One of the ESPN folks posted FPI odds today, retweeted by Ben Alamar. The numbers are very different from my playoff formulas. The nature of those odds made me suspect that FPI is intrinsically an offensive stat, with the advantages and disadvantages of such a stat.

One of the issues I’ve has with offensive stats is that the confidence interval of any I’ve looked at, in terms of predicting playoff performance, is that those confidence intervals are on the order of 85%. Whatever flaws of my formulas, they fit to confidence intervals of 95%. The effects they touch on are real.

But still, the purpose of this is to compare FPI odds to the odds generated by some common offensive stats. We’re using Pythagorean expectation, as generated by my Perl code, SRS as generated by my Perl code, and median point spread, also calculated by my code.

Results are below.

FPI Odds versus Other Offensive Stats
Game FPI Pythag Simple Ranking Median Pt Spread
Kansas City – Tennessee 0.75 0.75 0.79 0.73
Jacksonville – Buffalo 0.82 0.89 0.86 0.73
Los Angeles – Atlanta 0.62 0.75 0.74 0.68
New Orleans – Carolina 0.70 0.73 0.74 0.78

 

The numbers correlate too well for FPI not to have a large offensive component in its character. In fact, Pythagorean odds correlate so well with FPI I’m hard pressed to know what advantages FPI gives to the generic fan.

Note: the SRS link above points out that PFR has added a home field advantage component to their SRS calcs. I’ll note that our SRS was calibrated against PFR’s pre 2015 formula.

 

Advertisements

This would have been done earlier, but Pro Football Reference dropped its very handy chart of draft position versus AV. I started missing it more and more, and using the Wayback Machine I found it here.

The three major QB trades of 2017 were the trade for Mitch Trubisky, Patrick Mahomes, and Deshaun Watson. We will analyze them in sequence.

Mitchell Trubisky Trade
Chicago Bears 49ers Results
Pick Average AV Pick Average AV Delta AV Risk Ratio
2 46 3 45
67 21
111 13
(71) 13
Total 46 Total 92
46 2.00

 

The Bears have a trade risk comparable to a typical trade for a #1 draft choice and a quarterback at that. The trade has less fundamental risk than Goff or Wentz. The comparable that comes to mind is Eli Manning. By contrast, the delta AV of the other two trades are substantially less.

Patrick Mahomes Trade
Chiefs Bills Results
Pick Average AV Pick Average AV Delta AV Risk Ratio
10 41 27 25
91 20
(25) 24
Total 41 Total 69
28 1.68

 
Mahomes merely has to give six seven good years, and the trade ends up warranted. The issue in the case of Deshaun Watson is keeping him upright. A fistful of whole years almost as good as his freshman year in the NFL and he would end up bordering on Hall of Fame numbers.

Deshaun Watson Trade
Texans Browns Results
Pick Average AV Pick Average AV Delta AV Risk Ratio
12 35 25 24
4 44
Total 35 Total 68
33 1.94

 

So here is wishing Deshaun Watson a healthy career from now on.

As I’m really away from home and short of time, I’ll mention my methodology is described in depth here.

The first table is the worksheet. Values will not change throughout the playoffs.

2017 NFL Playoff Teams, C&F Worksheet.
NFC
Rank Name Home Field Adv Playoff Experience SOS Total Score
1 Philadelphia Eagles 0.660 0 -0.216 0.444
2 Minnesota Vikings 0.660 0.747 0.301 1.708
3 Los Angeles Rams 0.660 0.0 -0.046 0.614
4 New Orleans Saints 0.660 0 0.471 1.131
5 Carolina Panthers 0.0 0.747 0.629 1.376
6 Atlanta Falcons 0.0 0.747 0.575 1.322
AFC
1 New England Patriots 0.660 0.747 -0.377 1.030
2 Pittsburgh Steelers 0.660 0.747 -0.334 1.073
3 Jacksonville Jaguars 0.660 0 -0.842 -0.182
4 Kansas City Chiefs 0.660 0.747 -0.404 1.003
5 Tennessee Titans 0.0 0.0 -0.644 -0.644
6 Buffalo Bills 0.0 0.0 -0.146 -0.146

 

LA is not favored by this model and neither are AFC teams. The NFC South’s toughness shows through in the SOS marks for this data set. Minnesota and/or NFC South Teams largely have advantages over almost any matchup the AFC can offer. Finally, an open Q is, is Drew Brees elite enough that his team should be granted the PPX bonus? For now I’d consider this a question to be answered later.

