Another playoff conundrum

[Stoner]

This is right up carp's analytical and statistical alley. Yes, it's a dark and forboding alley.

 

If most teams that score the first goal of a game win the game, and most teams that win the first game of a series win the series, how do you measure the importance of scoring the first goal of a series?

 

Does the first-scorer win the series 60 percent of the time? 55% 51% 50.1% Even if it's a relatively small advantage, say 53-47, shouldn't that mean that coaches should spend a lot of time figuring out how to score that first goal?

[Canadian Sabre]

A team would have to essentially crawl into a hole and die if the first goal of a series effected the outcome of the series in any way. Let's assume Boston finish eighth in the East and play the Caps in the first round, with Washington obviously being at home first. A goal scored by Blake Wheeler at 1:21 of the first period is going to have little effect on a team as explosive as the Capitals.

 

The playoffs are a different story because you're possibly playing the same team seven times in a row. The more you play a team with the same players, the more you begin to learn their tendencies.

[Stoner]

A team would have to essentially crawl into a hole and die if the first goal of a series effected the outcome of the series in any way. Let's assume Boston finish eighth in the East and play the Caps in the first round, with Washington obviously being at home first. A goal scored by Blake Wheeler at 1:21 of the first period is going to have little effect on a team as explosive as the Capitals.

 

The playoffs are a different story because you're possibly playing the same team seven times in a row. The more you play a team with the same players, the more you begin to learn their tendencies.

 

I'm not alluding to the emotional impact of getting scored on. Just the known probabilities that if you score first you usually win the game and if you win the first game of a series you usually win the series.

[fallen627]

A team would have to essentially crawl into a hole and die if the first goal of a series effected the outcome of the series in any way. Let's assume Boston finish eighth in the East and play the Caps in the first round, with Washington obviously being at home first. A goal scored by Blake Wheeler at 1:21 of the first period is going to have little effect on a team as explosive as the Capitals.

 

The playoffs are a different story because you're possibly playing the same team seven times in a row. The more you play a team with the same players, the more you begin to learn their tendencies.

 

I'm not sure if a team's response to an opponent scoring the first goal is relevant to his statistical question, which should be true if the his first and second premise are valid.

 

@ PA, Great question, man. I'm really curious to find this out myself.

[fallen627]

I'm not alluding to the emotional impact of getting scored on. Just the known probabilities that if you score first you usually win the game and if you win the first game of a series you usually win the series.

 

beat me to it! :oops:

[Stoner]

beat me to it! :oops:

 

No problem, thanks for the backup.

[Canadian Sabre]

I would assume that what you're alluding to has to have some sort of mental impact on a team. Running into a hot goaltender, missing some of your key players, coming up against great defencemen etc.. would get anybody frustrated after a while.

[Canadian Sabre]

I'm not sure if a team's response to an opponent scoring the first goal is relevant to his statistical question, which should be true if the his first and second premise are valid.

 

@ PA, Great question, man. I'm really curious to find this out myself.

 

I understand what he's asking, I just don't think the first goal of a series is overly important.

[carpandean]

Just the known probabilities

Probabilities are rarely, if ever, known; that's why we need/use statistics. We merely infer something about the probabilities from the known statistics.

[Stoner]

Probabilities are rarely, if ever, known; that's why we need/use statistics. We merely infer something about the probabilities from the known statistics.

 

Mighty Casey has struck out?

[Stoner]

This is impressive.

 

In 16 playoff series involving the Sabres going back to 1998, the team that has scored first has won the first game and the series 12 times. Another time, the team scored first, lost the game and won the series. So, 13 of 16 times the team scoring first won the series.

 

Doesn't prove anything, but now I REALLY want the first goal next Wednesday or Thursday night!

[Canadian Sabre]

This is impressive.

 

In 16 playoff series involving the Sabres going back to 1998, the team that has scored first has won the first game and the series 12 times. Another time, the team scored first, lost the game and won the series. So, 13 of 16 times the team scoring first won the series.

 

Doesn't prove anything, but now I REALLY want the first goal next Wednesday or Thursday night!

 

Shhh don't jinx it! :D

[SabresRepublic]

This is right up carp's analytical and statistical alley. Yes, it's a dark and forboding alley.

