2 goal lead ("the worst lead in hockey")

[Hockey Outsider]

Commentators often say that a 2 goal lead is the worst lead in hockey. The most common explanation is players will become complacent (despite not really having a large enough lead to justify that complacency). The other explanation I've heard is it creates confusion, because a team isn't sure if they should attack further, or defend the lead. Regardless, the theory is a team will give up a 2 goal lead more often than they "should" as a result.

Has anyone studied this phenomenon? Is there any evidence that 2 goal leads are less secure than one would expect?

Click to read more...

 

[Doctor No]

I don't have perfect data on this (I have the score recorded at each intermission for NHL games back through 1999-2000, and scattered data prior), but this is something I've been interested in looking at. I'll put it on the short list if no one has it ready.

 

[Johnny Engine]

Some observations on this not backed by any kind of data:

- It's a pet peeve of mine when I hear 3-0 referred to as "the worst lead in hockey", and I think that's getting more common. I don't think that's a real thing. Stop saying it, people.
- It's also a pet peeve when commentators/analysts/social media personalities get into the smirky "of COURSE a one goal lead is worse than a two goal lead!" routine. Not only do I find it kind of vaccuous and not an interesting point, but I also think it's (deliberately?) missing the point, which is...
- I think it's specifically a sustained 2-goal lead that goes on throughout most of the middle portion of the game. For instance, Team A scores 2 quick ones by the 7-minute mark of the first period, and by the end of the second they're still up 2-0. When that second goal went in, there was no reason to worry about their "worst lead", because the pucks are going in and everyone's rolling. However, going into the third, there's an (irrational?) feeling that if Team B scores at any point in the next 17 minutes or so, Team B will lose the game, because it sets up a big momentum swing for the push for the tying goal, and then if that happens overtime is just a march to the gallows.
- As a casual fan watching, I don't feel the same way if it's been 1-0 all game and then Team A gives themself an insurance goal halfway through the third period. So that's where the "sustained through the middle of the game" part comes in.

 

[Spirit of 67]

Commentators often say that a 2 goal lead is the worst lead in hockey. The most common explanation is players will become complacent (despite not really having a large enough lead to justify that complacency). The other explanation I've heard is it creates confusion, because a team isn't sure if they should attack further, or defend the lead. Regardless, the theory is a team will give up a 2 goal lead more often than they "should" as a result.

Has anyone studied this phenomenon? Is there any evidence that 2 goal leads are less secure than one would expect?

I think it's the early 3 goal lead, tbh.

 

[Johnny Engine]

I think it's the early 3 goal lead, tbh.

I know I jokingly trashed that in my post just above, but I've specifically noticed a shift in the way people use the phrase over the last 20 years, from 2 goals to 3 goals. I'd be very surprised if 3 goals being a particularly risky lead holds any water at all, but who knows.

 

[Doctor No]

This is just data with little thought behind it other than how to pull and organize it.

Average points achieved, NHL regular season games (2018-19 through last night), based on score differential at each intermission:

SCORE AT 1st INTERMISSION	AVG PTS
-6	-
-5	-
-4	-
-3	0.13
-2	0.47
-1	0.75
0	1.12
1	1.49
2	1.72
3	1.92
4	2.00
5	2.00
6	2.00

[TBODY] [/TBODY]

SCORE AT 2nd INTERMISSION	AVG PTS
-9	-
-8
-7	-
-6	-
-5	-
-4	0.01
-3	0.09
-2	0.22
-1	0.64
0	1.20
1	1.64
2	1.88
3	1.95
4	1.98
5	2.00
6	2.00
7	2.00
8
9	2.00

[TBODY] [/TBODY]

Coarsely speaking, it's aesthetically pleasing that a first-intermission two-goal lead is about halfway between a one-goal and a three-goal lead.

After two periods, a two-goal lead becomes nearly as good as a "better" lead.

Caveat: data errors may be present in my files.

 

[Doctor No]

Anyone care to guess which game (from 2018-19 until last night) had a four-goal second-intermission deficit turn into points?

 

[RCP]

I think it is worth bearing in mind that although every two-goal lead blown became a one-goal lead before being overcome, not every one-goal lead blown was once a two-goal lead. Because of that, a one-goal lead is the worst lead in hockey,

 

[Filthy Dangles]

Anyone care to guess which game (from 2018-19 until last night) had a four-goal second-intermission deficit turn into points?

I remember that game.

