Is there such a thing as momentum? (Edit: Between games not shifts)

[SnowblindNYR]

Warning long post:

So I spent some time working on NHL analysis as a whole, not just the Rangers but wanted to share since it's stats. Particularly what I call "Next Game Analysis". Meaning what do NHL teams score in the next game after a particular game. I used two methods, one just simple cross tabs using pivot tables and one method to get into the nitty gritty, I used regression.

Cross Tab Analysis- I grouped 4+ goal wins and losses as "blowout wins and losses" and 1-3 goal wins and losses as "non-blowout wins and losses" and did a cross tab of the 4 types of scenario margins (of wins and losses) in the rows and 4 types of scenarios for the next game and took the percentage of rows (margins). Throughout the analysis I excluded all final games of seasons due to not having any regular season next games. In the aggregate the pattern that was followed was that blowout losses followed blowout losses at a higher rate than blowout losses followed non-blowout losses, which was at a higher rate that blowout losses following non-blowout wins, and this was in turn at a higher rate than blowout losses following blowout wins.

This is interesting and there are two hypothesis that I had. 1) Momentum and injuries. Maybe teams gain confidence from winning by 5 and win next game by 6 and the opposite with blowout losses. Or teams lose big consecutively because of injuries. 2) Bad teams are the ones that tend to lose big and they're also more likely to lose big in the next game because they're bad. The opposite is true of good teams.

I decided to control for record and broke down the league into 4 quarters. The range of points in the last 10 years go from 52 to 132 (132 is prorated for the 12-13 Blackhawks). The quarters are as follows 52-71 (Q1), 72-91 (Q2), 92-111 (Q3), and 112-132 (Q4). I created the same cross tabs but this time per quarter. And the relationship didn't hold as well as in the aggregate but it seemed to me that it held somewhat throughout.

Next, I decided to do this a bit of a more precise way using regression. I first looked at all point totals with the Next Game as the dependent variable and Margin as the independent variable. The slope I got was 0.031 (every 1 goal increase in current game's margin yields 0.031 increase in the next game's margin). Which is a positive slope meaning the higher the margin of victory or loss, the higher the margin of victory or loss of the next game. The p-value is less than 0.05 for the slope (p-value=0), which makes it statistically significant (roughly speaking 0 percent chance that the slope we get is random).

So, once again this may be because of how good or bad a team is. I controlled for those factors running regressions for point totals for each of the 4 quarters. When differences in points are taken away, the results are all over the place and insignificant. The least significant are the middle tier teams the most are the best and worst teams (which makes sense as they're significantly biased towards winning or losing), the most significant are the top tier teams with a slope of 0.034 and a p-value of 0.1471. Still, insignificant.

So finally I decided I'll run a multiple regression using point totals as one of two factors (margins once again being the other). Now the slope for Margin actually becomes negative but the p-value is off the charts high (0.5623) and insignificant. Points on the other hand are positive (0.0319) and are significant (p-value of 0).

In conclusion, while the cross tabs seem to point at possible momentum swings or injury factors, it appears that any patterns in the following game are related to the strength of the team. Basically, this was probably the safe bet for the hypothesis even before I began the analysis.

If you made it this far thanks for reading. My spreadsheet is attached.

Click to read more...

 

[Ohashi_Jouzu*]

Yes there is, no matter how many attempts to quantify it statistically fail.

 

[SnowblindNYR]

Yes there is, no matter how many attempts to quantify it statistically fail.

Ok, but this is the forum for statistics no? How else should I try to analyze it in a forum called "By The Numbers"?

 

[Canadiens1958]

Sitting At Home Waiting - Scheduling

Warning long post:

So I spent some time working on NHL analysis as a whole, not just the Rangers but wanted to share since it's stats. Particularly what I call "Next Game Analysis". Meaning what do NHL teams score in the next game after a particular game. I used two methods, one just simple cross tabs using pivot tables and one method to get into the nitty gritty, I used regression.

Cross Tab Analysis- I grouped 4+ goal wins and losses as "blowout wins and losses" and 1-3 goal wins and losses as "non-blowout wins and losses" and did a cross tab of the 4 types of scenario margins (of wins and losses) in the rows and 4 types of scenarios for the next game and took the percentage of rows (margins). Throughout the analysis I excluded all final games of seasons due to not having any regular season next games. In the aggregate the pattern that was followed was that blowout losses followed blowout losses at a higher rate than blowout losses followed non-blowout losses, which was at a higher rate that blowout losses following non-blowout wins, and this was in turn at a higher rate than blowout losses following blowout wins.

This is interesting and there are two hypothesis that I had. 1) Momentum and injuries. Maybe teams gain confidence from winning by 5 and win next game by 6 and the opposite with blowout losses. Or teams lose big consecutively because of injuries. 2) Bad teams are the ones that tend to lose big and they're also more likely to lose big in the next game because they're bad. The opposite is true of good teams.

