Discussion in 'The History of Hockey' started by SidGenoMario, May 23, 2011.
It can. All you have to do is regress all the stats of players vs if they make the HoF or not. Of course you'd have to normalize everything in someway and you might need to look at obscure things like # of season in top X of scoring which would be absolutely insane to compile (unless you are pnep). In fact, if someone has some some grand database, I'd love to analyze this.
The author does not mention how significant his results are or provide out of sample validation but it's a good effort. He'd be better served with a binary logistic regression IMO.
I guess one could make up a model predictor and modify it to the number of teams in the league and use adjusted statistics to some degree but hockey is much more than stats than some sports like basketball and especially baseball.
I don't think a lot of fans realize how little use the executives, the coaches and the players care about stats. It's probably the sport that has the least emphasis on stats by those in the game. Remember when Shanahan scored his 600th goal? The interviewer asked him about what it meant. He said it would be the victories and the players he played with that he would reflect on once his career is finished, and not the statistical milestones, like the 600th goal. I think you'd find that most players have the same attitude. Players who do care about their statistics are often regarded as selfish, troublesome, cancers, or as "addition by subtraction" types.
Part of the problem is that there are so many elements of the game that aren't measured on stats. There isn't a stat to measure a player's ability to win the battles in the corners, win the battles in front of the net, make the big defensive play (there's more to great defensive play than just takeaways), create intimidation, create a presence, or make the timely save at the right time. (Which happens more than just late in the third period of a close game). Bob McKenzie said it best in his book, "Hockey Dad," when he said that competitiveness is a skill. There's no stat for competitiveness, but it's a skill that's one of the most sought-after once you get to the game's upper levels.
Adjusted stats? Nobody in the game cares about that. They want to be remembered for who they were and what they did and what they accomplished, not for what a calculator would say they might have done. No NHL player would ever want to have a calculator involved in judging his career. Nobody cares really about their shooting percentage. Goalies might be able to tell you their career numbers for wins and shutouts, but none of them could tell you their career save percentage or their career goals against average. How do you quantify a defensive forward? You can't. How do you quantify a defensive defenceman? You won't.
I think stats are for fans and the media. I think most of us would be really surprised how unaware players are about their numbers. In many cases, I think the only way they would know is if a reporter told them, or if their agent tells them during contract negotiations. I noted in an argument for Martin St. Louis in the HHOF that he's been year-in, and year-out, the best playmaking winger in the league for nearly a decade. He's so slick and creative, so smart, so dynamic with the puck, and willing to take a hit to make a play, that he makes the players around him better. (When he chooses to play the goal-scorer role, he's damn good at that, too). But for the benefit of fans who like their numbers, I noted that at the time he'd been No. 1 in assists among wingers three out of six years. (It's now four out of seven, including the last two and three of the last four. He's also been No. 2 in assists among wingers once, finishing behind that Jagr guy in 2006-07). The thing is, we can go to www.hockey-reference.com and look at top 10s and find that out. Marty St. Louis probably couldn't tell you how many times he's been No. 1 for assists among wingers. He probably doesn't care. He couldn't tell you how many times he's finished in the top 10 in goals, assists or points. He might be able to tell you his point total from 2003-04, when he won the Hart and the Art Ross, but he's more likely to remember his game-winning goal in Game 6 of the Stanley Cup final.
The selection process for the HHOF certainly has its flaws. I've argued that we need a larger selection committee. Expand it from 18 (its current size) to 40. That'll prevent the occasional oddball selection (Clarke Gillies, for example). At the same time, the composition of the committee - with former players, executives, coaches and officials, along with media members - is far better in baseball.
Hockey's a completely different beast from any other sport in terms of using stats in evaluations. Maybe it's a Canadian thing. Maybe it's because many players have been told for a long time that statistics are for losers. Or maybe it's just because so many elements of the game aren't measured in statistics, and never will be. But to have a "quantifiable HOF system" is insulting to the game and the HOF, and most players would tell you it's nothing short of bull****.
