February 10, 2007
Stuart & W. Primeau for A. Ference & Kobasew
Stuart:
Calgary needed to get their power play going so they picked up a guy who is currently running at 1.8 GF/hr on the power play and if you include goals against essentially a big 0. His defensive numbers would make anyone cringe: -4.5GA/hr and +3GF/hr, for a total of -1.5 +/- /HR. In terms of the plus scoring rate he's one of the best on the team, but a -4.5GA/hr is something to worry about (ranked 229/231).
Primeau:
Popular probably because of his experience, because he doesn't help teams much. He's arguably the worst forward on Boston at even strength, doesn't help much on the PK and doesn't add much to the power play. Scores about 0.25 points per game, which is below what I consider marginal (0.25). However, I believe people consider him a leader. He seems able to stay in the NHL. According to Sportsnet he is a good penalty killer (hasn't panned out this year).
Ference:
He's D man #4 based on ice time, is the best defense man (statistically) on a reasonable poor PK. Signed for 4 years below market value.
Kobasew:
Thee guys are a dime a dozen, sure he's young, but I don't think he's going to be a top line center. He's probably more valuable than Primeau though.
I see this trade as Calgary throwing away some defense for some short term offense they believe they'll need in the playoffs. They may have thrown away too much defense (as a Canuck fan I can only hope). I generally think it's bad to take a #1 defenseman from a bad team as they are likely contributing to the problems. The numbers in this trade don't make sense on the numbers side. Scouting people might be able to make sense of it. It seems strange to me that a team would want the worst statistical players from a bad team for a couple average players on a good team, that appears to be what has happened here. Looks like a bad trade on the surface, but only time will tell.
February 7, 2007
Huselius Diving
There’s another triple threat player in the diving mix, Huselius now has three [1, 2, 3] diving calls against him, however all this appear mute as Hockey Operations disagrees with most of the on ice officials diving calls. When a dive (or unsportsmanlike conduct) occurs Hockey Operations takes over to determine the appropriate action:
In addition to the minor penalty for unsportsmanlike conduct/diving that may be assessed by the Referee during a game, Hockey Operations will review game videos and assess fines to players who dive or embellish a fall or a reaction, or who feigns injury in an attempt to draw penalties. (NHL.com)
The proverbial slap on the wrist will never rid the NHL of diving, but as I said before the NHL likes diving as it increases power plays and power plays increase goals and the current management believes that only goals make for good entertainment. It's no surprise they deal with diving the way they do: a lot of pretty ideas on paper with no action.
P.S.I posted a diving update last month that showed diving calls had fallen from 0.12/game to 0.06/game, however the next 100 games it has gone back to the 0.12/game level as there were 12 diving calls for games 700-799. Either the drop over games 500-699 was due to random variations or the NHL reads this site...
Chaos Theory
Why then do goals appear so random?
In simple terms, Chaos Theory can be summarized by the butterfly effect: "The phrase refers to the idea that a butterfly's wings might create tiny changes in the atmosphere that ultimately cause a tornado to appear (or, for that matter, prevent a tornado from appearing). The flapping wing represents a small change in the initial condition of the system, which causes a chain of events leading to large-scale phenomena. Had the butterfly not flapped its wings, the trajectory of the system might have been vastly different." (Wikipedia)
. The main point of chaos is there is a strong dependence on initial conditions as you can see in the graph to the right a difference 0.00127 resulted in the curves to eventually be on opposite sides of the graph. A chaotic system's dynamics are fully understood, so given any initial condition the result at any moment in time is known, however a small change has unpredictable effects. There are a number of real systems that are chaotic including the most classic of all examples: The Lorenzian Waterwheel. A number of other examples can be found here. A coin toss could be argued is a chaotic as could a roll of a dice, although most would agree that they are in fact random. I don't have an article: but a few years back I heard of a couple guys who could predict the numbers for the lottery as the lottery corporation used a random number generator and the individuals discovered all the initial conditions and knew the generating function and were therefor able to predict the winning numbers."So, is flipping a coin a random event? It really depends on the definition of the word 'random'. If we take it to mean that the outcome is theoretically unpredictable then no, the coin flip is not random. But in the real world it is practically impossible to predict the outcome of a good flip, so in that respect we can call it a random event." (Random Number Generator).
Alan Ryder concludes after going through great analysis that: "Goal scoring in a hockey game is basically a Poisson process." (Poisson Toolbox [PDF]). It is the result of hockey being chaotic with unknown variables that makes a completely determined system appear random. The more information we have on the dynamics of hockey and the more variables we know the less random the game will become. That being said, the chaotic properties would still exist and unless we had exact (infinite decimal places) values for all (infinite amount) of the information the game would still have significant random properties. And this in fact makes the game exciting. If we could predict with 100% certainty every event in hockey it wouldn't be too excite, how many people actually watch sports re-runs?
February 6, 2007
Determining an anomaly.