I decided to control for record and broke down the league into 4 quartiles. The range of points in the last 10 years go from 52 to 132 (132 is prorated for the 12-13 Blackhawks). The quartiles are as follows 52-71 (Q1), 72-91 (Q2), 92-111 (Q3), and 112-132 (Q4). I created the same cross tabs but this time per quartile. And the relationship didn't hold as well as in the aggregate but it seemed to me that it held somewhat throughout.

Next, I decided to do this a bit of a more precise way using regression. I first looked at all point totals with the Next Game as the dependent variable and Margin as the independent variable. The slope I got was 0.031 (every 1 goal increase in current game's margin yields 0.031 increase in the next game's margin). Which is a positive slope meaning the higher the margin of victory or loss, the higher the margin of victory or loss of the next game. The p-value is less than 0.05 for the slope (p-value=0), which makes it statistically significant (roughly speaking 0 percent chance that the slope we get is random).

So, once again this may be because of how good or bad a team is. I controlled for those factors running regressions for point totals for each of the 4 quartiles. When differences in points are taken away, the results are all over the place and insignificant. The least significant are the middle tier teams the most are the best and worst teams (which makes sense as they're significantly biased towards winning or losing), the most significant are the top tier teams with a slope of 0.034 and a p-value of 0.1471. Still, insignificant.

So finally I decided I'll run a multiple regression using point totals as one of two factors (margins once again being the other). Now the slope for Margin actually becomes negative but the p-value is off the charts high (0.5623) and insignificant. Points on the other hand are once again positive (0.0319) and are significant (p-value of 0).

In conclusion, while the cross tabs seem to point at possible momentum swings or injury factors, it appears that any patterns in the following game are related to the strength of the team. Basically, this was probably the safe bet for the hypothesis even before I began the analysis.

If you made it this far thanks for reading. My spreadsheet is attached.

Suggest looking at scheduling and the "Sitting at Home Waiting" phenomena.

Basically this looks at fatigue during the regular season schedule. A team playing three games in four nights on the road, going into the third game on the road against a fresh home team that has not played in three or four days has a distinct disadvantage in terms of fatigue vs rest, preparation time, treating minor injuries.

You would have to mine and interpret your own data.

 

[SnowblindNYR]

Suggest looking at scheduling and the "Sitting at Home Waiting" phenomena.

Basically this looks at fatigue during the regular season schedule. A team playing three games in four nights on the road, going into the third game on the road against a fresh home team that has not played in three or four days has a distinct disadvantage in terms of fatigue vs rest, preparation time, treating minor injuries.

You would have to mine and interpret your own data.

I agree with you that that's a factor. I was thinking about more macro level factors because I'm using around 24k data points. I will get back to you though with this analysis too.

 

[SnowblindNYR]

Ugh, I realized I screwed something up, but yes, days off plays a big role and actually a bigger one than points. I do want to make a subtle change that will answer the question I wanted to better.

 

[SnowblindNYR]

In the meantime an interesting bit of trivia.

The 05-06 LA Kings are the team that had the largest 2 game swing in the last 10 years. Game 46 they won 6-0 and game 47 they lost 10-1, a 15 point swing.

 

[SnowblindNYR]

Here it is with the days off accounted for. I even included the data this time (skipped it last time due to size issues).

 

[DearDiary]

If I said you're amazing, sexy and smart vs you're a piece of trash, how can a loser like you exist... Which situation would you play better in?

 

[SnowblindNYR]

If I said you're amazing, sexy and smart vs you're a piece of trash, how can a loser like you exist... Which situation would you play better in?

Is the purpose of this to show how psychology and thus momentum plays a role in sports? Or do you want me to do the research of how players perform when they're told they amazing, sexy, and smart, vs. a piece of trash?

 

[Hardyvan123]

Yes there is, no matter how many attempts to quantify it statistically fail.

pretty much this and this forum probably isn't the place for the question because hockey, like all sports, has an emotional base that isn't going to show up in large grouping of statistics that looks for trends or predictive behavior .

It's an eye test thing really.

Also momentum is more of an in game thing than a carry over one.

Each game does stand on it's own more or less but in game emotions and momentum can sweep teams (and players) up at times.

 

[Bear of Bad News]

pretty much this and this forum probably isn't the place for the question because hockey, like all sports, has an emotional base that isn't going to show up in large grouping of statistics that looks for trends or predictive behavior.

If people would like to discuss attempts (or provide attempts) to prove/disprove the existence of momentum in hockey, then yes, this is the place.

 

[Hardyvan123]

If people would like to discuss attempts (or provide attempts) to prove/disprove the existence of momentum in hockey, then yes, this is the place.

On second thought sure it's the place if one wants to try to look at it by numbers.

