Why pay attention to game winning goals?

[plusandminus]

But the "same value" that you mentioned isn't true. A GTG 2 minutes into the first period has very different value than a GTG with 2 second left in the third period.

I agree. A GTG with 2 seconds left of regulation is very important.

But on the other hand. Since most seems to agree that the 1st goal of the game is very important (especially in the playoffs), a quick equalizer would erase that advantage and thus change the game dramatically - sort of "back to zero" again.

And what happens when the next time one of the teams score? Then we are back again to a one-goal lead. I think in games where in the end both teams are within two goals, every goal is important.

Exceptions would be big wins. But those are not as common these days than they used to be. (In the Bobby Clarke vs Bobby Orr thread, it is being mentioned that Orr's Boston had many huge wins. Gretzky would be another example of someone playing seeing big wins, while also Gretzky seemingly being very keen to produce points even in those "already settled" games.)

I think GWG should be exchanged for better stats. 1st goals would be better. Or more sophisticated stats.


By the way, in tennis there was a commentator some decades ago that used to speak about "the important 7th game" of the sets. But I and others tended to almost make fun about that, while thinking that every game is about as important. It doesn't matter if you break the opponent's serve in the 1st, 3rd, 5th, 7th or 9th game, as long as you keep your own serve in the 2nd, 4th, 6th, 8th and 10th.
(No we got some tennis too, as a balance to all the baseball talk. )

You can give value to goals by when they occur and how they change the score by comparing it with average won-loss percentages that vary on time line and goal differential.

You are welcome to give examples.

 

[seventieslord]

But on the other hand. Since most seems to agree that the 1st goal of the game is very important (especially in the playoffs), a quick equalizer would erase that advantage and thus change the game dramatically - sort of "back to zero" again.

And what happens when the next time one of the teams score? Then we are back again to a one-goal lead. I think in games where in the end both teams are within two goals, every goal is important.

I agree with this big time. There are garbage time goals which account for maybe 15% of all goals... and then there are the rest, which are all pretty important goals.

 

[Iain Fyffe]

But the "same value" that you mentioned isn't true. A GTG 2 minutes into the first period has very different value than a GTG with 2 second left in the third period.

Same average value.

If you're going to get into measuring the time at which these important goals were scored, you're getting into advanced stats rather than columns in a player's stats summary.

 

[Hockey Outsider]

You are welcome to give examples.

I think it would be interesting to look at "probability added" goals and assists.

Let's assume that a team is expected to earn 1.00 points in the standings if a game is tied 0-0, ten minutes into the first. (For simplicity I'm assuming no points for losing in overtime... and since it's my example, there are no shootouts!) If a goal is scored ten minutes into the first to break a 0-0 tie, the team scoring the goal might, on average, be expected to earn 1.50 points in the standings (since they're more likely to win than their opponents, but it's far from a guarantee as there's a lot of time for them to come back). So the goal added 0.40 points to the team's expected record.

As another example, let's suppose a team is down 4-3 with 1 minute to go in the third. They're expected on average to earn say 0.10 points in the standings (as most times they'd lose but occasionally they'd make it into OT). If someone scores to tie the game with 1 minute left, the goal might be worth 1.0 points in the standings (there's a small chance they either give up another goal quickly and lose, or score quickly and win, but most likely they last into OT and have a 50% chance at winning and earning 2 points). This goal goal added 0.90 points to the team's expected record.

Another example - suppose a team is winning 5-1 halfway through the second. They're expected on average to earn say 1.9 points in the standings (as the probability of them losing at that point is extremely small). If a player on that team scores again to make it 6-1 halfway through the second, his team might be expected to earn 1.95 points in the standings - the goal is worth very little (0.05 points) because it had virtually no impact on the outcome of the game.

In order to actually calculate this you'd need a very large sample of data, or you can do calculations based on the theoretical probability. I'd imagine that, in practice, it would be extremely difficult to calculate this, but I think it could highlight the relative importance of each goal/assist, without relying on the flawed game-winning goal statistic.

 

[doktor2d]

Same average value.

If you're going to get into measuring the time at which these important goals were scored, you're getting into advanced stats rather than columns in a player's stats summary.

There seem to be two requests in this thread ... a quality stat for a column in a player's summary and a true measure of value of goals.

For the first, a redesigned game-winning goal is the answer. I'm not sure why a first goal would be better. A baseball-style GWG where the GWG is the one that puts a team on top for good would include first goals in games where that team didn't surrender the lead.

