Part IV -- Coaches and the Preseason
A popular theory with handicappers regarding the preseason games is to focus on the coaches and back the ones who have a good historical record in the exhibition season. The question we will try to answer here is whether in fact a coaches previous preseason record is of much help in predicting the outcomes of future preseason games.
To begin with, we are using for our data sample, the preseason games played from 1997 to 2003 (a seven-year span). The next step is to classify coaches by their past results -- for the 1997 season analysis we considered a coach's record in the previous two years, but thereafter we use the cumulative record from 1995 on heading into the season. We broke out the coaches by the following winning (straight-up) percentages:
Excellent -- 65%+ win rate
Good -- 50% to 64%
Poor -- 36% to 49%
Bad -- 35% or less
New -- the first season with a new team
Below are the results against the spread by Coach class:
(neutral site games were excluded)
Coach Class Home
Favorite Home
Underdog Away
Favorite Away
Underdog Overall
W-L Win%
Excellent 31-26
54% 2-2
50% 11-13
46% 18-14
56% 62 - 55 53%
Good 40-54
42% 12-9
57% 9-16
36% 47-42
53% 108 - 121 47%
Poor 21-28
42% 4-6
40% 5-4
55% 32-21
61% 62 - 59 51%
Bad 25-40
38% 9-8
52% 3-6
33% 43-35
56% 80 - 89 47%
New 31-41
43% 20-7
74% 4-8
33% 49-36
58% 104 - 92 53%
So at a glance it appears there is some very modest predictive value in a coach's historical preseason record. The top group hits 53%, the bottom group hits 47%. Meanwhile, newcomer coaches lend a hand to a team's preseason chances.
Now, we mentioned a few situations last year where more detailed situational specs showed some stronger extremes in performance one way or the other, but these angles collectively were just 18-21 following the trend last year.
The only really promising pattern we see is with
Teams with NEW coaches, as home underdogs.
THe 20-7 record includes a 3-1 effort last season, so this is the one piece of the pie to take with you. Other than that, perhaps you could view a situation like going against a 'bad' preseason coach as home favorite an additional plus if you are already leaning that way. At this point we are looking at coaches in isolation, rather than comparing the two coaches in a given match-up.
The final logical step if one believes there is some predictive value in a coach's preseason history would be to say "what if I back only the Excellent preseason coaches when facing coaches with bad preseason records?"
The answer is you would have done well!
- Excellent vs Bad: 20-12 (and 5-4 in 2003)
Of course that's a mere handful of games and could easily look different over a larger backdrop of years.
Here's the 2004 NFL coaching class groups, capping the WL record at 1998 on for the long time in one spot guys --
Excellent: Gruden (Tampa Bay 78%), Belichick (New England: 76%), Shanahan (Denver: 76%), Del Rio (Jacksonville: 75%), Erickson (San Francisco: 75%), Edwards (NY Jets: 69%), Fisher (Tennessee: 67%)
Good: Dungy (Indianapolis: 63%), Fox (Carolina: 63%), Billick (Baltimore: 63%), Vermeil (Kansas City: 62%), Sherman (Green Bay: 53%), Parcells (Dallas: 50%), Mariucci (Detroit: 50%), Cowher (Pittsburgh: 50%)
Poor: Wannstedt (Miami: 47%), Davis (Cleveland: 42%), Holmgren (Seattle: 40%), Schottenheimer (San Diego: 38%), Haslett (New Orleans: 38%), Tice (Minnesota: 38%), Reid (Philadelphia: 37%)
Bad: Martz (St. Louis: 35%), Lewis (Cincinnati: 25%), Capers (Houston: 12%)
New: Smith (Chicago), Mularkey (Buffalo), Green (Arizona), Mora Jr. (Atlanta), Coughlin (NY Giants), Turner (Oakland), Gibbs (Washington)
Some of the "new" coaches are of course NFL veterans, but for this analysis we treated every coach as new when getting a fresh start.
--------------------------------------------------------------------------------