 

[Filthy Dangles]

It's one of the dumber cliches in hockey. Would you rather be up by 1 or 2...it's hard to believe it's a real thing people say.

 

[Mickey Marner]

Some numbers from this season. I thought the all-situations GF% was ridiculous relative to the shot attempts, so I checked out 5v5 as well I'm pretty sure those are heavily empty-net inflated goals. If anyone cares they can check for ENG specifically. All data is downloadable from naturalstattrick, all you have to do is modify the URL from "u1" to "u2" when sorting by team stats. You can do this for game states as well, turn 5v5 into 3v3 or whatnot, though it doesn't seem to want to co-operate if you try this with multiple seasons, so you have to go year by year. Any difference between a one and two goal lead looks pretty negligible to me, but obviously you'd want to increase the sample size beyond 2021 before drawing any conclusions.

STATE	TOI	CF	CA	CF%	SF	SA	SF%	GF	GA	GF%
UP ONE All-situations	15872.02	13409	15141	0.470	7632	8218	0.482	810	715	0.531
UP TWO All-situations	7837.067	6265	7848	0.444	3548	4182	0.459	446	387	0.535
UP ONE 5V5	12608.05	10790	11726	0.479	6064	6351	0.488	526	511	0.507
UP TWO 5V5	6049	4837	5838	0.453	2717	3089	0.468	258	241	0.517

[TBODY] [/TBODY]

 

[plusandminus]

Commentators often say that a 2 goal lead is the worst lead in hockey. The most common explanation is players will become complacent (despite not really having a large enough lead to justify that complacency). The other explanation I've heard is it creates confusion, because a team isn't sure if they should attack further, or defend the lead. Regardless, the theory is a team will give up a 2 goal lead more often than they "should" as a result.

Has anyone studied this phenomenon? Is there any evidence that 2 goal leads are less secure than one would expect?

I think I studied it many years ago but don't remember the results.

To study this, I think one needs to factor in some things.
When? Here a good approach might be the end-of-period data published here.
Strength of teams? We might want to consider this. For example, if a top team leads by 2-0 against a bottom team, the likelihood of them winning is likely bigger than if a bottom team gets an early lead on a top team. Considering this is meaningful as most 2-0 leads likely will be by good teams that likely would have won anyway. The 50-50 or so games might be more interesting to us.
Control group? We might want to check out other leads as well, to have something to compare our results with. Like @Doctor No did.

 

[plusandminus]

This is just data with little thought behind it other than how to pull and organize it.

Average points achieved, NHL regular season games (2018-19 through last night), based on score differential at each intermission:

SCORE AT 1st INTERMISSION	AVG PTS
-6	-
-5	-
-4	-
-3	0.13
-2	0.47
-1	0.75
0	1.12
1	1.49
2	1.72
3	1.92
4	2.00
5	2.00
6	2.00

[TBODY] [/TBODY]

SCORE AT 2nd INTERMISSION	AVG PTS
-9	-
-8
-7	-
-6	-
-5	-
-4	0.01
-3	0.09
-2	0.22
-1	0.64
0	1.20
1	1.64
2	1.88
3	1.95
4	1.98
5	2.00
6	2.00
7	2.00
8
9	2.00

[TBODY] [/TBODY]

Coarsely speaking, it's aesthetically pleasing that a first-intermission two-goal lead is about halfway between a one-goal and a three-goal lead.

After two periods, a two-goal lead becomes nearly as good as a "better" lead.

Caveat: data errors may be present in my files.

Your tables shows that the better result after 1st or 2nd period, the better likelihood of winning.
It would be interesting if you could filter it in a way where you eliminate the best and worst teams, so that we focus on teams that are fairly evenly skilled.

I notice that for games being tied at period end, it seems as if both team will get on average 1.12 or 1.20 points. I suppose that would be true for both teams, as the NHL likes to award a bonus point for when teams are tied after regulation. It would be interesting to focus on the score after 60 minutes of play, and give 2 for a win/lead, 0 for a loss/trail, and 1 if game was tied.
Edit: But on the other hand, the trend seem clear enough as it is.

 

[Doctor No]

That's definitely possible - I'll add it to the list. "Fairly evenly skilled" is qualitative, of course, but I'll propose a definition that two teams' SRS values (*) are within 0.25 of one another (inclusive of home ice adjustment).

(*) Simple Rating System - scheduled-adjusted goal differential; I calculate my own although others (like H-R) have theirs.