Strength of schedule flawed

[TheGodInAGreyHoodie]

I have concluded that “strength of schedule” is a really bogus measure of determining how difficult a teams schedule actually is, because the more a team wins the weaker the schedule becomes.

For example the New England Patriots strength of schedule is .469 based on an opponent record of 120-136 and the Miami Dolphins strength of schedule is .539 based on an opponent record of 138-118.

But here is the problem: of the 136 loses the Patriots opponents had 19 of them came from the Patriots and of the 138 wins on the Dolphin opponents schedule 18 come from Dolphin loses.

Lets say we play a little what if... We change the outcome of 15 games... Miami goes 16-0. Now the Dolphin opponent record would be 117-139 instead of 138-118 giving them strength of schedule of .457.

Likewise if we have NE lose 16 games then the Patriots opponents record would be 142-114 or .554.

The reason the Dolphins strength of schedule appears so much harder than the Patriots is not that the Dolphins played better teams, it is because the Dolphins lost and the Patriots won.

A better measure would be not to count the teams own games.
If we do that the SOS would be as follows:
Miami 117-117 .500
NE 120-114 .521

BTW under this measure Colts still have a more difficult SOS than the Pats.
Colts new method: 128-106 .547 instead of 132-124 .516 (old method)

 

[PatsFaninAZ]

Most computerized SOS things I've seen do not count the subject team's wins (or losses) in calculating opponent's record. It's pretty standard. RPI and BCS formula are two pretty sophisticated SOS ranking schemes and both do it too, so it's pretty commonly accepted practice.

 

[godef]

I have concluded that “strength of schedule” is a really bogus measure of determining how difficult a teams schedule actually is, because the more a team wins the weaker the schedule becomes.

This is simply false. Perhaps one thing that perhaps hasn't occurred to you: the Pats schedule next year, on top of the 14 years-down-the-line-predetermined games, is rounded off by playing 1st place teams from the two other divisions in the AFC that they won't round-robin next year. OTOH, the 2nd place Bills will additionally play the 2nd place teams from those divisions, etc.

This is done because it is recognized that the better a team is, the easier their schedule gets if the schedule were a strict round-robin venture. In other words, the Patriots can't place a first team in their own division, whilst the Bills, Dolphins and Jets all have to do so twice. So they schedule other games against 1st place teams to make up for it.

Bottom line: next season, based on this season's results, ALL NFL teams will play four 1st place teams, four 2nd place teams, four 3rd place teams and four 4th place teams, no exceptions. This has been the case every season since 2003 (2002 was skewered by the change from 6 to 8 divisions). The only real irregularity about this setup is that you may sometimes have 3 home games and 1 road game (or vice versa) against the four 1st place teams, or 2nd place teams, etc.

First point being, there IS currently "balanced scheduling" in the NFL (ie there are no advanatges to 4th place schedules, nor disadvantages to 1st place schedules). Any imbalance in SOS comes strictly from the random variances of good or bad teams in the various divisions.

Second point being, considered logically, NONE of this has anything to do with SOS in a completed season; a .xxx winning percentage is a .xxx winning percentage, no matter whether those teams involved are all first place teams or all last place teams. The Pats in 2007 played teams that compiled a combined .450 winning perdentage, which was better than the sub .400 that the Fins in 1972 faced. In 2008, the Pats will apparently have an easy schedule, if all the teams hold to their 2007 form, all due strictly to the variances mentioned in the first point.

 

[MDPATSFAN]

I have concluded that “strength of schedule” is a really bogus measure of determining how difficult a teams schedule actually is, because the more a team wins the weaker the schedule becomes.

For example the New England Patriots strength of schedule is .469 based on an opponent record of 120-136 and the Miami Dolphins strength of schedule is .539 based on an opponent record of 138-118.

But here is the problem: of the 136 loses the Patriots opponents had 19 of them came from the Patriots and of the 138 wins on the Dolphin opponents schedule 18 come from Dolphin loses.

Lets say we play a little what if... We change the outcome of 15 games... Miami goes 16-0. Now the Dolphin opponent record would be 117-139 instead of 138-118 giving them strength of schedule of .457.

Likewise if we have NE lose 16 games then the Patriots opponents record would be 142-114 or .554.

The reason the Dolphins strength of schedule appears so much harder than the Patriots is not that the Dolphins played better teams, it is because the Dolphins lost and the Patriots won.

A better measure would be not to count the teams own games.
If we do that the SOS would be as follows:
Miami 117-117 .500
NE 120-114 .521

BTW under this measure Colts still have a more difficult SOS than the Pats.
Colts new method: 128-106 .547 instead of 132-124 .516 (old method)

There is a distortion the SOS equation caused by the putrid wretchedness of the Phins/Jets and the Pats having to play them four times....Those two teams are responsible for 54 of the 136 losses totalled by Pats opponents.

In other words they are drag on the division.

 

[TheGodInAGreyHoodie]

Most computerized SOS things I've seen do not count the subject team's wins (or losses) in calculating opponent's record. It's pretty standard. RPI and BCS formula are two pretty sophisticated SOS ranking schemes and both do it too, so it's pretty commonly accepted practice.

The NFL, however, do count the subject teams wins and loses.

http://www.nfl.com/draft/story?id=09000d5d805a3a57&template=without-video&confirm=true

 

[godef]

There is a distortion the SOS equation caused by the putrid wretchedness of the Phins/Jets and the Pats having to play them four times....Those two teams are responsible for 54 of the 136 losses totalled by Pats opponents.

In other words they are drag on the division.

Yeah, they do drag down the division. But still, the SOS is not based on division performance, it's based on absolute wins and losses of opponents. The whole point of determining an SOS is to compare it to that of other teams. If you want something more concise, one can go down further levels and determine the SOS of those opponents, but an SOS is still better that just simply looking at a single team's won-loss record.

 

[A.C Vegas]

In the end does it really matter the SOS is only for those who want to find fault with or hope for there team. the only place SOS matters is in tie breakers and even then for the playoffs its way down on the list. And for the draft it does make since

 

[AzPatsFan]

The NFL, however, do count the subject teams wins and loses.

http://www.nfl.com/draft/story?id=09000d5d805a3a57&template=without-video&confirm=true

From this I deduce that the Pats will pick either fourth or fifth in the third round from the Raiders pick. That would be either 67th or 68th pick. (31 + 32+ (4 or 5))