Archive for October, 2012

I had planned on continuing the work I started last week and looking more at shots generated per zone entry, but I have hit a snag in the data analysis, so that is on the shelf for now.

Instead, I wanted to go back to the Home/Road number crunching that I showed last week. I did a basic t-test on home and road numbers of total zone entries, controlled entries, and uncontrolled entries. This week I brought in some more even-strength data from the Holy Grail of hockey stats, Behind the Net. Specifically, I matched even-strength shots on goal and even-strength goals to corresponding games with my zone entry data.

It is anecdotally understood that the home-ice advantage is not as great in hockey relative to football, baseball, and especially basketball. Indeed, the numbers back up this assertion, and we see that many measurements are not significantly different on the road vs. at home:

Home Away Sig?
Even-Strength Shots Taken 20.12
(4.92)
20.80
(4.84)
No
Even-Strength Goals Scored 1.46
(1.14)
1.33
(1.05)
No
Uncontrolled Zone Entries 31.9
(7.6)
30.8
(7.7)
No
Controlled-Uncontrolled Entry Ratio 0.467
(0.079)
0.439
(0.092)
No

But with a good sample size, we see that the Wild have a significant change in their style of play at the friendly confines of the Xcel Energy Center. Specifically, they play better in the neutral zone at home, and hit the blue line with control more often.

Home Away Sig?
Total Zone Entries 59.5
(9.53)
54.9
(9.1)
Yes
Controlled Entries 27.6
(5.9)
24.1
(6.6)
Yes

On average, the Wild get about 4.5 more zone entries at home versus on the road, and about 3.5 more controlled entries. There are a couple reasons why this might be. The players may be more confident playing in their home building in front of a friendly crowd, they may benefit from better rest by not having to live out of a hotel room, or their opponents may play more conservatively on defense, not wanting to give up a big momentum-shifting play.

But it’s notable that although zone entries and controlled entries are significantly different, the number of shots taken does not differ. If and when I am able to look at shots generated off zone entries, I will be better able to shed some light on this. Also, I found 5v5 shots on goal numbers per game, but not 5v5 Fenwick totals. There may be some differences in SOG vs Fenwick numbers (though I doubt it.)

In psychology we look at statistical significance, but we also have a thing called clinical significance. Basically what this means is that although two measurements can be statistically different, in a real-world sense the difference may not be that notable. In hockey terms, we see that the Wild get about five more entries on average at home, and about three more controlled entries. But knowing that each controlled entry generates 0.6 shots and each uncontrolled entry generates 0.3, that difference may only account for two or three shots a game. Considering the most elite snipers in the game convert around one out of eight shots or so into goals, the difference in zone entries may not be ‘clinically’ significant.

Ultimately, the difference is notable because so many measurements show no difference between home and road, so the fact that we are finding an effect may be a clue that there is more to discover. And even though the actual difference may turn out to be just a few shots, the reason we look at zone entries in the first place is as a surrogate for possession. Sure it’s just a couple shots, but more shots means more possession and more time near the opponent’s goal, and that’s never a bad thing. I still want to look at shots generated from entries instead of just overall 5v5 shots, and Fenwick if available. But research is incremental–we take small steps and develop these building blocks that we use to develop our overall knowledge base.

What are your thoughts on why the Wild seem to play better in the neutral zone, and do you think there is such a thing as ‘clinical significance’ in hockey? I’d love to hear what you have to say, so please leave a comment below and don’t forget to follow me on Twitter, @Hashtag_Hockey. Thanks for reading!

I’m not sure if you’ve heard, but we may not have hockey for a while. What’s a blogger to do? You may have seen some of my previous posts about my other hobby, homebrewing. I’ve recently (read: today) gotten into home coffee roasting, and I thought I’d share my experience because…well, why the hell not!?

I bought my coffee roasting setup from the same place I get all my homebrewing stuff. It’s a website called MoreBeer! and their storefront in Riverside is right near where I work. Very friendly and knowledgeable guys. I bought the cheapest starter kit they offer, which is actually just a modified Whirley Pop Popcorn roaster. The only modification is a small hole drilled in the top to fit a thermometer. It’s a very simple setup, and the roasting is done on the stovetop. Here’s a photo of the gear:

I have never made my own popcorn so I had to take just a quick second to figure out the contraption. It’s just got a crank that spins the tines (you can barely see one there inside the pot) so you can turn the beans while you’re roasting them so they don’t scorch. The theory goes it’s an idiot-proof setup, and even a moron could figure it out. That theory was put to the test…

