Point Shares
- Thread starter BubbaBoot
- Start date
overpass
Registered User
- Jun 7, 2007
- 5,274
- 2,819
The system uses available data to allocate credit for team regular season success to individual players.
1. Team regular season points are split into offensive and defensive credit, depending on the team strengths.
2. Team offensive credit is divided among the players based on goals and assists. Goals are given more weight than assists.
3. Team defensive credit is split between the players and the goalies, according to shots against and save percentage data. Defensive credit for players is divided based on minutes played and plus-minus.
Problems can occur in the following cases.
1. Some data is not available for earlier eras. For example, the NHL does not have official shots against data pre-1983, which makes it difficult to credit goalies accurately for that time period. Time on ice is not available pre-1997, so it must be estimated before that time. Plus-minus is not available pre-1967, so it can't be used in the calculation.
2. Changing league sizes and schedule lengths mean that point shares can't really be compared across eras. For example, 50 years ago there were 6 teams that played 70 games apiece. So there were 420 point shares handed out in a season. Today, there are 30 teams that play 82 games, and three points are given out in some games. So there are over 2500 point shares handed out in a season.
3. Issues with the way point shares are calculated. Plus-minus is used to allocate defensive point shares, but it's just as much an offensive stat as a defensive stat. There is no extra credit given for killing penalties, an important defensive role. This is how Phil Housley ends up as one of the all-time leaders in defensive point shares despite rarely killing penalties and being a defensive liability for much of his career.
TheDevilMadeMe
Registered User
The short version: Point shares is a crap formula
The long version: No all-in-one stat is perfect, but if you want to use one, Tom Awad's Goals Vs Threshold, and Iain Fyffe's Points Allocation are much better, as they were created by people who actually understand hockey. H-R's points shares formula was basically an attempt to adopt a stat from other sports to hockey by people who don't know the difference between a power play and a penalty kill. The calculation of "defensive point shares" is particularly terrible.
See details here: http://hfboards.mandatory.com/showthread.php?t=889335
The long version: No all-in-one stat is perfect, but if you want to use one, Tom Awad's Goals Vs Threshold, and Iain Fyffe's Points Allocation are much better, as they were created by people who actually understand hockey. H-R's points shares formula was basically an attempt to adopt a stat from other sports to hockey by people who don't know the difference between a power play and a penalty kill. The calculation of "defensive point shares" is particularly terrible.
See details here: http://hfboards.mandatory.com/showthread.php?t=889335
TheDevilMadeMe
Registered User
It's worse than this, right? They actually give defensive credit for goals scored against per minute, but make no distinction between PP and PK time. So an offensive specialist who plays massive PP minutes (where you really get scored against) but rarely plays PK (where even the best will be scored against) will rack up the "defensive point shares."
Absolutely worthless "advanced stat."
BubbaBoot
Registered User
I figured there was a fair amount of subjectiveness, ( as in "Team regular season points are split into offensive and defensive credit, depending on the team strengths.") on a dubious stat....thanks.
overpass
Registered User
- Jun 7, 2007
- 5,274
- 2,819
Here's the details on how the stat is calculated.
http://www.hockey-reference.com/about/point_shares.html
I don't think it's quite as bad as you put it. Defensive credit is basically minutes played*team defensive success + a plus-minus adjustment. So penalty killing is not given any value in point shares, but it doesn't hurt the player either. It just isn't considered. The plus-minus adjustment is very simple, it literally just uses player plus-minus and team plus-minus. Nothing so advanced as goals against.
you are right about the other 2 formulas being better but the biggest problem is the lack of full data until recently.
The HR formula and system is okay at getting a good ballpark picture, some of the finer details can get lost though.
TheDevilMadeMe
Registered User
Thanks for the clarification. I honestly stopped paying attention to the details of the formula when I realized how awful it was, but it's good to have the correct details down for accuracy sake.
That said...
It's not even good at that. It has Phil Housley with more "defensive point shares" than Rod Langway!
It's kinda like the exception proves the rule, the overall defensive share points isn't really all that bad.
The biggest problem is the lack of detailed statistics to really have a consistent defensive stat throughout time in the 1st place IMO.
TheDevilMadeMe
Registered User
It's not an exception. The overall defensive point share system is horrendous.
Here are the "career defensive point shares" of some cherrypicked defensemen. In other words, hockey reference is claiming this is a list of the guys who prevented the most goals over their careers; offense is supposedly not a factor.
7. Larry Murphy 91.47
17. Glen Wesley 76.79
20. Sergei Zubov 74.08
22. Phil Housley 73.91
25. Rod Langway 71.05
26. Red Kelly 70.19
27. Mathieu Schneider 69.50
31. Paul Coffey 68.74
32. Roman Hamrlik 68.56
39. Mark Howe 64.01
40. Borje Salming 63.92
69. Bryan McCabe 56.96
82. Jacques Laperriere 54.04
95. Adrian Aucoin 52.62
98. Eddie Shore 51.37
http://www.hockey-reference.com/leaders/dps_career.html
Then "overall point shares" is created by combining this garbage with a halfway decent measure of their offense (that's still inferior to straight up points, IMO, but at least it's better than this crap).
It's an absolutely worthless formula created by statisticians who didn't understand the sport they were attempting to measure, and hockey reference should be embarrassed for publishing it.
Last edited:
Bear of Bad News
Your Third or Fourth Favorite HFBoards Admin
- Sep 27, 2005
- 13,602
- 27,403
For what it's worth, the phrase "the exception proves the rule" is a nonsense statement that's entered the English language. A common origin suggests that it stemmed from "the exception probes the rule", meaning that the exception tests the rule. Which makes a lot more sense, doesn't it?
An alternate etymology for the phrase is that "the exception proves (that) the rule exists" - for instance, if you saw a sign that said "PARKING ALLOWED ON SUNDAYS", it implies a rule about not parking there on other days of the week.
How in the world could an exception prove a rule?
Theokritos
Global Moderator
- Apr 6, 2010
- 12,552
- 4,974
Nice theory, but the phrase also appears in other languages where the probe->prove etymology doesn't work. In German for example it says Ausnahmen bestätigen die Regel ("exceptions confirm the rule").
The source of the phrase is actually Latin: exceptio probat regulam in casibus non exceptis ("the exception confirms the rule in the cases not excepted"). Which comes very close in meaning to what you suggest in the other etymology. The second part of the proverb (...in the cases not excepted) is usually left out, for the sake of brevity I guess, hence the confusion and the misuse of the phrase.
Bear of Bad News
Your Third or Fourth Favorite HFBoards Admin
- Sep 27, 2005
- 13,602
- 27,403
Theokritos
Global Moderator
- Apr 6, 2010
- 12,552
- 4,974
I don't have "problems" with either. The first one is a nice idea that just happens to be wrong. The second one is not 100% identical with the original Latin proverb, but pretty much amounts to the same thing. Which I have already stated:
You were wondering about the origin of a paradox phrase and made some good attempts to solve the mystery. I happened to know the solution you were seeking and posted it. No "problems".
