My summer “project” this year was to come up with some better ways to show something that I think is very important: the lines the Avalanche use, and how they perform in the game. I tracked them last year using excel, but I never really liked how those numbers copied into the blog entries. And, it was a lot of work. So, I dusted off my old Visual Basic books and set about finding an easier and prettier way.
Here’s an initial version of what I’ve come up with:
So far, these reports list the lines and pairings used along with some key stats for each players. As you can guess, my HTML skills are not “all that”, so bear with me. Hopefully, they’ll be easy enough to read.
I think most of the report will be self-explanatory. The only new concept is what I’m calling the ITCS game score. It’s an idea I got from playing EHM (a hockey text sim) - the game would apply a 1-10 rating for each player after a game. While I have no idea who that rating was calculated in EHM, I think the game score I came up is similar in spirit. Basically, a player starts with a game score of 5. If you dress, and do absolutely nothing good or bad in a game (*casts a glance at Tyler Arnason*), you get a 5. Positive stuff you do in the game gets you a higher score, while negative stuff counts against you.
Good Stuff:
- goals
- assists
- positive +/-
- fights (helping the team)
- blocked shots
- shots taken
- hits
- takeaways
- missed shots (yes, even these count, as they do reflect offensive chances)
Bad Stuff:
- negative +/-
- giveaways
- penalties (except fighting majors)*
*for now, roughing penalties don’t count either way
Most of the stats are weighted to reflect the average output of an Avalanche player at that position (forward or defense) last year. So blocked shots count a bit differently for Brett Clark than they do for Ian Laperriere. This is still very much a work in progress. I still need to work in faceoffs and a few other minor stats. And I also may tweak the formula a bit as the season goes on, if I see something that needs adjustment. In particular, I think the formula for D needs a bit more fine-tuning. Still, from the first 3 games, it looks like on the whole it’s more or less working as I intended it to.
Game score isn’t intended to be a be-all-end-all, bottom-line rating. But it’s another tool to help evaluate how well a player performed in a particular game.
This is just the beginning - I plan to tweak and add to this report throughout the season. Now that I’ve got the foundation laid, making additions should be relatively easy (and if the NHL would stop changing around the reports I’m pulling data from, it would be even easier…hint, hint). I still don’t have goalies represented, for example, but those should be coming pretty soon. I also plan to add in powerplay and penalty kill reporting. Next up, I’m hoping to add back the highly useful shift charts that the league, in all its wisdom, has deep sixed.
I always welcome feedback here at ITCS. When it comes to these reports, I’d like to encourage you to let me know what you think. Is there something you’d like to see added or changed? Comments are a great way to let me know (I see all of them). Or, there’s a contact form at the top of the page. Either way, let me know.
I dig it DD. Although I administer site myself for work I am no HTML guru either. I think you can make your tables a little more legible by simply applying some alternating color to the rows. See http://www.nhl.com/nhl/app?service=page&page=PlayerDetail&playerId=8471669&tab=prft
All they do is alternate the row colors so it’s easier to read. The source is a mess from the NHL site because they probably use asp.net or something but here’s an example of the tables they use…
Stuff...
Stuff...
More Stuff...
It looks like you’re using css so I think once you got it dialed in it would be easy to replicate for you.
Like you, I’m interested in the hidden stats of an NHL game. As you well know the nuances of play are rarely captured accurately in most common NHL statistics. I’d love to see the Earned Run Average (ERA) equivalent for defensemen at even strength and the PK. +/- just has so many things going against it that it’s not held up to a very high standard, and rightly so.
As far as what’s there. I think the faceoff title needs a different alignment to the stats below. It needs to be left justified to the stats when there’s no column lines. Right now it seems hard to look at. I know that’s picky but you want feedback right?
Also, in game 3 it shows that W. Smith was 1-2 for faceoffs but his % is listed at 67? Same goes for Hejduk who was 1-4 but is listed at 20%.
I think you’re approaching this the right way. Get some of the very basic things down solid along with some extras like your GS. Then expand it from there.
I’m looking at Game 3 and see that Lappy had TOI of 9:48. I know that’s probably because of his fighting major that he lost time. What I’d like to see preferrably in Red for less or Green for more is how that TOI compares with their season average TOI. So if Svato’s TOI was five minutes less than his season average TOI for three games in a row I could reliably conclude he’s once again in Q’s dog house. Either that or the Avs are on the PK more than average. There’s not lack of needed data here. I’m impressed that you’re taking it on. I think this is a great start.
