I am admittedly a bit of a fantasy hockey junkie. I am in several keeper leagues online with several stat categories, and also take part in a few yearly drafts that are points only. Since I am a math teacher by day, I try to use this to my advantage, especially in the yearly leagues where we just draft a dozen players and the only stat we use is points. It is a pretty straight forward league, whoever drafts the team that scores the most points during the season wins. No trades, no free agents, no drops, no keepers, and certainly no tracking shot blocks and hits.
It is old school, so like I said I try to use the fact that I am a math teacher to my advantage. First, I do lots of number crunching because I enjoy it and it usually leads to me being near the top by the end of the season, but I also have one other advantage on most fantasy poolies, and that is plenty of time on my hands in the month of August.
So I scour the internet looking for expert advice and try to find trends and any information that can help me find a steal in the 8th round of a 15 person draft. I was a bit underwhelmed at the amount of information out there, so I decided to create some of my own.
I spent some time crunching numbers and looking for patterns and anything I can use to better estimate who is going to score the most points in the NHL next season. I kept coming back to the fact that age is an enormous factor in any players point totals, as well as their past production of course. I checked the past several seasons and grouped all players into their age group, and tracked how they did at 22 compared to 23 and 24 and so on.
Though I expected this to lead to a huge breakthrough, looking at the NHL as a whole never seemed to go anywhere. During the years when individual players seemed to be doing their best, which was the mid-20′s, there was also a huge group of third and fourth liners who were dragging the age group down. Then older ages, like say 36, would have five or six decent scorers and that would be it. So when I realized most fourth line plugs are in their mid-20′s and most people do their best scoring at the same time, I realized I had to look at it a different way.
So, instead of grouping the entire league into their respective age group and trying to compare, I just took one player at a time and looked at when he had his most productive years as an individual. I then compared every other season he had to his best, and found the percentage of points he scored in each season compared to his peak year.
I thought that if I gathered the data for several players and compared when their peak season was, and when they came closest to reaching their peak season, I could find when NHL players are expected to score the most points in their career. I decided to picked several players to start this little project and make a spreadsheet with each player getting a column of their PPG/Peak at each age. Every player would have a 1 in the season in which they had the best PPG, and then every other season they played in the NHL would be compared to that season, and given a “mark” based on the PPG of each season compared to their best.
For example, Marian Hossa had his best season in 2006-07 when he scored 100 points at the age of 28, which is a points per game of 1.22. For this season I would give Hossa a 1, and then for every other season, I compared that year’s points per game to the 1.22 PPG he scored in his peak season. In 2010-11, Hossa scored 57 points in 65 games, which is 0.88 PPG. To find out his “mark” for the season, I take his 0.88 PPG and divide it by his peak season of 1.22 PPG and get 0.72. This becomes his “mark” for his 32 year old season, and says at the age of 32, Hossa scored at 72 percent of his peak season.
I wish I could plug in the data for every NHL player, or even a few hundred, but it is extremely time consuming to calculate the PPG average of each player for every season they played in the league. So I decided I needed a dozen or so players that were somewhat random so i could get started.
I picked players that have played in the NHL at the age of 40 since the 2005 lockout. This would keep the data fairly recent, while also getting players who had lengthy careers. This gave me Gary Roberts, Mark Recchi, Mario Lemieux, Joe Nieuwendyk, Radek Dvorak, Ray Whitney, Daniel Alfredsson, Jaromir Jagr, Scott Mellanby, Steve Yzerman, Doug Weight, Brendan Shanahan and Mike Modano. I found that only a few of them played at the age of 18 so I also grabbed Kirk Muller and Trevor Linden since they were comparable to the one’s I used and played at the age of 18.
I took each player and sorted his stats by PPG. Whatever age was first got a 1, and then I went down the list until each season was listed as a percentage of the player’s best season. Luckily hockey reference.com does a lot of this work for me. I then plugged in each value, which would be between 0 and 1 into the spreadsheet, and then went on to the next player.
Once I had all the data I took the row of each age and found the average. This would calculate the average PPG percentage of peak for each age. Basically it would combine all of the players and say at the end, a 19 year old should score at 57.5 percent of what his best season will be. The highest number an age could possibly get would be 1, and that would only be if every player in the data had their best season at the same age. This of course did not happen, but when all the data is formulated, you will get a graph that shows you when a player can be expected to have his best seasons. That graph looks a lot like this:
The x-axis represents a given age, and the y-axis represents the percent of peak performance. Basically the highest bars are when players normally have their best season, and the lowest bars are when a player is far from his peak.
The graph shows about what one would expect, but actually calculates it and takes out some of the guesswork. Based on the players who were in this “study” you expect a player to get better for a few years and then get worse in his 30′s. Admittedly, nothing groundbreaking is going on there, but a few interesting tidbits certainly came out of it.
First, these players had a very clear and defined “prime” which ranges from 22 to 27. This is slightly younger than I would have thought, but again, not entirely out of the perceived norm. What is a bit odd is how quick players appear to decline after they turn 28. Each season after 27, you can see the graph declines sharply, until it bumps up at the age of 35.
That bump at the age of 35 seems odd, and I would consider it an outlier since the sample size I used was only 16 players. Mark Recchi, Gary Roberts and Mike Modano had successful years at 35 that they were not able to match the year before or after which has a big effect on this graph when you also consider Teemu Selanne and Recchi lost their 34 year old season to the 2005-06 lockout which would have bumped that number up.
There are a few things to consider when looking at this data. First, I only used 16 players and though a trend can be seen throughout most of these players, it is a small sample size that I will continue to work on. Also, I only used forwards, and though I tried not to use all Hall-of-Famers, there is not a lot of variety in the players I used. Using this data to predict how P.K. Subban will do next season may work, but with no defencemen in the data, it could be off.
Also, I realize that all of the players I used are at least “good” NHL players, and that is by design. The point was to help me (and now you) in our upcoming hockey pools, so I didn’t bother using Sandy McCarthy’s point totals in the data. Also, if you are considering drafting Colton Orr I can’t help you anyway, except to tell you that I would love to have you join my draft at the end of September.
This is also sort of a “beta” graph since so few players were used to develop it, but I honestly think it would remain mostly unchanged if I were to continue adding players into the data, and I will add more players to see what changes, if any, occur.
I get that this is a Habs fan site and you are probably only reading this if you are a Habs fan, so I will throw a quick Canadiens spin on it now.
Looking at the Habs forwards for next season, and using the above graph, I think Brendan Gallagher and Alex Galchenyuk are about to have big years. Gallagher at 22 could be much closer to his prime than one would think which seems strange since he only has two seasons under his belt.
David Desharnais is going into his 28 year old season, and had his big breakout in 2011-12 when he was 25 years old. I wouldn’t be surprised to see his production dip slightly this season compared to his 0.66 PPG last year. His line mate Max Pacioretty however is heading into his 25 year old season and will be in the middle of his prime. His PPG over the past three season were .82, .89 and .82. The optimist in me suggests he will reach .90 and get 75 points, but expect 70 from the Habs top scorer.
Tomas Plekanec was 32 last year and saw his PPG drop to 0.53 from 0.70 and 0.64 the two years previous. A small bounce back may occur, but don’t expect Plekanec to get any more than 50 points this season. Newly signed P.A. Parenteau is an odd case since his play dropped off sharply in his 30 year old season. Injuries can be blamed on that, so don’t expect him to get back to his 0.90 PPG from two years ago, but he can bounce back from last year’s 0.60. He should be able to challenge 60 points based on his production before last year’s injury plagued season.
If you have any questions or comments, fire away in the comments section, or send a tweet to @FSAWinningHabit.