This second table shows odds for the first round, calculated for you. The only home team it favors is Kansas City.

First Round Playoff Odds
Home Team Visiting Team Score Diff Win Prob Est. Point Spread
Los Angeles Rams Atlanta Falcons -0.708 0.330 -5.2
New Orleans Saints Carolina Panthers -0.245 0.439 -1.8
Jacksonville Jaguars Buffalo Bills -0.036 0.491 -0.3
Kansas City Chiefs Tennessee Titans 1.647 0.838 12.1

Last set of regular data for the year. Going to try and grind out playoff stats before I head off for the first. Atlanta takes the final playoff spot in the NFC, and the Bills and Titans make the playoffs in the AFC.

Global Statistics:
Games  Home Wins HwPct Winning_Score Losing_Score Margin
256       145     56.6      27.63        15.81     11.81

Calculated Pythagorean Exponent:  2.69


Rank  Team    Median  GP   W   L   T  Pct   Pred   SRS    MOV   SOS
------------------------------------------------------------------------
1     NE      12.0    16  13   3   0  81.3  76.4   8.89  10.13 -1.24
2     MIN      9.0    16  13   3   0  81.3  75.4   9.12   8.13  0.99
3     PHI      8.5    16  13   3   0  81.3  76.4   9.41  10.13 -0.71
4     PIT      3.5    16  13   3   0  81.3  67.8   5.02   6.13 -1.10
5     NO       9.5    16  11   5   0  68.8  70.2   9.17   7.63  1.55
6     LA       5.5    16  11   5   0  68.8  73.2   9.16   9.31 -0.15
7     CAR      3.0    16  11   5   0  68.8  57.0   4.32   2.25  2.07
8     JAX      9.0    16  10   6   0  62.5  76.7   6.54   9.31 -2.77
9     KC       7.5    16  10   6   0  62.5  63.3   3.42   4.75 -1.33
10    ATL      3.0    16  10   6   0  62.5  57.6   4.26   2.38  1.89
11    BAL      7.0    16   9   7   0  56.3  67.1   3.40   5.75 -2.35
12    DAL      4.5    16   9   7   0  56.3  54.3   1.60   1.38  0.22
13    BUF      4.5    16   9   7   0  56.3  38.6  -4.04  -3.56 -0.48
14    DET      3.0    16   9   7   0  56.3  55.8   2.72   2.13  0.60
15    SEA      3.0    16   9   7   0  56.3  56.5   1.89   2.13 -0.24
16    LAC      3.0    16   9   7   0  56.3  67.2   3.64   5.19 -1.54
17    TEN      3.0    16   9   7   0  56.3  45.7  -3.50  -1.38 -2.12
18    ARI     -1.5    16   8   8   0  50.0  36.7  -3.72  -4.13  0.41
19    CIN     -3.0    16   7   9   0  43.8  37.8  -4.98  -3.69 -1.29
20    GB      -5.0    16   7   9   0  43.8  38.0  -1.90  -4.00  2.10
21    WAS     -5.5    16   7   9   0  43.8  41.6  -1.30  -2.88  1.58
22    SF      -2.5    16   6  10   0  37.5  40.3  -2.85  -3.25  0.40
23    OAK     -4.5    16   6  10   0  37.5  36.0  -4.73  -4.50 -0.23
24    MIA     -7.0    16   6  10   0  37.5  28.9  -6.26  -7.00  0.74
25    TB      -3.0    16   5  11   0  31.3  41.3  -1.28  -2.94  1.66
26    CHI     -4.5    16   5  11   0  31.3  37.3  -1.29  -3.50  2.21
27    NYJ     -6.0    16   5  11   0  31.3  33.9  -4.95  -5.25  0.30
28    DEN     -8.5    16   5  11   0  31.3  32.1  -6.73  -5.81 -0.91
29    IND     -5.0    16   4  12   0  25.0  24.0 -10.10  -8.81 -1.29
30    HOU     -7.5    16   4  12   0  25.0  33.5  -6.43  -6.13 -0.30
31    NYG     -8.5    16   3  13   0  18.8  22.7  -7.57  -8.88  1.31
32    CLE    -13.0    16   0  16   0   0.0  18.1 -10.95 -11.00  0.05

Playoffs become more and more decided. All division winners are decided but in the NFC South, where it will either be Carolina or New Orleans. Five of the six NFC slots are sewn up, with Atlanta and Seattle competing for the 6th playoff slot. In the AFC, all the division winners are decided, with the 5th and 6th playoff slots in play. Baltimore has over a 90% change to take one slot, according to Five Thirty Eight, with Tennessee at about 60%, and the Chargers and the Bills with marginal chances for a playoff bid.