If a player is significantly different from the average (the regression line) it would suggest that the player is an anomaly or that it's either been lucky or unlucky (you can generally determine which by looking at their past)
What I find interesting is Backstrom is doing amazing, but his results don't mesh with that of Luongo, Kolzig and Huet who have similar PK%'s, but worse SQN%'s. Cloutier is better than his 0.869 rating would suggest as is Esche. Dunham has been bit lucky and Auld with an excellent 0.915 PK% has done poorly at even strength. You can make your own concussions based on these results, but this is a tool I use to try and figure out to some degree if a player is lucky or actually good. Mason seems to be actually good, of course there are many other indicators that could prove otherwise.
February 5, 2007
Hooking, Holding and Tripping
|
Division Graphs






This is just a messy graph of all the teams together. I post it only so you can see the context of some of the divisional graphs above.
I figured since the graphs might be hard to read that I could post some of the numerical results.
|
February 3, 2007
Goaltender win statistic
W+ =60* (actual wins - expected wins)/GP
Here's the list [GP>10]:
|
It should be noted that luck plays a bigger part in these statistics as there aren't many events (<50), these results could be interpreted as random, however the list is a good representation of how man goaltenders on this list have played. I'll try going back a few season later on to see what happens when I include more data.
February 2, 2007
Scorers: West vs. East
The western curve is significantly different from the eastern curve. You can see that the west has quite a few more players in the 0.4-1.2 category, whereas the east dominates the 0-0.3. In other words the East is not as good as the west because it has more marginal players. Basically the west has more depth: better 2nd powerplay unit, better 5th and 6th defenseman etc.I looked into goaltending as a possibility, but both divisions have the same save percentage.
Statistically out of the playoffs
For example we currently have about 30 games left. A 50% team has a 43% chance of playing above 50% and 43% chance of playing worse (leaving 14% for 15 wins out of 30 games). Of course, what I care about here is that there's a 5% chance that a 50% team wins 20 or more games. In other words if a team needs to win 20/30 games to make the playoffs then they have a 5% chance of making the playoffs (if they're average). Of course most bad teams (those who don't make playoffs) are below 50%, but I'm assuming a team might play a little better to make the playoffs. This is done following the simple binomial distribution. As you get closer to the last game of the season there is a higher likelihood that you could win 100% the remaining games: (1/2)games left, and a similar relationship exists for winning 90% or 80% of remaining games.
Of course winning percentage doesn't make much sense in the new NHL. I use a system that assumes each team will get 8 additional points during the course of the season (10% of games) and so winning percentage equals:
w% = (W+OTL/2)/GP-0.05
Due to the fact that the binomial distribution is discrete you can't solve for an exact value so I did a regression on the last 30 games to show the sort of functional relationship between games remaining and statistically impossible winning percentage.
cut-off = 0.763 - 0.452 *ln(ln(ln(games left))) [or = 1 if games left < 7]
Although it looks complicated, the nested 'ln' functions are necessary to get the appropriate curvature.
Using this equation and a guess for the number of points to make playoffs (~93 points) you can estimate the required winning percentage
Required% = (93-2*W-OTL)/[2*(82-Games Played)]-0.05.
And if required% > cut-off then you have less than a 5% chance of making the playoffs.
Of course the opposite is true if required% < (1-cut-off) then the team is statistically making the playoffs.
Of course if the team is better than 50% for whatever reason (star player just got back from long term injury).
Most of the results are obvious, but in the Blue Jackets, Blackhawks and Kings are out and in the East just Philadelphia. Ducks, Predators and Red Wings are making the playoffs in the West and it appears that Buffalo has secured a spot out East. These statistics are presented on my website.
January 19, 2007
Attendance Demand.
Attendance is about as complicated as they come. I’m not sure I can tackle this with the limited data available and only partial information of the market forces. Attendance primarily depends on the demand for: “going to hockey games”, price of tickets, and population. I should quickly mention there was a change in how attendance was reported in 2000-2001 as such I had to add a $10 increase to the stated value and increase attendance by 200. [Many of the factors involved are complicated and I lack the information or time to look at them all in detail]. Normalizing ticket prices is trivial as there are nice simple CPI calculators (~2.5%/year) that do a great job and it appears that before the lockout the optimal average was about $55. In order to look at demand I had to also normalize attendance. Now there is no CPI for attendance, looking at the years where prices were stable I was able to estimate attendance growth to approximately 2.5% per year. Now the American Population grows at around 1%, so it’s somewhere in that range. I found a 2% growth rate for New York, which probably accurately measures most cities and so I used 2% natural growth rate for attendance, changing this number slightly has large effects on the results. I’m not sure how important ticket sales are in terms of the hockey business, but 16000*$55*41*30 = $1.1B or half of league revenue. If you use all the above data you get a graph that looks like:
