Please be advised that the following post contains subjects which are nerdy in nature. If you prefer not to hear anything about mathematics, statistics and other words with the suffix “tics”, then refrain from reading the following paragraph. If mathematics tickles your fancy, then you should probably ignore this post as well, as it contains mathematical deductions which are illogical and nonsensical. Thank you
Over the past two days, the Pucklehead has been working on his NHL statistics prediction spreadsheet. He thought he would be smart, and develop a foolproof formula for predicting NHL statistics for the coming year. His spreadsheet took all of the player’s statistics over the past three years (on a per game basis to account for injuries), and then fit a linear trend for each statistic and used it to predict the 2011-2012 season statistics. Well, that failed for obvious reasons. First of all, consider the case of Sidney Crosby. Crosby had an absolutely phenomenal, impossible to replicate season last year in which he scored 1.61 points per game. If you are trending statistics over 3 years, then having a season like that in the final year means that the trend will predict an even BETTER season in 2011-2012. Not happening. Under this formula, Crosby was on pace for some sort of borderline worshippable deity-like 200 point season.
It also failed to recognize cases like Jeff Carter; where their career statistics were all from a team which was considerably better than the pile of steamy, semi-solid poo of a club that they now play for. So there has to be team adjustments as well (cheekily named a “team multiplier” because the Pucklehead is highly cool). But how do you define the team multiplier? How can you predict how a player will adjust to a new team?
Because of the trending fail, the Pucklehead decided that using 3 year averages would be a more appropriate predictor of the upcoming season. That was all fine and good for some players with fairly consistent numbers over the past 3 years, but there is no way that a 20 year-old player is going to have the average of his age 18 and 19 seasons (see: Tavares, John). After testing this method with other players (Matt Duchene, Mikhail Grabovski and friends), the Pucklehead realized that this method was also completely unrealistic.
From this experience, the Pucklehead has learned a couple of things. Firstly, any basic model that the Pucklehead can construct with his pea-sized brain is going to be too simple to accurately predict anything. That’s why he doesn’t have a job in sports statistics. Secondly, this has proven what the Pucklehead has always thought. Statistics are meaningless (14% of Canadians know this to be true). Depending on what model was used, the Pucklehead could justify any prediction for the coming year. The results of these spreadsheets will be posted in the coming days, but if you read them, you should probably just ignore them. The ranking procedure is no more accurate than pulling names out of a hat. Just do yourself a favor and ignore what this shoddy, low-budget site is peddling for the next few days.