Games  Home Wins HwPct Winning_Score Losing_Score Margin
240       135     56.2      27.82        15.90     11.92

Calculated Pythagorean Exponent:  2.79


Rank  Team    Median  GP   W   L   T  Pct   Pred   SRS    MOV   SOS
------------------------------------------------------------------------
1     PHI      9.0    15  13   2   0  86.7  78.2  10.32  11.20 -0.88
2     NE       8.0    15  12   3   0  80.0  75.2   8.59   9.47 -0.87
3     MIN      8.0    15  12   3   0  80.0  75.0   9.14   7.80  1.34
4     PIT      3.0    15  12   3   0  80.0  68.9   5.79   6.27 -0.48
5     NO      10.0    15  11   4   0  73.3  73.3  10.45   8.60  1.85
6     LA       6.0    15  11   4   0  73.3  78.0  11.30  11.33 -0.03
7     CAR      3.0    15  11   4   0  73.3  60.0   5.01   3.20  1.81
8     JAX     12.0    15  10   5   0  66.7  79.0   7.46  10.27 -2.81
9     KC       8.0    15   9   6   0  60.0  64.1   3.97   4.87 -0.90
10    BAL      7.0    15   9   6   0  60.0  69.9   4.23   6.40 -2.17
11    SEA      3.0    15   9   6   0  60.0  57.7   2.40   2.40 -0.00
12    ATL      3.0    15   9   6   0  60.0  55.7   3.49   1.73  1.76
13    DET      3.0    15   8   7   0  53.3  51.9   1.41   0.67  0.75
14    DAL      3.0    15   8   7   0  53.3  53.3   0.76   1.07 -0.31
15    TEN      3.0    15   8   7   0  53.3  44.4  -4.38  -1.80 -2.58
16    BUF      3.0    15   8   7   0  53.3  36.2  -4.37  -4.20 -0.17
17    LAC      1.0    15   8   7   0  53.3  64.6   2.93   4.20 -1.27
18    WAS     -3.0    15   7   8   0  46.7  42.5  -0.32  -2.53  2.22
19    GB      -3.0    15   7   8   0  46.7  41.6  -0.71  -2.67  1.96
20    ARI     -5.0    15   7   8   0  46.7  34.8  -4.34  -4.53  0.19
21    OAK     -3.0    15   6   9   0  40.0  38.7  -4.00  -3.47 -0.54
22    CIN     -3.0    15   6   9   0  40.0  35.3  -5.81  -4.20 -1.61
23    MIA     -8.0    15   6   9   0  40.0  28.1  -6.02  -7.07  1.04
24    SF      -3.0    15   5  10   0  33.3  35.1  -4.87  -4.87 -0.01
25    CHI     -3.0    15   5  10   0  33.3  39.3  -1.20  -2.87  1.66
26    NYJ     -5.0    15   5  10   0  33.3  36.5  -4.56  -4.27 -0.29
27    DEN    -10.0    15   5  10   0  33.3  30.7  -7.28  -6.00 -1.28
28    TB      -3.0    15   4  11   0  26.7  38.8  -2.48  -3.60  1.12
29    HOU     -7.0    15   4  11   0  26.7  33.7  -5.62  -5.93  0.32
30    IND     -6.0    15   3  12   0  20.0  20.6 -10.94 -10.00 -0.94
31    NYG    -10.0    15   2  13   0  13.3  19.6  -8.61 -10.00  1.39
32    CLE    -14.0    15   0  15   0   0.0  15.9 -11.72 -11.47 -0.26

Things begin to clear in the playoff picture. Atlanta wins, they are in. New England beats the Steelers, again. Aaron Rogers has no magical comeback. Derek Carr tries a little too much to win a game. Jacksonville still appears to be a major spoiler come playoff time.

