KC822-14.25327224.7521
LAC1518-3151423-8.758
LAR1045.52014212.2529
LV1524-9.581730-13.253
MIA2020-0.51720119.2525
MIN1678.75231419-5.511
NE14104191587.2523
NO22219.532936.2522
NYG2330-7.5101529-13.752
NYJ2416821241410.2526
PHI18810251717017
PIT221392416412.2530
SEA161153118134.520
SF17313.5291316-315
TB69-3.2514927-18.51
TEN1521-6.51123121128
WAS1832-14425718.2531

Non RZ – DVOA rank in all Field Zones outside of the Red Zone (or 20 yards from the endzone)

RedZone – DVOA rank in the Red Zone
Net FZ – Difference in DVOA ranks between the Red Zone and the NON Red Zone field areas


Table Above Explained

“Redzone Rebound” is the exact same “play” as “3rd down Rebound”, with a different set of parameters. Just like 3rd downs, the Redzone is a highly leveraged area of the field. Teams tend to perform the same no matter where they are on the field OVER THE LONG-RUN. So when teams are outperforming the last 20 yards of the field relative to the first 80, our impression of their performance is likely skewed, and will eventually return to the team norm.

Opponent Field Goal Percentage

Team2021202020 vs 21 NETRk
ARI66.67%83.33%16.66%30
ATL85.71%85.71%0.00%17
BAL77.78%65.52%-12.26%22
BUF100.00%83.87%-16.13%4
CAR83.33%93.55%10.22%18
CHI90.91%80.56%-10.35%10
CIN71.43%86.84%15.41%25
CLE87.50%84.38%-3.12%15
DAL66.67%88.89%22.22%29
DEN66.67%92.86%26.19%28
DET83.33%80.00%-3.33%20
GB57.14%89.66%32.52%32
HOU92.31%80.65%-11.66%8
IND100.00%84.00%-16.00%3
JAX78.57%79.49%0.92%21
KC90.91%75.00%-15.91%11
LAC83.33%87.50%4.17%19
LAR86.67%86.67%0.00%16
LV88.89%82.05%-6.84%12
MIA88.24%73.91%-14.33%14
MIN92.31%89.74%-2.57%7
NE61.11%81.48%20.37%31
NO73.33%86.67%13.34%24
NYG91.67%88.89%-2.78%9
NYJ94.44%82.86%-11.58%5
PHI100.00%84.38%-15.62%2
PIT100.00%84.62%-15.38%1
SEA88.24%84.21%-4.03%13
SF70.00%89.19%19.19%26
TB69.23%85.29%16.06%27
TEN73.33%88.46%15.13%23
WAS92.86%90.32%-2.54%6

*2021 is the current opponent FG Conversion Rate
**Included 2020 as context to see how easily this metric fluctuates


Table Above Explained

Both the table above, and the next one below fall under the  “out of their control” type predictors. That is, things that happen in a game a team benefits (or not) from, that they had no control over. For the most part (sans drawing DPIs, offsides), the opponent’s success in these fields have nothing to do with the team in question. Hence, having marginal success with them is not a stable item a team can depend on continuing.

Opponent Penalties Committed vs Team

Team2021202020 vs 21 netRk
ARI6.16.5-0.415
ATL8.35.72.631
BAL5.14.50.64
BUF7.25.2227
CAR7.95.32.630
CHI4.75.8-1.13
CIN6.96.20.723
CLE5.65.30.310
DAL8.75.7332
DEN76.60.426
DET5.76.4-0.712
GB5.34.70.68
HOU5.74.80.911
IND5.35.20.19
JAX74.92.124
KC7.45.71.729
LAC6.35.90.418
LAR5.150.15
LV6.15.90.214
MIA6.75.31.422
MIN4.75.2-0.52
NE6.75.11.621
NO4.74.40.31
NYG6.45.50.919
NYJ66.2-0.213
PHI7.370.328
PIT6.56.20.320
SEA5.15.6-0.56
SF6.24.91.316
TB5.15.8-0.77
TEN76.40.625
WAS6.35.21.117

*2021 is the current opponent Penalty Rate
**Included 2020 as context to see how easily this metric fluctuates


See explanation ABOVE.

