With ESPN putting up an article from Timo Seppa (unless you have Insider, you can't see it yet) teasing the upcoming VUKOTA projections yesterday, I figured I should probably publish the recap from last years. When the article goes up on Puck Prospectus, you can find the Insider article it in the Fly By.
If you recall, last year we looked at the VUKOTA projections for the 09-10 Flyers. There, we looked specifically at the GVT projections, which I thought were a bit low. Well, it turns out VUKOTA did a pretty good job.
To make things simple - and fair to the system - I only looked at players who actually played in 40 games this past year and had a projection from VUKOTA. This created a 16 player sample - 11 forwards and 5 defensemen - to look at. The total team results were:
Looking just at this, the system as a whole did a pretty good job at predicting the 09-10 Flyers. But what about the projection in it's entirety?
Jump to find out.
Now, if you saw the actual Puck Prospectus article in the link above, the Flyers were projected to do this:
When it first came out, everyone thought the projections were too pessimistic. Turns out they weren't that far off. The problem is, when totaling all the players in the VUKOTA projections, the expected goals scored was 237.2. That's not what the Puck Prospectus article said was expected. So let's try to explain it.
Here's looking at the full chart of individual Flyers' performances and VUKOTA projections:
|Sum of Individual||GP||G||A||Pts||GVT|
While we see similar scores across the board, this is where we see the flaw: The VUKOTA projections didn't include enough games played. Every year, the Flyers have 82 games with 18 skaters equaling 1476 games played. VUKOTA was over 2 full seasons off (186 games divided by 82 games = 2.27 seasons) in their projections. While this isn't perfect, let's assume VUKOTA had the averages right and just extend their projections over a full NHL season:
And there we see how VUKOTA's team projections were off. Well, that or just how badly the team underperformed this year. I mean, we can't blame VUKOTA for not predicting what none of us could predict: that the team just wouldn't score.
So how did VUKOTA do predicting each individual player? For the entire spreadsheet, click here. For the file of those players who didn't reach the 40 game threshold, or who didn't have projections, click here.
Before looking at the chart - since it's long - here's the key: PGP = Predicted games played; AGP = Actual games played; PPts/G = Predicted points per game; APts/G = Actual points per game.
Italics indicate players who beat their projections. Bold indicates those who missed by 0.15 points per game or more.
By this metric, you can see that many regulars produced far less than expected. Mike Richards, Simon Gagne, Claude Giroux, Scott Hartnell, and Braydon Coburn all missed their marks by 0.15 or more. Jeff Carter missed by 0.14 per game. Danny Briere missed by 0.12 per game. The only person predicted to score half a point per game or more who didn't miss by 0.12 points per game or more was Chris Pronger. And yet, this group as a whole only produced 9 points fewer than projected. How is this possible, you ask? They played an additional 171 games. Had the group produced as projected - in the same amount of games - they would have scored an additional 102 points.
Again, whether you want to say VUKOTA is inaccurate or the Flyers had a miserable season is up to you. Really, it's probably a combination of the two. The only player to far exceed expectations was Blair Betts, and then you have to wonder why he was projected to score fewer points per game than Ryan Parent to begin with.
While we're not going to go over the second spreadsheet - those players under 40 games and rookies - know that James van Riemsdyk made up for a lot of the poor results. In the file, you may notice outlines around the cells. That was done to group players by games played to see if there were any correlations. There really weren't. To see VUKOTA fail, this is the chart to look at. But before criticizing it, know that Timo didn't want the Lukas Kaspar, Krystofer Kolanos, Jason Ward, OK Tollefsen, and Danny Syvret projections published because of small sample sizes. In other words, VUKOTA doesn't claim to project this chart. It's just too difficult to predict what the Jon Kalinskis and David Lalibertes of the world will do.
At first glance, VUKOTA seemed to predict the 09-10 Flyers performance fairly well. But when you look deeper and see that the team achieved the same amount of success in a far greater amount of games, it becomes apparent that VUKOTA missed the mark. Even then, look at this:
There are quite a few huge outliers there. Richards, Gagne, Giroux, Briere, Hartnell, and Coburn, all had their worst years in the past two years. Were the VUKOTA projections that far off their previous two years' production? Not at all. The projections seem accurate. The results seem to be outliers.
Just something to keep in mind when checking out the 10-11 VUKOTA projections, which will be coming shortly.