 

If most teams that score the first goal of a game win the game, and most teams that win the first game of a series win the series, how do you measure the importance of scoring the first goal of a series?

 

Does the first-scorer win the series 60 percent of the time? 55% 51% 50.1% Even if it's a relatively small advantage, say 53-47, shouldn't that mean that coaches should spend a lot of time figuring out how to score that first goal?

 

I am certain they do!

 

Why wait until the other team scores first before scoring one themselves!

 

The name of the game is to score goals.

 

But if you wish to bring statistical hubris into play here, I would be curious to see how many games in the playoffs have ended 1 - 0?

[Stoner]

I am certain they do!

 

Why wait until the other team scores first before scoring one themselves!

 

The name of the game is to score goals.

 

But if you wish to bring statistical hubris into play here, I would be curious to see how many games in the playoffs have ended 1 - 0?

 

I most heartily BELIEVE you have the skill and aptitude to look this up yourself! I do!!!

[Eleven]

What is "statistical hubris"?

[RuffMuff]

no, because the first team to score is just usually the better team. If you work to just be the first team to score, that in itself does not give you a better edge to win a series

 

But I like 2006 series were it seemed that we always scored first and won too!

[Stoner]

no, because the first team to score is just usually the better team. If you work to just be the first team to score, that in itself does not give you a better edge to win a series

 

That's a pretty good point.

 

I still think I've got something here.

[Midnight]

If most teams that score the first goal of a game win the game, and most teams that win the first game of a series win the series, how do you measure the importance of scoring the first goal of a series?

 

Here's a plot showing the effects of scoring the first goal in an NHL playoff series. I hope it's somewhat self explanatory.

 

 

The curves are estimates, obtained using the last 4 season's data. The observed effect of scoring the 1st goal (the difference between the red and green curves) is not formally "statistically significant," but is suggestive. In addition, it is reasonable to assume that scoring the first goal improves a team's chances of winning the series by at least a small amount. It'd be interesting to funnel in more data. (That was the slow part - I quit at the strike season.)

 

I used regular season scoring for competing teams in a series (both goals for and against) to determine the probability that the home team scores "the next goal." There are other ways to do it: Mine is, I think, acceptable. I can justify it. This variable serves as a proxy for seeding (the two are positively correlated). It doesn't take into account any late-season trending, nor any particular match-up tendencies (for example, Sabres tend to lose vs. Senators). Those things are difficult to model.

 

So don't lose heart when the Bruins score 1st. The Sabres still have a better than even chance (at least as estimated here) of winning the series.

 

Here are figures for the other 7 series. These are not probabilities that a team wins a game - they are probabilities the team scores the next goal. Use this with the black curve to find the pre-series probability that team wins the series. Compare to Vegas odds! (Except that they manipulate the odds to ensure profits.) Adjust as the first goal is scored

 

Caps: 0.545

Devils: 0.513

Sabres: 0.514

Penguins: 0.514

Sharks: 0.521

Blackhawks: 0.523

Canucks: 0.514

Coyotes: 0.500

[Stoner]

I'm impressed someone would take my question and put so much thought into a reply! I will study this in the AM when my eyes and brain aren't so blurry.

[carpandean]

Here's a plot ...

You complete me.

[billsrcursed]

You complete me.

 

:w00t:

[Stoner]

You complete me.

 

Your trend is pointing straight up.

[Stoner]

Here's a plot showing the effects of scoring the first goal in an NHL playoff series. I hope it's somewhat self explanatory.

 

 

The curves are estimates, obtained using the last 4 season's data. The observed effect of scoring the 1st goal (the difference between the red and green curves) is not formally "statistically significant," but is suggestive. In addition, it is reasonable to assume that scoring the first goal improves a team's chances of winning the series by at least a small amount. It'd be interesting to funnel in more data. (That was the slow part - I quit at the strike season.)

 

I used regular season scoring for competing teams in a series (both goals for and against) to determine the probability that the home team scores "the next goal." There are other ways to do it: Mine is, I think, acceptable. I can justify it. This variable serves as a proxy for seeding (the two are positively correlated). It doesn't take into account any late-season trending, nor any particular match-up tendencies (for example, Sabres tend to lose vs. Senators). Those things are difficult to model.