Now the regression on price for the pre-lockout years looks like:
[ Attendance = 22574 – 142 * Price ]
Or each dollar increase/decrease in prices costs the organization about 142 people. If you maximize that Attendance*Price you get $80, that certainly wouldn’t be popular and would only increase ticket revenue by about $100M (10%). Or a price increase of 50% would increase revenue by 10%; of course less revenue would be received from consumables and merchandise as well. So I can’t really say whether it would be beneficial without understand all aspects of their revenue. That being said raising ticket prices by a dollar would increase ticket revenue by about $7,000*1230 = $9M.
Some may note that it appears that price of tickets is increasing every year; this is to be expected as population increases:
[ Optimal price: 22574*(1+population growth rate)year/284 = Price ]
In other words prices should grow as fast as population. If not, there’s something else going on.
However the real question that everyone probably cares about is: what was the cost of the lockout in terms of attendance? So I did a quick regression on Price, and whether or not the information came after the lockout:
[ Attendance = 22622 – 143*Price - 1486 Lockout ]
Basically the lockout cost the NHL about 1500 people per game; of course it doesn’t look that way as attendance averages have increased. The past two years of attendance increases have been the result of lower ticket prices and population increases. This costs the NHL about $100M per year until it regains that 1500 loss (if never then you can discount [at 2%] $100M to infinity for a net cost of $5B). The GMs probably believe that the benefits of the new CBA exceeded the costs. Such as the immediate $400M dollar saving in salaries (guess) and possible future saving caused by the cap.
In conclusion I did all this to show that while the NHL is very proud of its attendance figures it’s actually nothing to be proud of. They’ve lost a significant amount of fans and this can be seen by the lower ticket prices. I’m not sure if the future is predicting that the NHL recovers these lost fans (attendance increases faster than population until it reaches it’s old trajectory), however there aren’t really any good signs that is happening at this point. Maybe the 2006-2007 figures will improve as we get closer to the playoffs. But don't panic, Betman isn't even as Colorado see s 12/23 home games not sell out.
January 17, 2007
Diving Update.
I’ve said a bit about diving in the past. And as people posted they wanted to see how it played out through the season. In the first 500 games or so this season the NHL wanted to make a statement that they were cracking down on diving. Interestingly those statements have faded since November. The media appears to have gone silent and interestingly enough it appears there no more upper management pressure on referees to call these penalties. Avery got his third diving call (next requires a suspension) costing him 0.2% of his current income (I guess it’s the cost of being in his business), however there appears to be no news coverage on the fine he should’ve received. I’m curious if the NHL will give Avery the fourth to make a point or if they’re just bluffing about the suspension.
The point of this post however is just to show the interesting trend in diving calls over the season:
| Games: | 2005-2006 | 2006-2007 |
| 0-99 | 11 | 14 |
| 100-199 | 10 | 14 |
| 200-299 | 5 | 12 |
| 300-399 | 4 | 10 |
| 400-499 | 9 | 13 |
| 500-599 | 8 | 5 |
| 600-699 | 8 | 6 |
It’s interesting that the 2005-2006 had the same sort of diving calls occurring in games 200-399 as 2006-2007 is having in games 500-699. Now my database is missing 6 games from the 600-699 set, however 4 of those played today had no diving calls and it’s unlikely that the two games left played tomorrow will have a diving call.
In my opinion, I think there is probably one significant dive per game and probably twice that number of additional minor dives however the NHL seems to believe there is around one dive every ten games and even less when they’re not marketing the diving rules.
January 15, 2007
Plus Minus Graphs II
Instead of graphing the team data as it is, I decided to remove the ice time and plus minus data of the given player so that the two are relatively independent of each other. This closely mimics that of Behind the Net On-Ice/Off-Ice statistics. However it only looks at the games they were involved in (where as behind the net takes the entire season average).
So to get me started I'll use Jovanovski as an example. A top defenseman on a terrible team, as you can see below, Jovanovski has done quite well in his circumstances. Even though Jovanovski has played on a relatively bad team he's managed reasonable numbers.