The beans I chose to roast first are “Brazilian Pulped Natural” and the description from the shop is as follows: “Soft fruity vanilla in aroma. Light acidity and medium body. Fruity roundness with a slightly nutty and buttery fullness. Awesome!” I had been reading up on the process and seemed confident enough to give it a try. I was shooting for a medium roast, and started with about 1/2 pound for my first try. Here is the BEFORE shot of the green coffee:

Brazilian Green Coffee

BEFORE

PROCEDURE: I heated the pot to 400 degrees with nothing in there, it didn’t take more than about seven minutes. Then I dumped in the beans and started the clock. I was very surprised that the temperature dropped very rapidly to about 350, and even a little below that, closer to 325. The beans must have sapped A LOT of that heat right off the bat. I had to goose the flame up to just over medium heat to try to get that temperature up to roasting (I was shooting for a roasting temp of 375.) Next time I will probably go to 425 or 450 even before I throw in the beans.

After about three minutes, the temp got up above 350 again. The whole point to coffee roasting is to listen to the “first crack” and “second crack” where the beans actually make popping noises, not dissimilar to popcorn. After about 5 or 5 1/2 minutes I was pretty confident I was at the first crack. The beans were still a little green and I was shooting for a medium roast. I ended up stopping at the 8:00 minute mark. The smell was definitely pungent–it actually smelled a little burned but I figured I hadn’t actually burned the beans…that’s just what the normal smell is. I quickly moved the beans to a strainer and aired them out. Next time I will see what I can do to get the temperature down more rapidly–as in homebrewing, they are at their most vulnerable while they are cooling down like that. Bacteria are most likely to start growing, and oxidation is most likely to occur. I need to read up more on it, but I believe there is a way to sort of just spray water on the beans to cool them down.

AFTER

All in all, the whole process took around 20 minutes, 30 with clean-up. The pot did not get much of any residue on it, and was very easy to wipe down and then scrub out.

I will be grinding these up starting probably tomorrow morning, so I’ll update this post and give my thoughts on how it tastes. I will be interested in different recipes in the future, this is just a straight roast with nothing extra. My favorite kind of coffee is a vanilla french roast, so I’ll probably try to roast with real vanilla beans sometime.

Thanks for reading, and please follow me on Twitter @Hashtag_Hockey. Here’s a good song to have a cup of java and groove to.

I finally finished data entry for this Minnesota Wild Zone Entry project, so I now have all 81 games of the 2011-12 season that were available through the NHL.com Game Vault. For detailed methodology, check out my recent post with a specific rundown of all the work that went in to collecting the data, including the Mystery Game that disappeared into the Bermuda Triangle of teh interwebs.

Now comes the fun part, which is an arduous and thorough statistical analysis…a-yep that’s what’s fun to us stat heads. Now you know why the ladies are just throwing themselves at us.

I have way too much to cover in just one results post, so I will be choosing a few research questions per week and looking at outcomes based on particular themes. To start, I wanted to look at team-level results. Specifically, the questions for this week are:

1) What were the overall outcomes for the Wild?

1) Do the Wild perform significantly better or worse on the road compared to at home?

2) How do the Wild perform against the other teams in the Western Conference, and specifically the Northwest Division?

Mean Greenies

The first thing to do when making sense of a dataset is to look at basic descriptive statistics. From the Philadelphia Flyers research done by the boys over at Broadstreet Hockey and my research on the Wild, here’s the 10,000 foot look at what we know:

  • A team generates 0.6 Fenwick shots (shots on goal plus missed shots) per controlled entry (carry-in or pass-in).
  • A team generates 0.3 shots per controlled entry (dump-in, tip-in, or other).
  • The Wild, on average, had a ratio of 46% controlled entries to 54% uncontolled entries.

These are very interesting findings, and it is remarkable to note that for the Wild and Flyers, those numbers are almost identical over the course of a season. However, they are also a bit reductive–to boil them down to simply an average does not tell the whole story. As they say, “The average Canadian citizen has one breast and one testicle.” On the first day of Stats class, they teach you that any look at a distribution of data should include two parameters: mean and variance. That is, where is the average, and how spread out are the data?

The distribution of the Wild’s entry count is pretty normal–they had a handful of games where they got up into the low-seventies, but for the most part, they settled in around 50-65. There are a couple of games down into the 40s and even mid-30s…ouch. The club was more consistent on the number of controlled entries, and you can clearly see a few games where they chucked the puck down the ice upwards of 40 times. The spreadsheets that we have do break down the entries for score effects (close games vs blowouts) but I’ll have to dig into that data another day. The ratio chart is pretty well grouped, but you can see how often they were favoring uncontrolled entries, there are just a few games where they were north of the 50-50 mark. For a club that is trying to shake off their neutral-zone trapping reputation, they are still reliant on regaining possession after a dump-in.