 

So don't lose heart when the Bruins score 1st. The Sabres still have a better than even chance (at least as estimated here) of winning the series.

 

Here are figures for the other 7 series. These are not probabilities that a team wins a game - they are probabilities the team scores the next goal. Use this with the black curve to find the pre-series probability that team wins the series. Compare to Vegas odds! (Except that they manipulate the odds to ensure profits.) Adjust as the first goal is scored

 

Caps: 0.545

Devils: 0.513

Sabres: 0.514

Penguins: 0.514

Sharks: 0.521

Blackhawks: 0.523

Canucks: 0.514

Coyotes: 0.500

Hey midnight, I guess it wasn't the late hour. Would you humor me and other dolts on the board and kind of start from the beginning and type r e a l s l o w ?

[MattPie]

no, because the first team to score is just usually the better team. If you work to just be the first team to score, that in itself does not give you a better edge to win a series

 

It's be interesting to compare the 'mismatches' and the more evenly matched. For example, in 3-6 and 4-5 series' compared to 1-8 and 2-7 matchups. I could see the even matches depending more on the first goal, or the mismatches showing a strong trend because the good teams will score first.

[Midnight]

Hey midnight, I guess it wasn't the late hour. Would you humor me and other dolts on the board and kind of start from the beginning and type r e a l s l o w ?

Well - let's take the Sabres v. Bruins. The Sabres are the Home team for Game 1. For this series I put the Sabres' (Home team's) chance at scoring the first goal at 0.514. That's x = 0.514. So, run your finger vertically until you reach the...

 

...Black curve: Before the game starts, the Sabres have a 0.64 = 64% chance of winning the series.

 

Now, suppose you're watching and the Sabres score the first goal of the game/series. Now go vertically over x = 0.514 until you reach the...

 

...Green curve: the Sabres have (improved their) chance of winning the series to 0.71 = 71%.

 

Now, suppose you're watching and the Bruins score the first goal of the game/series. Now go vertically over x = 0.514 until you reach the...

 

...Red curve: the Sabres have (reduced their) chance of winning the series to 0.54 = 54%.

 

If you look at the Black curve, there's a reference point. If the two teams playing have had, over the regular season, equal ratios of goals scored to allowed, then the probability of the Home team scoring the first goal is set at 0.50 (how could it be anything else?). For such a situation, the Black curve tells us that the chance of the Home team winning the series is (go vertically from x = 0.5) 0.50! Note that this did not "have to happen" when I fit these curves. That it did confirms what common sense dictates and indicates that in some sense the data produce a not unrealistic curve. (With more data one might tease this out to be slightly greater than 0.50 - if there really exists a "Home field advantage." It's going to take a lot of data to do that. That's because this effect is really somewhat covered by the "chance of scoring the next goal" on the x-axis. Home teams for game 1 tend to have a better chance of scoring the first goal going in, because Home team for game 1 is assigned on seeding, which largely comes about by scoring more and giving up fewer goals. That's why the x values run from only 0.45 - its rare to find a Home team for game 1 that has worse scoring stats than the visitors.)

 

OK: This is too much, but here goes...

 

In the 2007 playoffs New Jersey faced Ottawa. New Jersey was Home team (for game 1). Over the regular season, New Jersey had scored 206 goals and given up 193; Ottawa had scored 286 and given up 213. While the Devils managed to get seeded higher, the data suggests that Ottawa has a better chance of scoring the "first" goal. (Forget about Brodeur in the playoffs, etc. I can't model that.) This is the smallest x in the data: x = 0.473. (Anything below is really an extrapolation.) Therefore (go vertically to the Black) New Jersey's chance of winning the series is set at about 28% before the puck is dropped. Ottawa went on to score the first goal, dropping New Jersey's chances to about 22%. Of course Ottawa went on the win the series (4-1). (Which proves nothing, I might add. There's uncertainty in all this.)

 

The previous year Ottawa was Home to Tampa Bay. Ottawa was again, statistically, way better: x = 0.557 (the highest value in my data, so anything beyond is again an extrapolation). Before the puck was dropped, Ottawa had 92% chance of winning the series. After Tampa Bay scored the first goal, Ottawa's chance dropped to about 88%. Ottawa won the series 4 - 1. Again proving nothing, but suggestive of how it works.