Week 14 stats

Global Statistics:
Games  Home Wins HwPct Winning_Score Losing_Score Margin
208       115     55.3      27.94        16.16     11.78

Calculated Pythagorean Exponent:  2.63


Rank  Team    Median  GP   W   L   T  Pct   Pred   SRS    MOV   SOS
------------------------------------------------------------------------
1     PHI     10.0    13  11   2   0  84.6  77.9  11.51  11.85 -0.33
2     PIT      3.0    13  11   2   0  84.6  65.4   4.49   5.31 -0.82
3     NE       8.0    13  10   3   0  76.9  73.4   8.26   9.08 -0.81
4     MIN      8.0    13  10   3   0  76.9  67.2   7.70   5.69  2.01
5     JAX     12.0    13   9   4   0  69.2  78.3   7.51   9.77 -2.26
6     NO       9.0    13   9   4   0  69.2  71.0  10.33   8.23  2.10
7     LA       6.0    13   9   4   0  69.2  74.2  10.12  10.08  0.05
8     CAR      3.0    13   9   4   0  69.2  58.8   5.14   2.92  2.22
9     SEA      3.0    13   8   5   0  61.5  64.1   3.71   4.77 -1.06
10    ATL      3.0    13   8   5   0  61.5  57.8   4.04   2.54  1.50
11    TEN      3.0    13   8   5   0  61.5  45.1  -4.64  -1.62 -3.02
12    DAL     11.0    13   7   6   0  53.8  54.7   1.23   1.69 -0.46
13    KC       7.0    13   7   6   0  53.8  58.4   2.47   3.08 -0.60
14    BAL      7.0    13   7   6   0  53.8  66.2   4.83   5.54 -0.71
15    DET      3.0    13   7   6   0  53.8  51.8   1.83   0.69  1.14
16    GB       3.0    13   7   6   0  53.8  46.2  -0.10  -1.31  1.21
17    BUF      3.0    13   7   6   0  53.8  37.8  -4.37  -3.85 -0.52
18    LAC      1.0    13   7   6   0  53.8  67.7   4.26   5.62 -1.35
19    OAK     -1.0    13   6   7   0  46.2  40.8  -4.62  -3.08 -1.55
20    MIA     -3.0    13   6   7   0  46.2  31.4  -5.07  -6.31  1.24
21    ARI     -6.0    13   6   7   0  46.2  30.3  -5.75  -6.62  0.86
22    CIN     -3.0    13   5   8   0  38.5  38.3  -5.90  -3.46 -2.44
23    NYJ     -5.0    13   5   8   0  38.5  39.9  -4.78  -3.46 -1.32
24    WAS     -8.0    13   5   8   0  38.5  37.9  -1.12  -4.54  3.42
25    CHI     -3.0    13   4   9   0  30.8  37.1  -1.23  -3.85  2.62
26    TB      -5.0    13   4   9   0  30.8  39.2  -3.10  -3.69  0.59
27    HOU     -6.0    13   4   9   0  30.8  45.3  -2.74  -1.77 -0.97
28    DEN    -10.0    13   4   9   0  30.8  30.2  -7.05  -6.62 -0.44
29    SF      -3.0    13   3  10   0  23.1  30.1  -6.66  -6.62 -0.04
30    IND     -4.0    13   3  10   0  23.1  22.0 -11.02 -10.08 -0.95
31    NYG    -10.0    13   2  11   0  15.4  22.2  -8.32  -9.38  1.06
32    CLE    -12.0    13   0  13   0   0.0  19.9 -10.96 -10.62 -0.34

Week 15 stats

Global Statistics:
Games  Home Wins HwPct Winning_Score Losing_Score Margin
224       124     55.4      28.00        16.15     11.85

