Fast Starts and Full Season Effect
I have read quite a few previews of Michigan's upcoming season, wherein most of them indicate that Michigan is due for a season of 6-7 wins, but their eventual win total hinges upon how they acquit themselves in the early going. Should they stumble early, they may find the rest of their season insurmountable. However, if they win early, that, they say, will build momentum to carry them through the tougher portion of their schedule.
Well, I've heard talk of early season momentum and early season letdowns so often over the years, I wanted to see if there was anything at all to this thing. As such, I visited the NCAA football website and got some stats. I had originally planned to map this over many years, but as I got looking, I saw that this was one statistic that was really quite consistent year in and year out, so I've only got the data points for last season. Should equate fairly well to any year.
The first thing I needed to establish was how many games made up the "start" of the season. I hung that one up at 4 games, because by that point, the kinks should be worked out and pretty much every team should have played at least one conference game, or one or more highly challenging OOC games. I then divided the teams by total games won (Regardless of bowl games/longer seasons), and figured the respective group's records in the first four games of the season. Results below:
Group A: 10 - 13 Wins Group B: 8-9 Wins
20 Teams 31 Teams
4-0: 13 Teams, 65% 4-0: 5 Teams, 16%
3-1: 6 Teams, 30% 3-1: 16 Teams, 52%
2-2: 1 Team, 5% 2-2: 8 Teams, 26%
1-3: 2 Teams, 6%
Group C: 7 Wins Group D: 6 Wins
14 Teams 7 Teams
4-0: 3 Teams, 21% 4-0: 0 Teams
3-1: 4 Teams, 29% 3-1: 1 Team, 14%
2-2: 5 teams, 35% 2-2: 4 Teams, 57%
1-3: 2 Teams, 14% 1-3: 2 Teams, 29%
Group E: 5 Wins Group F: 4-3 Wins
16 Teams 20 Teams
4-0: 0 Teams 4-0: 0 Teams
3-1: 4 Teams, 25% 3-1: 1 Team, 5%
2-2: 6 Teams, 38% 2-2: 7 Teams, 35%
1-3: 6 Teams, 38% 1-3: 11 Teams, 55%
0-4: 1 Team, 5%
Group G: 0-2 Wins
9 Teams
4-0: 0 Teams
3-1: 0 Teams
2-2: 2 Teams, 22%
1-3: 5 Teams, 56%
0-4: 2 Teams, 22%
Now, taking this data, we can make some projections. If Michigan were to have a "Fast Start", as the experts are saying is necessary, and we assume "Fast Start" to be 3-1 or better, then we can see where the chips are most likely to fall.
Excluding a 10-13 Win season, startin 3-1 or better makes 8-9 wins the most likely. That seems obvious. You have to win a lot of games to get 8-9 wins. However, it seems interesting that a team winning their first four is more likely to win only 7 than 8-9 total. Similarly, it's odd tht those going 3-1 in their first 4 are more likely to fall to 5 wins overall than reach 6, by a substantial percentage margin. And as can be seen from the 2-2 mark, this is not because those teams are winning more.
I also wondered if the Big 10, with a penchant for scheduling weak non-conference games in the early season, might be less than reflective of the sport as a whole. As such, I checked that out.
Group A
2 Big 10 Teams, 4-0&3-1
Group B
3 Big 10 Teams, 4-0, 3-1, 3-1
Group C
2 Big 10 Teams, 4-0, 3-1
Group E
1 Big 10 Team, 2-2
Group F
3 Big 10 Teams, 2-2, 2-2, 2-2
Big 10 Teams fell into the 4-0 Win total 27% of the time. Average over all groups containing 4-0 Teams was 34%.
Big 10 Teams fell into the 3-1 Win total 36% of the time. Average over all groups is 26%.
Big 10 Teams fell into the 2-2 Win total 36% of the time. Average over all groups is 31%
The Big 10 lands a little below average in one win category, a little above in another, and just about dead on in the third. The total falls slightly over the national average in wins for the first four games, but not substantiantially. As such, Michigan's expected final win total shouldn't be grossly inflated or deflated because of OOC strength. Big ten teams don't perform noticably better than other conference teams during the OOC portion of the schedule, and so Michigan shouldn't get noticably worse (in wins and losses) during the conference season.