But why would a first goal be a more valuable stat than a GWG? If a team goes up 3-0, then gets into a 3-3 tie, then scores with 10 seconds left to win 4-3 ... that first goal isn't as meaningful as the fourth goal.

I think there's room for both first goal and GWG in a player summary. But a baseball-style GWG is more important because it includes the most important first goals.

 

[LeBlondeDemon10]

The career leaders of GWG's isn't exactly filled with no names. Most posters here seem to be underestimating the value of the GWG.

 

[Bear of Bad News]

The career leaders of GWG's isn't exactly filled with no names. Most posters here seem to be underestimating the value of the GWG.

That's because if you score a lot of goals, you're going to score a lot of game-winning goals. It has nothing to do with the "value" of the GWG.

 

[Ohashi_Jouzu*]

I'd rather see a statistic like "tie-breaking goals".

The 4th goal of the night is hardly the "clutch" if you go up 8-0 in the 1st and the losing team scores 3 late in the 3rd.

Tie breaking goals makes a bit more sense--and first goals of the night would always be considered tie-breaking. If it is 3-3 and someone scores to make it 3-4--that'd be another tie breaking goal.

I kinda like this. Most importantly, it accounts for the importance of the first goal in the game (25/30 teams had a record of 0.600+ when scoring first, and no team in the league was above 0.500 when trailing first, so it's obviously an important goal). Unfortunately, it doesn't account for the importance of tying goals which are obviously really important for a team trying to mount a comeback/reset the "advantage". The way I see it, if a tie-breaking goal is an important enough stat to track, then so is a goal that effectively wipes it off the stat sheet.

Maybe the "clutch" stat could be restricted to these two types? I don't know if it would provide anything overly useful when comparing players from different teams (similar problem with +/-), but within a team it might be interesting. Maybe they could just track 1st goals of games in wins? I dunno, it's too easy to come up with different approaches, and too difficult to precisely determine what you're actually going to "measure" in the end.

 

[BM67]

Teeder Kennedy won 5 Cups with the Leafs. Teeder scored 12 of Toronto's 65 goals. The five Cup winning goals were scored by five different players, including Teeder. Teeder had 5 GWG in the 5 Cup Finals, for 25% of Toronto's GWGs in the Finals.

Is Teeder clutch or is this just random noise?

 

[plusandminus]

Teeder Kennedy won 5 Cups with the Leafs. Teeder scored 12 of Toronto's 65 goals. The five Cup winning goals were scored by five different players, including Teeder. Teeder had 5 GWG in the 5 Cup Finals, for 25% of Toronto's GWGs in the Finals.

Is Teeder clutch or is this just random noise?

If I understand you right... Toronto won 20 games. Teeder scored 18.46 % (12/65) of all the goals, including 25 % of (5/20) the GW ones. He scored 20 % (1/5) of the cup winning goals.

I don't know of this guy, so I don't know about his usual performances during regular season and playoffs. If he scored much more in the finals than otherwise, I think he is "clutch" (I'm no expert on the term). But that would have little to do with his GWG goals. It would rather have to do with him scoring lots of goals in itself.
His GWG numbers looks nothing particular. Small sample too.

Perhaps he scored lots of 1st goals?

I haven't seen any research regarding if some players consistantly tends to "save" their goals for big occasions or moments. If they do, one could perhaps wonder why they do.

 

[Bear of Bad News]

Teeder Kennedy won 5 Cups with the Leafs. Teeder scored 12 of Toronto's 65 goals. The five Cup winning goals were scored by five different players, including Teeder. Teeder had 5 GWG in the 5 Cup Finals, for 25% of Toronto's GWGs in the Finals.

Is Teeder clutch or is this just random noise?

If I'm reading this correctly, Kennedy scored 18% of Toronto's Cup Finals goals (were those 12 and 65 for just the Finals) and 25% of Toronto's GWG in the Cup Finals.

That seems like statistical randomness to me.

 

[plusandminus]

Another way of looking at this might be the following:

Take the numbers of goals needed to win the game, and include up to one "insurance" goal too since they are often important. Examples:
3-2=3, 6-1=4, 2-1=2, 3-1=3, 5-2=4 (if a 3-2 game, then 3 goals, and so on)
Give the scorers of the winning team an equal share of GW "points". If three players contributed to a victory, then give them 0.67 each. If four contributed, give them 0.50 each. If Jagr scored 3 goals in a 5-4 victory, give him 3/5*2=1.2. There is always 2 points handed out.

