Monday, July 6, 2009

Predicting Points

Whenever fans and teams look at signing potential unrestricted free agents to fill an important offensive role, the offensive track record of the player in question is obviously a major factor. The main statistic used in determining offensive success is points, i.e. a vital contribution to a goal scored either by shooting or passing the puck. I decided to take a brief look at the repeatability of points at the NHL level. To do this I looked at players who played in at least 50 NHL games in their 25, 26, 27, 28 and 29 year-old seasons and who were still playing in the NHL last season. I took an average of the first four seasons in terms of points per game (so seasons of 0.4, 0.45, 0.55 and 0.6 would average out to 0.5 regardless of the number of games played in each season so long as each season had at least 50 games played) and disqualified any player that averaged less than 0.5 points per game over the four year period. This left me with only 24 players (the lockout caused many players to lose one of these seasons). I then used the average of the first four seasons as a prediction for the performance of the player in his 29 year-old season. These are the results:

Admittedly, we're looking at a very small sample of players here. Still, the results look pretty close to what I would have expected: 75% of the time the player performs within 20% of his established ability over the previous four-year period. For the forwards the player performs within 20% of his established ability over 85% of the time. Obviously a larger sample of players will be needed in order to test these results. Fortunately, we are now entering the fifth year since the lockout. As such, I can predict the point totals for players entering their 29 year-old season this year using the previous four years as a base. If nothing else, this will be a fun experiment to revisit (triumphantly!) in nine months. Here are the predictions for the forwards who meet all of the criteria (have played at least 40 NHL games in each of the last four years, are entering their 29 year-old season and have achieved an average of 0.50 point per game over the previous four seasons). I included the cap hit as well as a high and low estimate, in terms of points per game, for next season.

A few notes for Oiler fans... Dany Heatley looks like a pretty good bet. As much as the situation with Heatley has been frustrating, there really aren't many players that have his offence and durability. I can see why the team is pursuing him to the point of embarassing themselves. If Heatley doesn't work, Jonathan Cheechoo might not be a bad idea. Granted, his last two years haven't been anything special, but that's part of why he's available. A bounce-back year from him could make for a big payoff. Finally, a guy like Robert Nilsson probably has some value to a team that thinks he can put up between 40 and 60 points. That's a talent that costs mucho money on the UFA market.

No comments: