|
Scheduling issues:
Sandwich games
A common belief is that even the best of teams can fall victim to unfortunate scheduling and suffer a letdown game. Outclassed by a seemingly inferior side, the pundits point to the better team's lack of motivation coming into the event.
|
Schedule & Motivation analysts will try to convince you that the key to predicting future performances can be as easy as finding "schedule letdowns" spots. You examine a team's schedule of games and try and find a situation where a squad will be looking ahead from this week's game for a big matchup the following week, or alternatively a team ready for a "down" game after a series of tough matchups. You can find these situations (horse racing folks like to call it a "bounce") several times a season...after the fact.
Therein lies the problem -- while it's easy enough in hindsight to say that a team "didn't show up" for a particular game, doing it advance is much trickier, and perhaps statistically there is no value in this type of assessment (if for instance for every game a team under-performs, there are five where a team in similar circumstances.over-achieves).
So how do you spot the letdown beforehand? The general tip-offs according to some is as follows -- you have a good team, who probably think they're one of the big boys, taking on a poor team, who are easy to overlook. The good team is favored by a bunch, often playing at home (in an away game the hostile opposing crowd will often shock a team out of complacency) against the "sad-sacks," thus making it very easy to slip into the comfortable rest-up for the one that counts mode. Generally the good team will have a "big" game the following week, although sometimes the "letdown" seems to come after a series of pressure matchups.
To try and gauge the predictive value of this type of forecasting, we built a ten year database of matchups and started off with some assumptions:
- Only look at games played in weeks 9 through 16 (the first half of the season teams are still finding out where they stand in the NFL hierarchy, and the season is fresh enough that no one is presumably taking anything for granted). In addition, week 17 games can see starters rested for the playoffs.
- The key data points are: the quality of your last opponent, the quality of this week's opponent, and the quality of your next opponent
For simplicity sake we break out teams by "quality" into two categories: .500 or better record, or below .500 record. While we can run more advanced setups in the future (eg have three classes, say <.400, .400 to .599, and .600+) by limiting it to two groups you have eight possible "schedule situations" to assess, which should produce a large enough sample in any set to have some merit.
The most obvious schedule letdown, as alluded to above is: good team, playing a bad team after playing a good team, with another good team on the horizon. In other words the team has a game against a soft opponent "sandwiched" by games against good foes:
Last week: good team
This week: bad team
Next week: good team
The following chart shows the results of a side in such a spot, overall (the furthest right column) and by the team's own winning percentage.
Results against the spread,
good (last) - bad (this) - good (next) opponent schedule
Team Win%
---> |
<35% |
35%-44% |
45%-54% |
55%-64% |
65%+ |
ALL |
| HomeFavs |
6 - 8 |
10 - 5 |
6 - 5 |
9 - 10 |
11 - 8 |
42 - 36 |
| HomeDogs |
3 - 6 |
1 - 0 |
0 - 1 |
0 - 0 |
0 - 0 |
4 - 7 |
| AwayFavs |
3 - 0 |
6 - 3 |
2 - 3 |
6 - 12 |
15 - 13 |
32 - 31 |
| AwayDogs |
7 - 11 |
3 - 6 |
2 - 2 |
1 - 3 |
1 - 2 |
14 - 24 |
| Favorites |
9 - 8 |
16 - 8 |
8 - 8 |
15 - 22 |
26 - 21 |
74 - 67 |
| Underdogs |
10 - 17 |
4 - 6 |
2 - 3 |
1 - 3 |
1 - 2 |
18 - 31 |
| Home Teams |
9 - 14 |
11 - 5 |
6 - 6 |
9 - 10 |
11 - 8 |
46 - 43 |
| Away Teams |
10 - 11 |
9 - 9 |
4 - 5 |
7 - 15 |
16 - 15 |
46 - 55 |
| ALL PICKS |
19 - 25 |
20 - 14 |
10 - 11 |
16 - 25 |
27 - 23 |
92 - 98 |
| WIN % |
43 % |
59 % |
48 % |
39 % |
54 % |
48 % |
| Spread Range |
10+ points |
5 to 9.5 |
0 to 4.5 |
| Favorites |
13 - 13 |
37 - 29 |
24 - 25 |
| Underdogs |
0 - 1 |
2 - 8 |
16 - 22 |
The opposite to what the common wisdom says appears to be true -- a team facing a lowly opponent sandwiched by good opponents is not significant if the team is favored, however, if the team is an underdog, then the "lowly opponent" may well sneak up on them, especially when the team is getting big points (2-9 as a five plus point dog). The very best teams (65%+ record on the season) are actually above .500 against the spread overall.
The sample size is of course tiny when you break it down so fine, but the overall mark of 48% versus the point spread in such a spot makes us think there's not a lot to be gained by trying to isolate teams in this kind of situation.
As we said though, there are eight combinations of scheduling setups when you use the good/bad opponent designation:
good-good-good
good-good-bad
good-bad-good (the above stats)
good-bad-bad
bad-good-good
bad-good-bad
bad-bad-good
bad-bad-bad
Some of these other categories were actually much more revealing and potentially worth looking for in the upcoming season. All eight charts can be found in the Subscriber article below.
See Also: Sandwich Game Research, part II (Subscriber Feature)
Then of course we can also expand our search to include the contest two games prior to today's matchup (and so track down what happens to teams off say two straight games against tough opponents), which will be highlighted in an upcoming research article.
| 