- Individual (blue) Slope (red): 0.54 ±/HR
- Team (grey-red) Slope (light pink): -0.76 ±/HR
- Difference: 1.3 ±/HR

- Individual (blue) Slope (red): 2.47 ±/HR
- Team (grey-red) Slope (light pink): 0.38 ±/HR
- Difference: 2.09 ±/HR

- Individual (blue) Slope (red): 1.63 ±/HR
- Team (grey-red) Slope (light pink): -1.81 ±/HR
- Difference: 3.43 ±/HR
I'm not sure who's responsible for making Phoenix so bad, but Morris and Scatchard certainly aren't the best two players in the NHL, but on a bad team both of these players are making the team look average.
Morris: ($3.9M)
Scatchard: ($2.1M)
And if anyone is wondering who is responsible for Philadelphia's -1.71 ±/HR you need not look any further than Calder ($3M):
The rest of these graphs can be found on my website by clicking on a players name.
January 14, 2007
Plus Minus Graphs
Hockey often looks at highly variable statistics in order to determine averages that aren’t very accurate. What we care about is the true figure, but it often masked by a week of great play or a bad week. A great start to the season could give an average player a great plus minus for the entire season, whether it’s a measure of his true skill or not it will appear in the final results. A way of fixing this is to look at all the data between the start of the season and the current time and see what the trend is doing rather than just looking at the final result. For this example I’m going to use plus minus statistics. This will remove some of the luck in the results and make the results more real. Take for example Heatley with a 4.1 +/hr and 2.2 -/hr, he is measured as a ±1.9/hr (+23 in 12.1hours), however if you do a regression on his game by game data you’ll find that it looks like:

The spike at end allowing him to go from +12 to +23 in just 4 games, which appears to be an anomaly. The regression spits out a slope of +1.18/hr, which seems a little more reasonable, but significantly different from the 1.9 +/hr, which is the official average. You might also notice the slope doesn’t start at the origin (he is +2 before the season starts), this is an attempt to scale out luck at the beginning and so the regression doesn’t chase the results at the end of the regression. The goal is to find the best slope to fit the data and if one fixes the starting point you can only swivel around that point and so the regression will chase the final value because that is the one it can change the most as a small change is slope has the most effect on the values furthest from the origin.
The benefit of such a method is in terms of error. In general the standard deviation for the plus minus statistic is: sqrt(plusses + minuses), so Heatley works out to about ±8 of course it’s not quite that simple because plusses aren’t purely independent of minuses, although the best error it could be is sqrt(plus-minus), for Heatley this is sqrt(23) or 4.8. Using the regression method we get a standard error for the fit of 0.7, which is much better than the simple average. You can see the same effect with Crosby who is +4.2 and -2.5 (difference = 1.7), the regression says he’s playing at +2.3/hr or 35% than his simple average suggests due to the fact he’s on a mini even strength slump.

It’s a rather simple adjustment to how to look at the data, and it can produce some striking results. Most of the time this should produce the same results as a simple average due to the fact the most likely place for a variable to be at is close to the true mean then anywhere else, but in cases where there is a divergent short term trend that is significantly effect the final results this should produce better results.
The rest of the NHL results can be found on my website.