Calculated Pythagorean Exponent:  2.74


Rank  Team    Median  GP   W   L   T  Pct   Pred   SRS    MOV   SOS
------------------------------------------------------------------------
1     PHI      9.0    14  12   2   0  85.7  77.5  10.65  11.36 -0.70
2     MIN      8.0    14  11   3   0  78.6  72.2   8.76   7.21  1.55
3     NE       7.5    14  11   3   0  78.6  73.1   8.22   8.64 -0.43
4     PIT      3.0    14  11   3   0  78.6  64.2   4.52   4.71 -0.20
5     JAX     14.0    14  10   4   0  71.4  83.1   8.90  11.79 -2.88
6     NO       9.5    14  10   4   0  71.4  72.4  10.40   8.50  1.90
7     LA       8.0    14  10   4   0  71.4  78.6  11.89  11.86  0.03
8     CAR      4.5    14  10   4   0  71.4  59.9   5.33   3.21  2.12
9     ATL      3.0    14   9   5   0  64.3  58.1   3.81   2.57  1.24
10    BAL     10.0    14   8   6   0  57.1  69.4   4.77   6.36 -1.59
11    KC       7.5    14   8   6   0  57.1  61.6   3.75   4.07 -0.32
12    DAL      7.0    14   8   6   0  57.1  55.3   1.31   1.79 -0.47
13    BUF      4.5    14   8   6   0  57.1  40.0  -3.80  -3.00 -0.80
14    DET      3.0    14   8   6   0  57.1  53.7   2.47   1.36  1.12
15    SEA      3.0    14   8   6   0  57.1  56.0   1.74   1.93 -0.19
16    TEN      3.0    14   8   6   0  57.1  44.9  -5.46  -1.64 -3.81
17    GB       0.0    14   7   7   0  50.0  44.9  -0.37  -1.71  1.34
18    LAC     -0.5    14   7   7   0  50.0  63.3   3.33   4.00 -0.67
19    OAK     -2.0    14   6   8   0  42.9  40.4  -4.15  -3.07 -1.08
20    WAS     -5.5    14   6   8   0  42.9  39.0  -1.01  -3.86  2.85
21    MIA     -5.5    14   6   8   0  42.9  30.2  -5.50  -6.43  0.93
22    ARI     -5.5    14   6   8   0  42.9  29.7  -5.80  -6.50  0.70
23    CIN     -3.5    14   5   9   0  35.7  32.4  -6.96  -5.14 -1.81
24    NYJ     -5.0    14   5   9   0  35.7  37.8  -4.27  -4.07 -0.20
25    DEN     -8.5    14   5   9   0  35.7  33.2  -6.51  -5.29 -1.22
26    SF      -3.0    14   4  10   0  28.6  31.3  -6.77  -6.00 -0.77
27    TB      -4.0    14   4  10   0  28.6  38.9  -2.63  -3.64  1.01
28    CHI     -4.5    14   4  10   0  28.6  34.9  -1.71  -4.29  2.58
29    HOU     -6.5    14   4  10   0  28.6  38.3  -4.73  -4.36 -0.38
30    IND     -5.0    14   3  11   0  21.4  20.6 -11.70 -10.21 -1.49
31    NYG     -8.5    14   2  12   0  14.3  22.9  -7.21  -9.07  1.86
32    CLE    -13.0    14   0  14   0   0.0  17.8 -11.28 -11.07 -0.21

This question came up when I was looking up the last year in the playoffs for seven probable NFC playoff teams. Both New Orleans and Philadelphia last played in the playoffs four years ago, in 2013. And then the thought came up in my head, “But Drew Brees is a veteran QB.” This seems intuitive, but wanting to actually create such a definition and then later to test this using a logistic regression, there is the rub.

There are any number of QBs a fan can point to and see that the QB mattered. Roger Staubach seemed a veteran in this context back in the 1970s, Joe Montana in the 1980s, Ben Roethlisberger in the 21st century, Eli Manning in 2011, and Aaron Rogers last year. But plenty of questions abound. If a veteran QB is an independent variable whose presence or absence changes the odds of winning a playoff game, what tools do we use to define such a person? What tools would we use to eliminate entanglement, in this case between the team’s overall offensive strength and the QB himself?

The difference between a good metric and a bad metric can be seen when looking at the effect of the running game on winning. The correlation between rushing yards per carry and winning is pretty small. The correlation between run success rate and winning are larger. In short, being able to reliably make it on 3rd and 1 contributes more to success than running 5 yards a carry as opposed to 4.

At this point I’m just discussing the idea. With a definition in mind, we can do one independent variable logistic regression tests. Then with a big enough data set – 15 years of playoff data should be enough, we can start testing three independent variable logistic models (QB + SOS + PPX).