For draws (and overtime losses?), do the same but multiply by 1 instead of two. For example, Colorado-Vancouver 2-2, goal scorers Forsberg, Hedjuk, Naslund, Bertuzzi. Give them 0.5 each.
(I know draws don't occur nowadays, but anyway.)

However, I'm not very interested in this. Basically, I just think things will probably even themselves out. The great scorers are probably also the great GWG (or whatever similar technique) scorers.
Being on winning teams, especially high scoring ones (but not too superior), will help.

I'm not a big fan of the "point shares" stat neither, that is sometimes being used.

 

[Bear of Bad News]

Bill James did something like that (for RBI/GWRBI) in one of his Baseball Abstracts.

If it went anywhere after that, I'm not aware of it.

 

[plusandminus]

By the way... If measuring goals, then why not measure the importance of the wins too? Surely, all the wins of a team of a team going to the playoffs with a margin of 1 point is being more important than the wins of a team winning President's trophy by 7 points. And shouldn't wins against biggest rivals count more? And should perhaps wins during the last month of the season count more than wins early in the season?

Which teams had most "extra important" wins?
Should players on teams not reaching the playoffs have their GW stats removed? After all their goals were just worthless. It perhaps even had been better if they hadn't scored at all, since that would have given their team a bigger chance in the draft lottery.

I think ideally one might consider things like these too. But in reality, again, I think it's enough to just focus on "which teams gathered most points during the whole season", and "which players gathered most goals or points during the whole season".

There would otherwise be so many "what if":s. If a team misses playoffs by 1 point, was it because player X hit the goal post with 18 seconds left (during a then 2-2 tie) during that November game, leading to a turnaround for the other team which scored with 6 seconds left? So many "what if":s.

 

[SmellOfVictory]

I'd rather see a statistic like "tie-breaking goals".

The 4th goal of the night is hardly the "clutch" if you go up 8-0 in the 1st and the losing team scores 3 late in the 3rd.

Tie breaking goals makes a bit more sense--and first goals of the night would always be considered tie-breaking. If it is 3-3 and someone scores to make it 3-4--that'd be another tie breaking goal.

Not bad, not bad at all. I don't know whether there would be statistical significance to that or not, but logically it feels like it could.

 

[BM67]

In the 5 Finals, Toronto scored the first goal of the game 12 times, with 11 assists awarded. Kennedy scored the first goal 3 times, and assisted on another 4. Sid Smith also scored 3, two of them assisted by Kennedy, and assisted on none. Bentley scored one, and assisted on another. The other 11 points went to 11 different players.

 

[Iain Fyffe]

That's because if you score a lot of goals, you're going to score a lot of game-winning goals. It has nothing to do with the "value" of the GWG.

Beat me to it.

If you look at the career leaders in "goals scored after the game was well in hand", the list would also be dominated by the best goal-scorers. Doesn't mean it's a stat worth keeping.

 

[reckoning]

I've been doing a lot of research in the 74-75 season over the last few months, and one of the stats I did look at were "blowout points"; just to see if any players were racking up an inordinate amount of their points late in one-sided games. I used a very conservative definition of that as being any point a player scored in the third period while his team was ahead by 4 goals or more. The top "blowout point" getters for 74-75 were:

Phil Esposito|14|
Peter Mahovlich|9|
Bobby Orr|8|
Rene Robert|8|
Larry Robinson|8|
Steve Shutt|8|
Steve Vickers|8|

And for a comparison I jumped ahead 10 years for the 84-85 season:

Peter Stastny|9|
Tim Kerr|8|
Michel Goulet|7|
Paul Coffey|6|
Mike Gartner|6|
Wayne Gretzky|6|
Dale Hunter|6|
Larry Murphy|6|
Dave Poulin|6|

I don't know if this is anything worth pursuing further. One variable that would have to be factored in would be the fact that players on top teams will obviously have more opportunities to rack up these points than players on weaker teams would.

 

[Canadiens1958]

Developmental Time

I've been doing a lot of research in the 74-75 season over the last few months, and one of the stats I did look at were "blowout points"; just to see if any players were racking up an inordinate amount of their points late in one-sided games. I used a very conservative definition of that as being any point a player scored in the third period while his team was ahead by 4 goals or more. The top "blowout point" getters for 74-75 were:

Phil Esposito|14|
Peter Mahovlich|9|
Bobby Orr|8|
Rene Robert|8|
Larry Robinson|8|
Steve Shutt|8|
Steve Vickers|8|

And for a comparison I jumped ahead 10 years for the 84-85 season:

Peter Stastny|9|
Tim Kerr|8|
Michel Goulet|7|
Paul Coffey|6|
Mike Gartner|6|
Wayne Gretzky|6|
Dale Hunter|6|
Larry Murphy|6|
Dave Poulin|6|

I don't know if this is anything worth pursuing further. One variable that would have to be factored in would be the fact that players on top teams will obviously have more opportunities to rack up these points than players on weaker teams would.