I'm just suggesting that momentum, especially in game momentum, is something that is going to be very hard to quantify much like leadership.

 

[Bear of Bad News]

Agreed that would be hard to quantify, at least with current data available.

It's worth exploring, though.

 

[SnowblindNYR]

In game momentum would be really hard to quantify. I was thinking of looking at goals scored consecutively in a short span of time. But a) what do you consider short and b) when does the momentum start and end? Almost impossible to quantify.

Not saying this is perfect, but I'm surprised at some of the complaints about an attempt to quantify something in the "By The Numbers" forum. I guess you don't have to think everything is quantifiable, but given how much work went into this, I would have appreciated something more constructive than "you'll never be able to quantify this, any attempt to do this is ********".

 

[Canadiens1958]

Sitting at Home Waiting

In the meantime an interesting bit of trivia.

The 05-06 LA Kings are the team that had the largest 2 game swing in the last 10 years. Game 46 they won 6-0 and game 47 they lost 10-1, a 15 point swing.

Yes, won 6-0 in Boston on a Thursday after playing on the road in San Jose and Anaheim the previous Saturday and Monday, then flying across the country against three time zones.

http://www.flyershistory.com/cgi-bin/hspgames.cgi

Buffalo was at home since December 30th 2005, rested. They had played at home the previous Saturday while the Kings were in San Jose, then at home Thursday against the Coyotes. They were definitely enjoying the benefits of a homestand, waiting for the Kings Saturday January 14.

 

[SnowblindNYR]

Yes, won 6-0 in Boston on a Thursday after playing on the road in San Jose and Anaheim the previous Saturday and Monday, then flying across the country against three time zones.

http://www.flyershistory.com/cgi-bin/hspgames.cgi

Buffalo was at home since December 30th 2005, rested. They had played at home the previous Saturday while the Kings were in San Jose, then at home Thursday against the Coyotes. They were definitely enjoying the benefits of a homestand, waiting for the Kings Saturday January 14.

Interesting, didn't realize that.

 

[Ohashi_Jouzu*]

Ok, but this is the forum for statistics no? How else should I try to analyze it in a forum called "By The Numbers"?

By starting off with the premise "Given that momentum exists is sports" (and you don't have to take my word for it, as academia has covered it plenty in subjective/psychological terms) and continuing with the question "will it ever be possible to quantify it, and how?".

As opposed to starting off wondering if it exists at all, that is. It's like questioning if oxygen exists, because it can't be seen. Everyone who has ever been on the ice has felt it, been aware of it. Academics who have studied and consulted athletes for the purpose of research acknowledge it (qualitatively, at least). It's only those who haven't or watch externally who question it, and it's simply because of the difficulty in "measuring" it (function of the group? function of the individual(s)? function of the circumstances? function of all four, or more factors? etc, etc). Is it mostly relevant in terms of which players who are able to "tap into it" most constructively to help their team? Do players who can't constructively "tap into" the momentum swing hinder the group's ability to turn it into something measurable to statisticians?

I mean, you could go on and on. But you have to start with the opening premise that it DOES exist, imo.

 

[Czech Your Math]

This paper may (or may not) be helpful to you at some point:

Win Probabilities

It might be better to use each team's GF/GA ratio (excluding SO goals... perhaps OT goals as well) in your model:

A = GF/GA ratio team A
B = GF/GA ratio team B
E = pythagorean exponent

Expected Win % of Team A vs. Team B = (A^E) / (A^E + B^E)

However, it doesn't seem so easy to predict the probability of a team winning by (at least) a certain margin.

As you must know, it generally comes down to trying to include the other variables that may significantly influence the dependent variable (in this case, team winning % in a single game) and that are relatively easy to measure. Binary variable for Home/Road, discrete variables for rest for each team (0 days, 1 day, 2+ days), expected win% using some version of pythagorean formula, etc., in addition to variable(s) chosen to represent "momentum."

You might also try doing a separate regression (and/or individual correlations) using different variables for "momentum" (e.g., GF/GA ratio in past Z games, team points in past Z games... using and possibly discarding various values for Z). In the case of team points, it may be best to count OT/SO games as ties worth one point for each team, for both the dependent variable and independent variable(s). Once you identify seemingly significant variable(s) for "momentum" via correlations (or some separate regressions), then you can use the best of those variables in the larger regression and hopefully it/they will have a greater chance of staying significant in the larger study.

 

[Canadiens1958]

Main Focus

Interesting, didn't realize that.

Realize that you titled based on momentum but what is the main focus?

Scheduling, strategies - player selection and momentum related(how to gain or neutralize momentum), scoring overall, etc.

Besides curiosity would help directing replies in the proper direction.

 

[hatterson]

By starting off with the premise "Given that momentum exists is sports" (and you don't have to take my word for it, as academia has covered it plenty in subjective/psychological terms) and continuing with the question "will it ever be possible to quantify it, and how?".

