End of the season momentum. Does it exist? (1 Viewer)

[Birdnals]

While working on my draft prep and projections for this season, I have been collecting quite a bit of data on players from the last five seasons. I’ve always been a big believer in end-of-the-season momentum carrying forward to the next season but this year I decided to analyze it further. My RB data consists of 262 running backs from 2008-2012 that includes RBs drafted in the first 14 rounds that year and the RBs who finished in the top 40 of all RBs that season. To analyze the effect of momentum, a section of my statistics includes the stats for that player in the last four games of the previous season (adjusted to a per game basis to account for players sitting towards the end of the season).

I broke the players down into two groups, those who finished with a points-per-game average higher than their season average (momentum group) and those who did not (non-momentum group). For the following season, the average point total of those who finished with momentum was 125 vs. 121 for those without momentum. A two-sample t-test revealed that from this dataset, a randomly selected player from the momentum group had a 31% chance of having a better fantasy season the following year than those who not in the momentum group.

A nearly one in three chance of having a better season seems to support the idea of momentum. Now the question is, how much of a better season? Once again, the two-sample t-test comes in handy. I tested the odds of a player from the momentum group having a 1, 2, and 3 point per game advantage over those from the non-momentum group. The results: a momentum player had a measly 3.5% chance of a 1 point-per-game advantage over the non-momentum one, a 0.016% chance of having a 2 points-per-game advantage, and a whopping 0.000000000754% chance of a 3 point-per-game advantage. Needless to say, I stopped there.

I think what is perhaps most sobering is the fact that out of the 148 players who finished with momentum, only 59 of them had a better season the following year when compared to the season they finished strongly (40%). Meanwhile, out of the group finishing without momentum, 34% had a better season the following year. This is hardly convincing in support of momentum leading to a better following season.

While I was expecting to find support of momentum, this information is just as valuable to fantasy players. Relying too heavily on momentum, or relying on it at all, is not a winning play. Does it exist? Maybe, but we don’t have any evidence here to support it. And even if it does exist, it appears to be so minute to the point of be insignificant. I haven’t gotten the opportunity to examine other positions so it’s possible that relying on momentum is just a bad play for RBs. So just remember, as you get ready to rank your RBs, use extreme caution when looking at end-of-the-season momentum. Maybe I'm missing something here but the evidence seems pretty damning: use caution when relying on end-of-the season momentum when ranking and drafting your players.

 

[GregR]

Huge kudos for taking the time to test something like this and share it.

 

[Birdnals]

No problem, Greg. There's a lot I can do with the data I've collected so if you or anyone else have any requests let me know (I have amassed career averages up to that year, fantasy round drafted in, previous season numbers, and last four games of previous season numbers) . As a side note, I now realize the title should be 'Does it matter?' and not 'Does it exist?'.

 

[Biabreakable]

Very interesting. Could you list the 40% who did improve?

I think perhaps looking at the sample that did improve by their age compared to the ones who didn't might be a useful question to ask.

Like everything else it becomes somewhat situational to each individual player. The reason the player didn't improve could be because of injury, or perhaps a coaching change or a change at QB or other key support player. By looking at other criteria such as age, staying with the same team or other questions would be the next step.

Thanks for sharing this. I have heard many people talk about momentum and use the previous seasons last 8 games as part of their analysis. What you are saying you found so far does not seem to support this idea being valid. However looking further into it, it is possible that players within a certain age group may be more useful than another for this type of analysis.

Great post!

 

[Birdnals]

Very interesting. Could you list the 40% who did improve?

I think perhaps looking at the sample that did improve by their age compared to the ones who didn't might be a useful question to ask.

Like everything else it becomes somewhat situational to each individual player. The reason the player didn't improve could be because of injury, or perhaps a coaching change or a change at QB or other key support player. By looking at other criteria such as age, staying with the same team or other questions would be the next step.

Comparing age groups is on tap next. I can compare different factors such as age or experience amongst the groups who finished with momentum and improved vs those who finished with momentum and did not improve to see if there are any significant factors. The coaching/personnel changes are likely to be a factor (or at least I suspect they would be) but unfortunately I don't have that data entered right now. I'd like to examine this but unfortunately I'm not sure that I have the time to research/enter this data for 200+ players right now. I actually did this testing just to give myself a break before doing the same data collection for WRs while I attempt to build a regression model predicting future production of players (I'm likely not posting this now because I know how inaccurrate it's likely to be but I'm interested to see if it can be used at all to help guide projections). Once I'm done with my initial projections I plan on revistting some of the data and looking for more tidbits like this. Given the intense debate going on right now about Foster, I'd also like to look at historically similar RBs and see if the past can shed any light on this debate.

 

[DropKick]

Your initial conclusion was that you had almost a 1 in 3 chance of a better season with a "momentum" player. Then you talked yourself out of it because, on a per game basis, the points difference was small. But the "average point total" you're quoting (121-125) for the following season is very low - less than 8 points per game? Is this due to a very conservative scoring system or an average brought down with part time players?

I think we can mislead ourselves with mathematical analysis. How many variables does the model consider? As another poster said, every situation is different. Teams, coaching staffs, etc. change each year. The player's age could be a factor. The reasons for the strong finish could now be irrelevant (an injured starter, a huge game against a weak D, a flukey week 17).

I've used a strong finish to target emerging players with good success. Of course, talent and opportunity also factor into the decision.

 

[Birdnals]

Your initial conclusion was that you had almost a 1 in 3 chance of a better season with a "momentum" player. Then you talked yourself out of it because, on a per game basis, the points difference was small. But the "average point total" you're quoting (121-125) for the following season is very low - less than 8 points per game? Is this due to a very conservative scoring system or an average brought down with part time players?

I think we can mislead ourselves with mathematical analysis. How many variables does the model consider? As another poster said, every situation is different. Teams, coaching staffs, etc. change each year. The player's age could be a factor. The reasons for the strong finish could now be irrelevant (an injured starter, a huge game against a weak D, a flukey week 17).

I've used a strong finish to target emerging players with good success. Of course, talent and opportunity also factor into the decision.

Yes, I did convince myself out of using momentum as a source of moving players up my draft board. There are obviously other factors in identifying emerging talent but the purpose of this test was to show how strong of a factor momentum plays going into the next season and the testing done here in fact shows that momentum does not play much of a factor, if any. As another poster suggested, we can break the situation down further to see how it affects different groupings of players, but overall the data overwhelmingly shows that you're wasting your time factoring end of the season momentum into your projections and that the cause your success when targeting emerging players is due to something else.

In regards to the points/scoring system, I used the Footballguys scoring system and the point totals from the Historical Data Dominator and Game Log Dominator. I was a bit shocked by the average but keep in mind that last season, only 10 RBs finished with 200+ points. When collecting data, I used the top 40 finishers (these are what I considered rosterable RBs in a 12 team league) in a given year as well as all drafted RBs according to Fantasy Football Calculator (because the consensus was that these guys could have some value and were worthy of a roster spot). Because some of the lower draft picks (and some of the earlier round flops as well) there are some players who did next to nothing and brought the numbers down. I figured these players were still important to include because they were at one point or another all considered rosterable players and it was worth looking at why they did or did not eventually pan out.

 

[Tool]

admittedly only skimmed through your post. i think momentum is an interesting thing to look at. you may want to look at a group that performed well above their average as just beating their average may not be a great indication of momentum. just an idea.