A variation of what you looked at would be developmental time. How many points were scored by players who were rookies or prospects trying to establish themselves in a role or position.

1974-75 you have two such players. Steve Shutt and Larry Robinson who had significant upward point improvements compared to 1973-74 but needed the extra playing time in blow-outs to build confidence and improve.

This could also be found on losing teams in blow-outs. Rookies / prospects getting points late in such games.

 

[seventieslord]

A variation of what you looked at would be developmental time. How many points were scored by players who were rookies or prospects trying to establish themselves in a role or position.

1974-75 you have two such players. Steve Shutt and Larry Robinson who had significant upward point improvements compared to 1973-74 but needed the extra playing time in blow-outs to build confidence and improve.

This could also be found on losing teams in blow-outs. Rookies / prospects getting points late in such games.

As usual, finding something in nothing.

There was one clear "winner" in this category, Esposito. Then one guy alone in 2nd, and then "the pack".

There is nothing significant to be gleaned from looking at who had 8 blowout points, despite your attempts.

That five of the top 7 were either habs or bruins tells me all I need to know.

 

[DJ Man]

Bill James did something like that (for RBI/GWRBI) in one of his Baseball Abstracts.

If it went anywhere after that, I'm not aware of it.

Bill James had a statistic he called VI-RBI, or Victory-Important Runs Batted In. It was pretty simple. I'll quote from James' 1985 Baseball Abstract.

(1) Only RBI in wins are counted as victory-important.
(2) The importance of each RBI in a victory is measured by:
(Opposition Runs + 1.00) / Runs Scored

That is, if the team wins by one run (4-3, 7-6, 1-0), then each RBI is essential to the win, and each RBI has an "RBI importance" of 1.00. But if you win 10-3, then the importance of each RBI is only .40.
(3 + 1) / 10 = .40
And if the team wins 16-1, then the importance of each RBI is only .125
(1 + 1) / 16 = .125

Whatever applies to runs should have some relevance to goals. Maybe his formula would work better for hockey. (You can produce runs four at a time with a grand slam, so maybe insurance runs are more important than insurance goals, but he doesn't give extra credit for the fifth run of a lead.)

 

[Hardyvan123]

in basketball, they have a useful stat for "crunch time scoring." i think it means the last five minutes of a game, if the score of the game is within five points.

so that would be useful if translated to hockey. the first goals of a game are also really important. i'd like to see a stat or that.

This would be usefull if king James actually knew what clutch meant as well eh?

 

[Bear of Bad News]

Bill James had a statistic he called VI-RBI, or Victory-Important Runs Batted In. It was pretty simple. I'll quote from James' 1985 Baseball Abstract.

That's exactly what I was thinking of - thanks!

(I should have guessed it was in 1985 edition - that's my all-time favorite. I still remember programming the Brock-2 algorithms into my Apple II+).

 

[DJ Man]

That's exactly what I was thinking of - thanks!

(I should have guessed it was in 1985 edition - that's my all-time favorite. I still remember programming the Brock-2 algorithms into my Apple II+).

I did those formulas on my Commodore 64 ... and a CDC Cyber supercomputer at work as well!

 

[Guest]

I used to be a big fan of the GWG, but after time I agree it doesn't really capture as much as you would like it to.

One thing that I have been looking to more in recent years has been the points a player accrues in wins versus losses. I've had a lost of resistance in valuing this stat, but I think it's a good indicator to see what players produce to drive team success versus what players get that 1 goal in a 6-1 loss. Again, you can typically guess that players who score a lot of points will score more points in wins, but like most statistics you need to dig a little deeper. Some players perform nearly equally offensively in wins and losses which could be interpreted that they are consistent or that they are a non-factor. There are lots of ways you can interpret it.

Similar to that I like to see goalie stats in wins versus losses. Again, I think you can tell a lot about the goalies performance in those stats. If the losses are still pretty respectable numbers you could derive that the goalie is keeping his teams in games. If the numbers in losses are quite inflated compared to his winning stats you could consider that goalie may be inconsistent or when he's off her really blows the game.

These stats can produce meaning in the wins as well, although I've mentioned the losses first. Goalies that have great winning stats (low GAA + SV%) likely carry their team but you could say that they play in a great defensive team.