As opposed to starting off wondering if it exists at all, that is. It's like questioning if oxygen exists, because it can't be seen. Everyone who has ever been on the ice has felt it, been aware of it.

I don't think anyone would deny that the feeling of momentum exists, but it appears the question the OP is attempting to answer is not "Do players feel momentum?" but rather "Does momentum have a real and tangible impact on games?"

Perhaps the OP worded his title in a sub-optimal way, but I think it's clear what he's looking for and semantic arguments don't really help anyone improve their knowledge of the game.

 

[Ohashi_Jouzu*]

I don't think anyone would deny that the feeling of momentum exists, but it appears the question the OP is attempting to answer is not "Do players feel momentum?" but rather "Does momentum have a real and tangible impact on games?"

Perhaps the OP worded his title in a sub-optimal way, but I think it's clear what he's looking for and semantic arguments don't really help anyone improve their knowledge of the game.

But whether or not it exists is not a matter of semantics, that's the point. How you go about trying to find sources for numbers to measure it, sure.

 

[SnowblindNYR]

Realize that you titled based on momentum but what is the main focus?

Scheduling, strategies - player selection and momentum related(how to gain or neutralize momentum), scoring overall, etc.

Besides curiosity would help directing replies in the proper direction.

Well I started off with a question of whether teams that get blown out are significantly more likely to "bounce back" next game because they were embarrassed. A weak hypothesis and the opposite of momentum. Thinking about it now, it was silly. But in my data I found that blowout losses leads other blowout losses and the same for wins. So my thought was, is this a product of both positive and negative momentum.

To the point of psychological studies. I honestly don't put much credence to interviews or whatever about whether players feel there's momentum. It basically puts too much weight on people's intuition and memory that can be deceiving. There's a reason qualitative research isn't quantified because it's just a guide rather than scientific. I made a pretty primitive study here and it may well be that momentum does exist, but I don't think psychological studies hold much water.

 

[Ohashi_Jouzu*]

Well I started off with a question of whether teams that get blown out are significantly more likely to "bounce back" next game because they were embarrassed. A weak hypothesis and the opposite of momentum. Thinking about it now, it was silly. But in my data I found that blowout losses leads other blowout losses and the same for wins. So my thought was, is this a product of both positive and negative momentum.

To the point of psychological studies. I honestly don't put much credence to interviews or whatever about whether players feel there's momentum. It basically puts too much weight on people's intuition and memory that can be deceiving. There's a reason qualitative research isn't quantified because it's just a guide rather than scientific. I made a pretty primitive study here and it may well be that momentum does exist, but I don't think psychological studies hold much water.

Why not? I submit that the psychological side/approach/impact is a HUGE determining factor in why games aren't pre-decided on paper. The problem you're going to encounter is that momentum measured at an individual level is almost unmeasurable without real-time CT scans on the players monitoring brain activity, and on a team level things like chemistry and synergy (also non-quantifiable to statistics, but very real factors) are going to play a huge part in whether or not momentum manifests itself in anything quantifiable to stats. Coaches, nevertheless, preach about things like "building off a good shift/PK/whatever", "getting mojo back", "breaking out of a funk", hot/cold streaks, etc. And that's because, if "harnessed", momentum is something that you can build on - even if "only" psychologically.

 

[SnowblindNYR]

Yes, won 6-0 in Boston on a Thursday after playing on the road in San Jose and Anaheim the previous Saturday and Monday, then flying across the country against three time zones.

http://www.flyershistory.com/cgi-bin/hspgames.cgi

Buffalo was at home since December 30th 2005, rested. They had played at home the previous Saturday while the Kings were in San Jose, then at home Thursday against the Coyotes. They were definitely enjoying the benefits of a homestand, waiting for the Kings Saturday January 14.

Ok, so I have both games when a team has a home game after a 3 day or more break and when a team has a home game after a 3 day or more break AND their opponent is playing a back to back on the road. Both are significant. The latter has a higher coefficient, but also a higher p-value, though both are way too many Standard Errors away for it to matter. For each regression I used a column that was simply binary, 1 if it met the criteria, 0 if it didn't.

Variable Coefficient Std. Error t-Statistic Prob.

HOME_AFTER_LONG_LAYOFF 0.366756 0.043587 8.414319 0
C -0.056009 0.016852 -3.323594 0.0009

Variable Coefficient Std. Error t-Statistic Prob.

HOME_AFTER_LONG_LAYOFF_&_OPP_B-TO-B 0.545389 0.085958 6.344851 0
C -0.01848 0.015823 -1.167946 0.2428

Sorry I don't know how to make this look nicer. Also, there's a chance, though not a great one that I screwed up on the Opp B-To-B. It's harder to double check since my data is mostly predicated on the home team and I was using a nested If function that I'm not a pro at my any means.