What conclusions can we draw from this? It looks like Michigan does indeed need to win as many early season games as possible, but only in that it bolsters their win total overall. About half of all 7 win teams did not have a "fast start" (3-1 or 4-0) record, and likewise, a quarter of all 5 win teams had a fast start (and 2/3rds had at least an even record after 4), and failed to put forth a winning season.
So if Michigan loses 2 or 3 of their first 4 (I hope not) and somebody tells you their toast, tell them the stats only say, "Eh... maybe". Likewise, if they go 4-0 and somebody tells you the've got it made, tell them that 21% of 4-0 teams last year barely made a bowl game.
Just take the season one game at a time, and take a "W" for what it is. A "W".
Well, I've heard talk of early season momentum and early season letdowns so often over the years, I wanted to see if there was anything at all to this thing. As such, I visited the NCAA football website and got some stats. I had originally planned to map this over many years, but as I got looking, I saw that this was one statistic that was really quite consistent year in and year out, so I've only got the data points for last season. Should equate fairly well to any year.
The first thing I needed to establish was how many games made up the "start" of the season. I hung that one up at 4 games, because by that point, the kinks should be worked out and pretty much every team should have played at least one conference game, or one or more highly challenging OOC games. I then divided the teams by total games won (Regardless of bowl games/longer seasons), and figured the respective group's records in the first four games of the season. Results below:
Group A: 10 - 13 Wins Group B: 8-9 Wins
20 Teams 31 Teams
4-0: 13 Teams, 65% 4-0: 5 Teams, 16%
3-1: 6 Teams, 30% 3-1: 16 Teams, 52%
2-2: 1 Team, 5% 2-2: 8 Teams, 26%
1-3: 2 Teams, 6%
Group C: 7 Wins Group D: 6 Wins
14 Teams 7 Teams
4-0: 3 Teams, 21% 4-0: 0 Teams
3-1: 4 Teams, 29% 3-1: 1 Team, 14%
2-2: 5 teams, 35% 2-2: 4 Teams, 57%
1-3: 2 Teams, 14% 1-3: 2 Teams, 29%
Group E: 5 Wins Group F: 4-3 Wins
16 Teams 20 Teams
4-0: 0 Teams 4-0: 0 Teams
3-1: 4 Teams, 25% 3-1: 1 Team, 5%
2-2: 6 Teams, 38% 2-2: 7 Teams, 35%
1-3: 6 Teams, 38% 1-3: 11 Teams, 55%
0-4: 1 Team, 5%
Group G: 0-2 Wins
9 Teams
4-0: 0 Teams
3-1: 0 Teams
2-2: 2 Teams, 22%
1-3: 5 Teams, 56%
0-4: 2 Teams, 22%
Now, taking this data, we can make some projections. If Michigan were to have a "Fast Start", as the experts are saying is necessary, and we assume "Fast Start" to be 3-1 or better, then we can see where the chips are most likely to fall.
Excluding a 10-13 Win season, startin 3-1 or better makes 8-9 wins the most likely. That seems obvious. You have to win a lot of games to get 8-9 wins. However, it seems interesting that a team winning their first four is more likely to win only 7 than 8-9 total. Similarly, it's odd tht those going 3-1 in their first 4 are more likely to fall to 5 wins overall than reach 6, by a substantial percentage margin. And as can be seen from the 2-2 mark, this is not because those teams are winning more.
I also wondered if the Big 10, with a penchant for scheduling weak non-conference games in the early season, might be less than reflective of the sport as a whole. As such, I checked that out.
Group A
2 Big 10 Teams, 4-0&3-1
Group B
3 Big 10 Teams, 4-0, 3-1, 3-1
Group C
2 Big 10 Teams, 4-0, 3-1
Group E
1 Big 10 Team, 2-2
Group F
3 Big 10 Teams, 2-2, 2-2, 2-2
Big 10 Teams fell into the 4-0 Win total 27% of the time. Average over all groups containing 4-0 Teams was 34%.
Big 10 Teams fell into the 3-1 Win total 36% of the time. Average over all groups is 26%.