Excellent post God Bless Canada.
One addition. Players like Bill Gadsby, Jean Ratelle, Brad Park, Rod Gilbert have often said that they would gladly have sacrificed individual stats and honours for a Stanley Cup win.
pnep has created such a system, linked in his post # 4.
Iain Fyffe of Hockey Prospectus has also created a system, see here.
Of course these models simply look to see how a player's statistics and accomplishments stack up to past HOF inductees, instead of directly evaluating the player and his career. As such, I don't think many people would advocate replacing HOF voting with such a system, it's a fun exercise for fans more than anything else.
Quality over quantity?
Because if there was a quantifiable system, you can bet Stanley Cups would be on there and there are some quality players who have not won the Cup. Also, there are probably players who aren't Hall of Fame worthy, but played a really long time and won a few Cups.
what's the fun in that?
Firstly, just because things aren't measured doesn't mean they can't be measured. You could measure competitiveness or defensive ability, you just need to figure out how. It could be a mental analysis, some sort of questionnaire or personality profile that you administer to known "competitive" players and find underlying commonalities. You could measure how many shots a player disallowed and against who he disallowed them. You could measure puck battles engaged and how man they won and against who, etc.
Secondly, the attitude against stats is antiquated. Stat analysis is coming in sports, like it or not. Hockey might be the hardest to analyze but it can be done and eventually it'll come. You might think the front office doesn't use them but they do. If you saw 24/7 you saw Bylsma and Shero rating players on a 5 point scale...that's doing statistical analysis on a crude level. Stan Bowman admits they do something similar but more intensive. Buffalo has an analytics department to do stat work.
Finally, it's not so much a calculator will decide who is an HOF player but you can find out what the voters place emphasis on by doing a simple regression. The voters might not even know what they are looking at but if, for example, 95% of HOF forwards have a career 12.5 shooting percentage, or higher, then clearly something is there. But it can be robust too. Maybe Cups matter but on partially so a player could make it without a cup. Of course things have to be standardized because of more teams, lower scoring, things like that.
NHL decision makers definitely track statistics to help them make better decisions.
This doesn't mean that published hockey statistics from the history of hockey contain all relevant information about a player and his career.
So you just want everyone to adopt pnep's monitor points system?
I like the concept of pnep's system, but its too biased towards forwards. Goalies don't get credit for leading the league in save percentage, but a forward gets creit for goals and points. Also, how can you quantify the impact of a 'defensive-defenceman'? I would rather have derian hatcher, rod langway or adam foote on my team over phil housley and mike green any day of the week.
Kevin Lowe's name is popping up in my head right now. Then again, even under the current criteria he has his supporters for the HHOF
Does anyone else feel like the obsessive compulsive tendency for some to have to quantify everything and adjust statistics in hockey seem a little too baseball-like? I mean this in a pejorative sense: whereas baseball can be broken down into individual actions and quantified, hockey is way more fluid and statistics aren't always a good or reliable measure of an individual's achievements and contributions. I never understand why there is the tendency for some fans to hyper rationalize everything or else try to raise the exclusivity of the Hall, to make it more elite.
Very true with a few clarifications and explanations. Each team at all levels on down from management thru coaching keeps internal statistics on the players. Players are evaluated and graded on a game to game basis. These stats or evaluations are never made public. What a team releases from a statistical standpoint is heavily filtered.The most important element in the internal team stats, evaluations and grades is that they are the only compilation of data that takes into account the coaches' game plan and the execution of objectives. Also they would be the only source where the extent of injuries and other variables may be factored in.Good luck accessing this data.
Teams also gather extensive data on all opposing NHL players, potential NHL players, pro and juniors. The advance scouting and game preparation is one aspect, the acquisitions part is another. How a player plays against a team may be very different to how a player plays with a team or upon leaving a team. Again this data is rarely accessible but you may get hints as to which teams do a better job when game plans are analyzed after execution or trades produce unexpected results.
In terms of the HHOF criteria being quantifiable via criteria in a data bank the answer is rather simple. You would not be able to get appropriate data going backwards in time. At best you would generate educated estimates.