Damn, the code tag didn’t protect my tr and td html tags from being wiped out. Basically I was just looking at the NHL page source and they are just putting alternating colors on their rows. It looks good.
the shading is a great idea. I’ll definitely take a look at that.
I’m glad you noticed the faceoff title - it shows me you’re really paying attention. I actually meant to fix that earlier this morning, but got sidetracked.
faceoff percent is correct, but maybe I need to be clearer on that - Smith won 2 and lost 1, so he won 2 out of 3 (67%). ditto for Hejduk - 1 win, 4 losses. would you rather see wins-chances instead of wins-losses?
I like the idea of comparing ice time to season averages. I also want to use some color to point out abnormally high (or low numbers) - like Lappy’s 5 hits in that game.
thanks for the feedback - lots of good stuff to work with there.
The game score idea is good in theory, but when you realize that a lot of the stats are tabulated subjectively (hits, giveaways, takeaways, etc), the idea of a game score is really no more valid than if you just arbitrarily decided on a score for each player.
Good idea, but without the stats that the score is based on being collected objectively, the idea is pointless.
Ah, I get it DD. It’s fine now that I know how it works. I guess to me it wasn’t obvious.
Stat tracking may be subjective in some arenas but I don’t think you can let that force your hand. We watch every game and know when the scorer is taking liberties. The same goes for baseball in hits vs errors. I think if ITCS establishes a good statistical system then it will only be as good as the NHL provides of course. But at least the system is there and if the NHL gets more consistent then the statistics will also. Regardless of the bias’, I think you’ll find that Guite will finish the season with 600% more hits than Arnason. No one is going to be overly stingy or generous when players are displaying obvious trends. Subtlies may be deluted because of home ice bias but that’s not something anyone can change but the NHL. What DD is doing is a great step in breaking down the statistics. Some stats are black and white so it will always be accurate in those areas.
At mid season I think obvious tendancies will be noticible regardless of the various inaccuracies of hits/giveaways etc…
I do know that a lot of the stats are highly subjective. Remember, though, that this isn’t meant as some sort of statistical panacea. Think of it more as an amalgamation of a player’s secondary stats for an at-a-glance score.
I’m not really all that thrilled with the numbers for the D, but I think if you look at the 3 games so far and the game scores for the forwards, you’ve got a surprisingly good representation of who played great and who didn’t.
And, again, this is really kind of an experiment - we’ll see if that continues to work as the season goes along.
I like it.
The red/green for the TOI is a superb idea. As, indeed, is different colours for an abnormally good game.
Personally, for the faceoff stats, I would prefer wins-chances as opposed to wins-losses. But that is me being picky. :p
Stat tracking is biased in *ALL* arenas. And nobody said anything about a home-ice bias. I am talking about the stats being 100% subjective in every arena.
Need proof?
06-07 Hits:
Vincent Lecavalier - 112
Scott Hannan - 29
This isn’t a situation where the stat-tracking is biased from time to time, and the home-town guy will give his team a little edge every once in awhile, which will balance out over time.
If you ran the numbers like DD is talking about doing, you would think that Scott Hannan shies away from contact and is a generally useless player, while Vincent Lecavalier is a hard-hitting power forward.
Hits, Giveaways and Takeaways are not the least bit valid as statistics, and to put them into DD’s equations would render them pointless.
1) I calculate it differently for forwards and defensemen.
2) even without the stat-bias (which, again, I’m well aware of), the impact of any one stat - say hits - on a players game score isn’t incredibly large
3) plus, I’m comparing vs the average av player from last season. assuming that bias carries from season to season, you’re still comparing apples to apples.
4) game score doesn’t tell you at all about how Hannan shies (or doesn’t) from contact. I’ve made no claims to that effect. And it certainly isn’t meant to compare an Avs defenseman to a forward on a different team who I don’t even calculate a game score for.
if you want to argue (and I know you do), pick a Col game, choose two players from that game, look at their relative game scores, and make a case for why one should be higher or lower than it is. that’s what game score is - a quick and dirty comparison of a player’s contribution compared to his teammates in any one game.
The bias doesn’t always carry from game to game, much less season to season. The people the NHL hires to tabulate these stats aren’t static from year to year. Last I heard, they were still having interns do it.