Big 10 Teams fell into the 2-2 Win total 36% of the time. Average over all groups is 31%
The Big 10 lands a little below average in one win category, a little above in another, and just about dead on in the third. The total falls slightly over the national average in wins for the first four games, but not substantiantially. As such, Michigan's expected final win total shouldn't be grossly inflated or deflated because of OOC strength. Big ten teams don't perform noticably better than other conference teams during the OOC portion of the schedule, and so Michigan shouldn't get noticably worse (in wins and losses) during the conference season.
What conclusions can we draw from this? It looks like Michigan does indeed need to win as many early season games as possible, but only in that it bolsters their win total overall. About half of all 7 win teams did not have a "fast start" (3-1 or 4-0) record, and likewise, a quarter of all 5 win teams had a fast start (and 2/3rds had at least an even record after 4), and failed to put forth a winning season.
So if Michigan loses 2 or 3 of their first 4 (I hope not) and somebody tells you their toast, tell them the stats only say, "Eh... maybe". Likewise, if they go 4-0 and somebody tells you the've got it made, tell them that 21% of 4-0 teams last year barely made a bowl game.
Just take the season one game at a time, and take a "W" for what it is. A "W".
this is pretty cool but I'm not sure those wired trend, like going 3-1 is more like to win 5 than 6, would stand up over more seasons
I think a lot of teams that suck in their conference (K-State, Minnesota) schedule extremely weak non-conference games, and lose almost all of their conference games.
"I think a lot of teams that suck in their conference...lose almost all of their conference games."
I smell a Harvard degree here.
I'd like, as a fellow numbers geek, to thank the OP for this analysis. It's that time of year when we think of every angle we can and to those who have the ballz to put it out there, Captain Morgan says, "Hello!"
It doesn't. The stats stay similar, but the individual team up and downs vary too much. See, that's what I mean, that doesn't make sense, but it DOES happen quite a bit. It just jumps around where it happens. So when somebody says you can't win with a slow start, or Michigan needs to start fast, it's not really true. They just need wins. Starting fast has nothing to do with it.
I wonder if this doesn't just show that on average, a team's wins are spread evenly throughout their season.
For instance: a team that wins 10 games will only have two or three losses. That's one loss every 3-5 games, thus the vast majority of these teams have 0 or 1 loss in the first four...
This assumes that the games early in the season and late in the season are equally difficult of course, which isn't true most of the time for Michigan.
Interesting analysis nonetheless, thanks for taking the time to crunch the numbers!
That was my point exactly. Thank you for reading closely. There is very little to show that "momentum" has anything to do with how the season plays out.
I think BlockM is right about these "trends".
We need to get off to a fast start in 2009 if we want to succeed, because we have some of our less heralded opponents early and if we can't beat those teams (WMU, EMU, Indiana), we will certainly struggle against the better teams (OSU, PSU, Illinois).
Maybe I'm reading this incorrectly, but I think there are some errors here. For example, you say a team winning the first 4 is more likely to win 7 than 8 or 9, but it seems that 5 out 21 teams that started out 4-0 finished with 8-9 wins and only 3 out of 21 teams that started out 4-0 finished with 7. Similarly, you say that 21% of 4-0 teams barely made a bowl, but it seems more 14% (3/21) of 4-0 teams barely made a bowl. Instead, 21% (3/14) of teams that finished with 7 wins started out 4-0.
I apologize if I'm misunderstanding you.
I was afraid this would happen. I was coming at it from the reverse way. My stats are subdivided by wins FIRST, so instead of saying, out of teams that started 4-0, 21% of those barely reached a bowl, you would say, "21% of teams that finished with 7 wins and barely made a bowl started with 4 wins."
AGAIN, Point here was to say that fast or slow starts are NOT a good reflection of season performance.
I really wish you had done BCS conference teams only. That being said...
This sort of analysis is far less meaningful than the fact that losing to Western means 1) we don't have that as a win and 2) we are probably a mediocre team if we lose to Western. So, the diary was interesting but not useful as a predictor of success.
If you're going to use statistics, something on turnover margin would be useful, particularly if you could factor in the effect of an inexperienced DL, a freshman QB, and a new DC. That's a highly random statistic with some nonrandom influences.
Well, like I said, my point was simply to say that when people say we "must have a fast start" to be successful, that's not really the case. It wasn't too in depth because I was saying there's not really anything here to support that.