Going forwards and building a data bank is possible but has many obstacles. SV% being a prime example. Shots on goal do not take into account degree of difficulty.Even if the stat was upgraded to include degree of difficulty and the provenance of each shot you would still have the obstacle of the defending team objectives. Teams defend to a goalies strengths as much as possible while the opposition attacks to a goalies weakness as much as possible. Filtering thru the conflicting game plans with all the misdirects that coaches produce would be a herculean task.
Adjusted stats aren't about what a player might have doen but rather a pretty good baseline to compare different seasons into a common denominator. It's not perfect but it's a good tool to try and compare offensive output from different seasons for any player IMO.
Overall a lot of this post is dead on but for many players the starting point is really about stats and the rest is icing on the cake that stats is built on.
You try hitting a ball thrown 90+ mph, it's the hardest thing to do in all of sports, there's more to hitting a ball than swinging a bat, and there's more to pitching than just throwing a ball. The amount of variables in a baseball game are infinite.
Hockey is just a bunch of 'tough' guys on skates, trying to put a small black object in a giant net. That's all it breaks down to, very simple concept.
But a players ability to effectively do that can be broken down in many different very accurate ways.
Where as a hockey players ability to help his team score and his opponent not score cannot be.
There is a reason why hockey seems to follow baseball with terms like the dead puck era instead of what it really was (the clutch and grab era). There simply has been a richer tradition of baseball writing in both terms of quality and quantity and hockey has always followed the lead of baseball in terms of player evaluation from a statistical point of view.
You quite rightly point out that there is alot more to hockey than just simply stats but on a basic level players can be determined by how much they score, or in goalies case how much they prevent the opposition from scoring because the team with the most goals in a game wins the game a point that is lost sometimes IMO.
At the end of the day pure stats should probably account for 70-80% of a HOF resume and intangibles maybe up to 20%.
Put another way stats are the cake and intangibles are the icing.
I prefer MLB
it is a ***** to get into that hall and people who get into it--there is very little argument about it
God Bless Canada gives you the "official" version of guys that are all about winning and have no selfish interest whatsoever. I think that attitude is actually more the exception than the rule. He's right though in that this is what people will tell you because it is downright reputation poison to be seen as a guy who pursues individual stats.
Yet I bet there's a lot of guys out there who care about their stats, especially forwards and goaltenders. Why? Better stats usually means more money, more recognition. It can be the difference between a new contract or being sent packing.
The situation is pretty easy for goalies, their stats tend to be the most directly linked to team success, have a terrible save % and GAA and your team probably isn't doing fantastically well either. And of course, it makes a difference to a team whether you have a .895 SV% and a 5.27 GAA or a .925 SV% and a 2.35 GAA in terms of signing you and for how much.
But even with forwards, the type of forwards primarily aid to provide offense especially, the amount of points is linked to pay scale, a 90 point player will as a rule command a higher salary than a 50 point player, it's one thing an arbitrator will look at when deciding on a salary for example. People who dismiss stats tend to do without realizing that for a significant group of players stats are the direct count of how they help the team. Goals and assists don't happen in a vacuum, they occur only if your team scores, the more you score the more you tend to win.
It is simple really: the statistics collected in hockey are a very vague description of the value of the players to the win.
So there is a lot more to determining player value than their individual statistics.
Save percentage isn't always linked to success of a team. Dominik Hasek in 1996 played for a crappy buffalo sabres team, yet he was still better than any other goalie at stopping the puck. Belfour and Broduer played behind defensive systems throughout the deadpuck era, some years they simply played better than others and that is why thier save percentage was better.
Yes hitting a baseball is the hardest thing to do in sports but what does that have to do with my post?
In baseball there is more isolation and control of both the hitter and pitcher as they battle one-0n-one and it is a sport that is more easily defined by stats than hockey IMO were there is a 6-on-6 battle and players playing both ways at the same time and a different gameflow as well.
Separate names with a comma.