Home Sweet Home

It is generally regarded that the home-ice advantage is less prevalent in hockey than other sports. Still, it’s always a good idea to look at the numbers to see how the data shakes out. How to read the following tables: mean values are presented in each cell, with standard distribution presented below in italics. The ‘Sig?’ column represents whether the home and away performance was significantly different (t-test, p < .05 benchmark.)

Home Away Sig?
Total Entries 59.5
(9.53)
54.9
(9.10)
Yes
Controlled Entries 27.6
(5.9)
24.1
(6.6)
Yes
Uncontrolled Entries 31.9
(7.6)
30.8
(7.7)
No
Ratio 0.467
(0.079)
0.439
(0.092)
No

The Wild played better in the neutral zone at home than on the road, and while they were able to possess the puck into their offensive zone more at the Xcel Energy Center, the number of dump-ins and the ratio of controlled-to-uncontrolled entries was not significantly different. My first thought is that these numbers reflect the youth of the team–when in front of a friendly crowd, they seem to play more confidently and look to carry or pass the puck into the zone rather than letting it fly. In a future post, I will add in the shots generated data and do some more thorough analysis to see if there is anything more to be learned about the club’s home/road tendencies. I am also interested in an extra bit of data collection where I track dump-in recovery rates in addition to just zone entry tracking.

We Sucked, But At Least We Beat Edmonton

One of the main questions I have wanted to look at for a while has been whether and how the Wild change their strategy or perform differently against different teams. I know I have mentioned a couple times some things I will get to in future posts, but definitively my next one later this week will be to do a power analysis to determine whether we can look at single-season opponent performances–the Wild played the Canucks, Flames, Avalanche, and Oilers six times, other teams in the Western Conference four times, and a couple teams from the Eastern Conference twice. Particularly for division opponents but also conference opponents, is that enough of a sample size to draw conclusions? Even though six games is a pretty small sample, I’ll have to wait and see what the tests say. For now, it is important to interpret the following charts as snapshots of one season. We’ll know in a week if there is more to be learned. **Click on these charts to see full-size versions.

There, did you see it!? The Wild played on their heels against every single team in the National Hockey League…except for one, they of the three consecutive first-overall picks, the Seattle Edmonton Oilers. They did play a 40-60 controlled game, but they ended the season with a 4-2 record against the Oil. The Wild had an easier time gaining the zone with control vs the Avalanche, against whom they managed to play a 50-50 controlled game on their way to a 3-3 record. Considering the Canucks won the President’s Cup, I think the Wild held their own in the zone entry portion of the game. They managed to almost match the Nucks in number and ratio of entries, but wound up with a 2-3-1 record. From the Pacific Division, the Coyotes and Sharks were interesting opponents, as each heavily favored a dump-and-chase game vs Minnesota. Although they only played twice, the Jets proved to be a very even opponent for the Wild, though both games went to Winnipeg.

So what have we learned? The only significance tests I ran for this post have revealed that the Wild are better able to gain entry with control while at home, but their uncontrolled entries and ratio of entries are the same. Breaking down performance versus specific opponents is interesting to look at, but more tests are needed before I can say anything definitive.

I am *always* interested to hear how others interpret the results, so please let me know your thoughts! I would be very happy for any comments left on this page, but you can always reach me through e-mail, hashtaghockey [at] gee mail [dot] com, or through Twitter, @Hashtag_Hockey. Thanks for reading!

I can barely walk and drink coffee at the same time–the OSU marching band covers all the classics: Tetris, Mario, Space Invaders, Pac Man, Zelda, Halo… plus don’t miss the galloping Epona at the 6:00 mark

In my last post, I offered a breakdown of how much each MLB team spent per win, on average. For a different analysis, today I put the teams together by division standing, and showed how much each team added or subtracted between Opening Day and the end of the season. For reference, the bubbles are a reminder of how many wins the team had. They are on a secondary axis which is kept at scale, so some teams’ bubbles overlap a bit clunkily. **Charts are compressed to fit on the page. To see an enlarged chart, click on it.**

As always, there’s a reason I have a category called “Damned Lies and Statistics.” The Orioles Yankees, and Blue Jays each added between $6M and $7M, but they started from vastly different amounts–New York just under $200M and BAL/TOR each between $80M and $90M. The Rays are disappointed that they didn’t make the Tournament, but they continue to win the Moneyball game: the club started the year with a $64M payroll, cutting almost $3M by season’s end, and still winning 90 games. The less said about the Red Sox here, the better. Less than two years after signing Adrian Gonzalez to a seven-year, $154M contract, they bailed and traded him to the Dodgers. If you’re going to be a 70-win team, you might as well be a 70-win team with a $100M salary than a 70-win team with a $175M salary. To borrow a hockey term, the Red Sox are back to being a Gong Show–go Sawcks!