I honestly think that we need a fast start. With a schedule that starts with:
Western Michigan
Notre Dame
Eastern Michigan
Indiana
We really need to start at least 3-1 in the first 4 games if we want to get a decent bowl game. Additionally, if we go 2-2 (or worse) against the above competition, the odds are that we will fare worse when we are playing better competition like Ohio State, Penn State, @ Iowa, @ Illinois, @ Wisconsin, @ MSU.
You are correct that we need a fast start in that those are some of our best chances to get wins. A fast start is NOT going to build "momentum" for us later in the season though. So if we start slow, don't panic.
"So if we start slow, don't panic."
Well if we can't beat EMU and Indiana, I won't bet money on the PSU or OSU games!
I think there are a number of problems with drawing conclusions from your analysis.
1. Why does an undefeated or one loss season give validity to a "fast start" momentum theory? If USC has an 11 win season next year, do we think it will be owed to their "fast start?" I think teams that win a lot of games win a lot of games--including their first 4.
2. "Fast starts" can be the result of pathetic non-conference schedules. During the darkest PSU days of the early 2000s PSU would struggle to break .500, after pummeling Temple and friends. They would falsely undermine the fast start argument. By the same token, small schools that are unafraid to play tough teams (e.g. Troy) may lose a few early and then pummel everyone in their weak regular conference. The degree of difficulty does not remain constant from the non-conference warmups to regular conference matchups, and that is a problem.
3. "as I got looking, I saw that this was one statistic that was really quite consistent year in and year out, so I've only got the data points for last season. Should equate fairly well to any year." You found the statistic was consistent year in and year out, but limited your inquiry to last season (one year)--how did that work, exactly?
4. "the Big 10, with a penchant for scheduling weak non-conference games in the early season" -- I don't think this is true. There was one ESPN column this offseason mocking our 2009 OOC schedules. I don't think there is any historical validity to this claim.
1. Read closer. That's what I said. It DOESN'T have any validity, or very little. Especially not for teams winning a lot of games anyway.
2. Please note my reference to the Big 10 and the soft early schedules. They fall more or less within the bounds of all other teams. It certainly has an effect, but not a dramatic one.
3. I.E. - I was doing more years, but the distribution of wins and losses versus starts was falling about the same, and I didn't feel like taking more time to say that something was an invalid argument to make.
4. Go back to point two. I was reaffirming what you said already.
this needs "research"? a team that has a good season will also have done well in their first few games. this is not a discovery or revelation or any such thing. this is just useless "research".
You say:
"However, it seems interesting that a team winning their first four is more likely to win only 7 than 8-9 total."
you silly fellow! look again at the stats you posted and interpret them correctly. your conclusion is wrong and absurd.
A: I was pointing out that there was very little or nothing to the idea of momentum. My research showed just that. There was nothing there to RESEARCH. As I noted. Teams that win a lot of games HAVE to have fast starts, to win a lot of games.
B: Percentage wise. Seriously. A higher percentage of teams that won 7 started by winning 4 than did those who won 8 or 9 total. That's exactly what the stats said.
you truly don't understand what you're talking about. you just mentioned a completely different thing in your followup than in your OP.
In the OP, you said
"However, it seems interesting that a team winning their first four is more likely to win only 7 than 8-9 total"
Obviously, this is wrong. More teams (5) won 8-9 total than 7 total (3 teams).
What you're now saying is something completely different. I thought only your math was bad. now i see even English and/or analytical skills suck.
For the record, I didn't down vote you.
Also for the record, you apparently can't read or don't understand the word "percentage". A greater percentage of teams that won 7 games total started 4-0 than did those that won 8-9 games total.
Three teams is indeed fewer than 5, but it is a greater PERCENTAGE of the total teams in each respective win category.
Before you go deriding other people's logic, reasoning, and communication skills, perhaps you should evaluate your own interpretation of what they were saying and see if you could be mistaken.
at least he put forth an effort to construct this diary.
for this season's squad is important considering the debacle last year and the youth and inexperience as a whole. The team had zero swagger last season. Even though the non-con games may be considered "gimmes" (except for ND), soaring to a 3-1 or 4-0 start would inject some much needed confidence for the tougher games on the schedule.
Comments