Now to put all variables together, evenly weighted, here is how these items in aggregate shape up:

Aggregate Scoreboard

TeamSOSFumbPWToYPPdYPPo3DRd3DRoRZRdRZRFGPenAvgChange RkWin %PAST Win % Rk
ARI2326323123128912301521.731100%1
ATL3192013322613279173120.72950%16
BAL291626169132223222419.02571%8
BUF18241732191514161642718.42267%11
CAR3231414181326306183018.52443%21
CHI17271963327282710315.51243%18
CIN2452326312842614252320.83071%9
CLE26621121661510151011.6357%13
DAL11292827672317293218.22183%5
DEN252813729243613282618.42343%20
DET7201413977420129.510%32
GB30153123282525121932822.53286%4
HOU24528325182481110.8214%30
IND9301222212230253915.01143%17
JAX21133248201318212414.2717%28
KC8111517201124321112915.51343%19
LAC110241942729158191815.81667%10
LAR15143028171818202916519.12686%3
LV2717272023121283121415.91771%6
MIA32128152361725142214.2814%31
MIN19311697201723117214.7950%14
NE2232824114211923312119.62843%22
NO1682230222910322224119.62767%12
NYG14237525211910291914.0629%26
NYJ282291261915212651316.81817%29
PHI52618121716251722813.5429%24
PIT131218115309243012015.71550%15
SEA62542130231312013617.21929%27
SF1271115141082915261614.81033%23
TB1018292911161114127715.71486%2
TEN4132525104221128232517.32071%7
WAS20191010275243161713.7529%25

Biggest Takeaways

One of the issues with these types of studies is they tend to result in “bad teams will get better, and good teams will get worse” situations. Because we know this is a natural phenomenon, we should highlight most, the teams that are still relatively high or low on the list, yet aren’t coming in as simply a good/bad team respectively.

  • The Browns jump out here, regardless of Baker Mayfield’s injury situation. They have a good record, yet have still been very unlucky (especially in high leverage situations)
  • Minnesota in in a similar situation, with a much healthier team to boot
  • On the other end of the spectrum, the Falcons are an interesting team. Especially given how many halftime leads they gave up last year, there record has been poor, despite being a “lucky” team over the first half of the season


Finally, tying this information in with the current market odds for NFL Team Futures, you can see appropriate opportunities below given their “expected variance changes” along with the betting win totals below:

NFL Win Totals + Avg Variance/Luck Rank

Win TotalAvgChange RkWin %WLWins to TotalRemaining Win %Win % Needed vs Cur Win %
ARI13.521.731100%706.565%-35%
ATL7.520.72950%334.541%-9%
BAL11.519.02571%526.565%-6%
BUF12.518.42267%428.577%11%
CAR6.518.52443%343.535%-8%
CHI6.515.51243%343.535%-8%
CIN10.520.83071%525.555%-16%
CLE10.511.6357%436.565%8%
DAL11.518.22183%516.559%-24%
DEN7.518.42343%344.545%2%
DET2.59.510%072.525%25%
GB12.522.53286%616.565%-21%
HOU3.510.8214%162.525%11%
IND8.515.01143%345.555%12%
JAX4.514.2717%153.532%15%
KC9.515.51343%346.565%22%
LAC10.515.81667%426.559%-8%
LAR12.519.12686%616.565%-21%
LV9.515.91771%524.545%-26%
MIA6.514.2814%165.555%41%
MIN8.514.7950%335.550%0%
NE7.519.62843%344.545%2%
NO9.519.62767%425.550%-17%
NYG5.514.0629%253.535%6%
NYJ3.516.81817%152.523%6%
PHI7.513.5429%255.555%26%
PIT7.515.71550%334.541%-9%
SEA6.517.21929%254.545%16%
SF8.514.81033%246.559%26%
TB13.515.71486%617.575%-11%
TEN11.517.32071%526.565%-6%
WAS5.513.7529%253.535%6%

Win Total – Current Vegas Odds
Avg – Aggregate Average of all Variables considered, ranked in order of teams (1) most likely to have better luck in the second half of the season
Wins to Total – How many wins a team needs to hit their Betting Line number

Remaining Win % – what percentage of games does a team need to win to hit the Betting Line number

Win % Needed vs Cur Win % – Win percentage needed – current winning percentage. This is used to give a quick depiction of how the market thinks win likelihood will change vs. how a team has performed thus far

We feel most strongly about the Vikings and Browns OVER plays, but use the data as you see fit. I simply wanted to give you information to make better, more profitable EV decisions on Futures Bets. This was Part I of II. Next week we will dive into the micro and present “Using Variance to target the best NFL Player Futures Bets”.

Leave a Reply