Despite winning the division, the Tigers were projected to win quite a bit more games going into this season, and if you listened to Jonah Keri and bet the under, you’d have a bit more coin in your pocket. There’s a number of jokes I’d like to make that start out with “Winning the AL Central is like…” but they say discretion is the better part of valor so I’ll just say signing Prince Fielder turned out to be a good move, and the Tigers have themselves a Ringer! The White Sox added salary as fast as they could, but fell short of winning the Division by 3 games and the Wild Card by 8 games. Better luck next year boys! I grew up in Minnesota and am a Twins fan, so it pains me to see how they have gone from an underspending, overachieving club to an overspending mess of underachievement. Justin Morneau is one of my favorite players, but the contract they gave him sure looks a whole lot worse than it did before he took that knee to the side of his head…

The Athletics are this year’s Moneyball posterchildren once again, but as of this afternoon they are down 2-0 to the Tigers in the ALDS after Don Kelly hit a walk-off sac fly. It’s a shame that the lasting quote from the movie may be, “Our stuff doesn’t work in the playoffs…” The Rangers added almost $40M to their payroll, but in a sort-of parallel to the A’s, Texas fans know all too well that if you don’t win the last game of the year, it doesn’t matter what you did in the regular season. The Angels will probably be good next year, and their payroll spike came before this season started, but for all the talent on the that roster if they don’t figure it out, that franchise might be in trouble. Here’s a fun fact: Albert Pujols made $162,337.66 per game this year, while Mike Trout earned $3,018.88 per game. That’s almost 54 times as much. Yay big contracts!

The Washington Nationals continued their upward mobility, after finishing almost exactly at .500 in 2011 and a third-place finish in their division, the Nats ended this year with the best record in the NL and a modest $97M salary at year’s end. The Phillies have not one but two bloated contracts, and continue to underperform. As for the Mets and the Marlins…yeah…it’s the Mets and Marlins.

Cincinnati is having a nice go at it recently, with the NL’s second-best record on right around a $100M salary. But Joey Votto’s contract alone accounts for 21% of the club’s payroll. If they want to keep the momentum going, they will have to find some good production out of some team-friendly contracts. The Cubs are doing Cubs things, although Theo Epstein will probably need more than a year to turn around that tire fire of a franchise. Then there’s the Astros…what to say about this club. They are likely going to get their own post–as you can see, they got rid of $44M over the course of the season. The thing is, their Opening Day payroll was just over $60M, and their end-season payroll was just $16.1M! What’s more, every contract they had (according to spotrac.com) was up in 2012! There’s rebuilding and then there’s what this club is doing.

This chart is not to scale…because if it was it would just look like one bar dwarfing all the others–the Dodgers added $110M to their roster this year (they started at $95M) so they more than doubled their Opening Day payroll. I couldn’t figure out how to have what’s called a “broken axis” in Excel, so if anyone knows how to do that, please let me know! The club won’t really have to answer for their spending until next year, and one would think that over time, the difference between James Loney and Adrian Gonzalez and Dee Gordan and Hanley Ramirez will become evident. One would think… The D-Backs, Padres, and Rockies were squarely in the “also-ran” camp, and didn’t really seem interested in adding or subtracting to their roster.

That’s all for today, thanks for reading! Please let me hear your comments, and make sure to follow me on Twitter, @Hashtag_Hockey

Moneyball got me into sports analytics (the book not the movie…hipster nerd slam!) and after every season I like to take a look and see how much money each team spent per win. The methodology is pretty quick-and-dirty: I looked at Opening Day salaries (source: USA Today Online) and year-end salaries (compiled from Spotrac.com) and averaged the two together, then divided by number of wins. There are more sophisticated ways of determining how much the team actually spent, but this is intended to just be a quick look. **The charts have been compressed to fit into the format of my site, click on them to enlarge.**

And for reference, here are the tables with the actual values:

.

Thanks for reading, please remember to follow me on Twitter, @Hashtag_Hockey!