As I alluded to in my introductory post, I love stats and I love that stats give us an opportunity to better understand performance. As this blog progresses I hope to ask a bunch of questions and attempt to use stats to see if an answer can be found. The fun will be in the trying, though I warn up front that I may not always have the grey matter to ensure victory.
Like so many sports fans, I at some point became aware of Bill James. I am a bit ashamed to admit it was relatively late in life when I read the book “Moneyball”*. The book rocked, not so much at the hard core statistical analysis that was being used by the Oakland A’s, but rather the fairly straight forward economic principles being used to find an advantage in the market. The market in this case being Major League Baseball.
* Ok it wasn’t that late in life. I’m 34 now so I must have been 28 or 29 when I read the book
It took some time to percolate in my mind, but my wife opened the floodgates when she complained about the Oilers pulling the goalie for the upteeth time. Her comments were something to the effect that MacT was stupid and pulling the goalie is a horrible decision. That got my brain going, asking all kinds of questions about in game management and strategy (questions that remain unanswered). Remembering Moneyball, I also started wondering about whether or not there was a better way to draft, to trade and to sign free agents. Essentially, I wondered if there were market inefficiencies in hockey that could be exploited if a team better understood the market compared to the other teams. MoneyPuck as many have come to call it.
I started with the pretty straight forward theory that team success is due to scoring more goals than you give up. As a team you need to put together a team that has a mix of guys that are good at scoring and/or good at preventing goals. To assess a player, ‘+/-‘ seemed like a pretty good guage of whether or not a player contributed towards wins. A half minute of thought later I had worries that an individual players ‘+/-‘ was dependent on a lot of factors: the team they played for, time spent on the power play, time spent on the penalty kill, the quality of players they play against, etc. Isn’t the shootout super important now though it isn’t part of +/-? My head hurt at this point so I started surfing the net to see what I could find.
I found there to be a lot of good information out there and a lot of good people trying to figure out the game of hockey. The one site in particular that blew my mind was hockeyanalytics.com . I’m not sure the work there is better than the other sites, but the analysis was put together in a way that was in line with my thinking, but 100 times better (ok 10,000 times better). It included a value for drawing penalties, something that I thought was underrated*. A team is roughly 16% to 24% likely to score while on a power play, a significantly higher probability rate than playing even handed so why not try to maximize your power play time and minimize your pk time? It also included situational analysis for even handed play, power play, and even shoot outs. Best of all, it quantified a players defensive contribution, a concept that really made my head hurt when I thought about it.
* I have a theory that a team should play the game in a manner that maximizes the probability of drawing penalties due to the increased probability of scoring a goal. This isn’t to say you need a team of Darcy Tuckers flopping around the ice, rather you identify the activities in the neutral and offensive zone that put the opposition in a position where they are more likely to take a penalty. Check back for that analysis in about 8 years.
About a year went by before I decided to get up off my butt and start this blog. While I’ve struggled to find my original list of questions, it was fun to think about hockey critically again, especially with the season about to start. I decided to check out hockeyanalytics.com again last week and lo and behold, the 2009 season analysis had arrived! Immediately my creative juices flowed as I now had the tools to analyze the game and the Oilers in a way that would make use of stats, just as I had hoped.
I’m going to spend a bit of time walking you through some of the key concepts developed by Alan Ryder over at hockeyanalytics.com, as I intend on using them in a number of future posts. Rather than try to reinvent the wheel and develop a methodology to evaluate players, I’m going to use the best wheel I’ve come across to see what I can learn from application of Ryder’s analysis. If you love hockey and you have an interest in stats and some acumen for mathematics, I highly, highly recommend you check out his site.
Note: the following section is essentially a rehash of certain portions of Ryder’s documentation on his Player Contribution System. It continues no original material on my part.
Ryder has used the term ‘Player Contribution’ (“PC”) to denote his methodology to attributing a team’s wins to the individual players on the team. It quantifies offense, defense and goalie contributions to a team’s performance. For players it quantifies their offensive and defensive contribution, breaking the game into pieces. Those pieces included even handed play, power play, short handed and shootouts. It also factors in penalty taking and the ability to draw penalties. I told you it was awesome!
For goalies, it quantifies their contribution during the game and in shootouts. During the game, the simplified explanation is that the more shots a goalie faces and the higher the quality the shots, the bigger the contribution all else equal. If Roloson and Brodeur had the same save percentage, Roli would get the nod as making a bigger contribution on the basis that the Oilers give up more shots than the Devils and tend to give up more point blank opportunities.
To make PC a bit easier to read, Ryder has multiplied the score by a factor of 10. If a team scored 85 points in the standings, there would be 850 points allocated across the team to denote each player’s contribution. A rough rule of thumb for non-goalies is that 100 points is a first team/ second team all star, 80 points is a team star, 60 is a team leader, 40 is a solid contributor and 20 is a weak link. Goalies tend to score a fair bit higher, which makes sense given they are on the ice the entire game and quite fundamental to team success.
For the Oilers last year, Souray topped the team at 71 points with Gilbert and Hemsky in the 50’s. It’s no wonder we stunk up the joint!
My last point on Ryder’s methodology is that he also includes cap dollars so we can clearly see the ‘bang for the buck’ obtained on each player in terms of how much it cost per PC point. I can already smell the Penner haters sniffing around for the answer to that question.
I’ve deliberately tried to stay out of the details of the PC methodology. If you are a stats-head, be sure to check out his site and read the full documents.
So now we step outside of the Matrix. I’m not going to commit right now to where I’m going to go with my PC analysis other than to say that the next hockey post will look at the Oilers roster last year in terms of PC to see if we can glean any insights into their poor performance. I’ll then look at the players we brought in to guess at how they might help or hurt our chances this year.
Until next time, keep fit and have fun!
PS- Penner had a PC of 29 last year, 12th on the team just behind Brodziak.