Tuesday, November 18, 2014

Guest Post from Burnt Boats: SHOOT IT

BurntBoats with a guest post on stats, because I’m a #nerd. Note, unlike most of my #math, in this case shots means just shots on goal (meaning they either were saved or went in), I excluded only shots and goals on an empty net. Now no more words, let’s get to the numbers.

Shots per line by game this season, remember each line gets more (sometimes substantially) ice time than the next:

1: 12.6

2: 7.4

3. 4.4

4. 2.4

Expected shooting percentage of each line:

1: Well Eichel is on this line so is 100% realistic? But seriously there is really no way to guess preseason what this line would shoot.

2: Hohmann, Rodrigues and Baillargeon (the 3 most common members of the 2nd line) took 220 shots last year, 24 went in. Good for 10.9% which is good but not exceptionally high and it is reasonable to expect them to repeat it.

3. Roberto, Lane (the 2 most common members of the 3rd line) and Moran (who has received frequent minutes on this line when they shorten the bench to 3 lines) took 186 shots last year, and 19 of those went in which equates to 10.2%. This seems like a good proxy as well, and it is slightly lower than the 2nd line, which is logical.

4. Lots of rotating freshman on this line, also hard to predict preseason as well; however, I think 8.5% which was BU’s (not good) team shooting percentage last year is a reasonable number since this year’s 4th line is about as talented as last year’s team. lol. (Plus this includes the D shooting percentage in this 8.5% average, and they bring the average down as they shoot from much farther away, except on Grzelcyk wraparounds)

So in the small sample size of 8 games I would expect that if the 2nd or 3rd line were substantially above or below their expected shooting percentage it is due to luck, because these are largely the same players as last year and they haven’t gotten substantially worse (or better, probably) at shooting the puck past the goalie.

Actual shooting percentage by line:

1. 16.8% - Very good, but I don’t think it is unrealistically high, given the talent of the line

2. 5.1% - Low, fluke-ishly low, since no one would believe that Hohmann, Rodrigues, Baillargeon, and sometimes Olsson/Piccinich are less than half as good at shooting as those first 3 names were last year. I fully expect this to rise to around 11% in time.

3. 11.4% - About right, 1.2% higher than what I would’ve expected going into this season, but Olsson was an unknown at that time.

4. 5.3% - Also low, yet still higher than line 2 (do you believe line 2 is getting unlucky yet?). I expect this to go up as well, but I have less data on it.

Do you still not believe line 2 is getting unlucky so far?

Ok, fine. Let’s look at just Rodrigues. He has played all season on the 2nd line except for 5 periods when he was on the 1st.

How is he doing? He has taken 26 shots and 3 have gone in, good for 11.5%, so he is definitely helping that low 5.1% 2nd line shooting percentage, right? Nope, because of luck (and Thatcher Demko). Of his 26 shots taken 21 were when he was on the 2nd line, just 5 were on the 1st line. Of the 21 just 1 went in, while 2 of his 5 first line shots went in (before you ask, both of those goals were individual efforts against BC, nothing that his linemates did on the 1st line that 2nd liners couldn’t/wouldn’t do). So his shooting percentage while on the second line is just 4.8% so he is helping to bring down the average of the 2nd line, despite his 11.5% shooting percentage this year. These statistical anomalies tend to straighten themselves out as the sample size grows later in the season

So, what do I expect each line to score per game?

Well that is just shots per game times expected shooting percentage:

For line 1 we will use their actual shooting percentage, since there was no expected amount: 12.6*16.8%=2.12 (ok that’s absurd)

Line 2: 7.4*10.9%=0.80

Line 3: 4.4*10.2%=0.45

Line 4: 2.4*8.5%=0.20

Oh, the Defense can score also?

Yes, that is a part of hockey. It is completely unrelated to what provoked this post, but why stop now.

Last year the D took 298 shots and scored 15 times, so 5% went in. So far this year the D has scored 3.1% of the time, 2 goals on 64 shots in 8 games. It seems reasonable to expect our D will be at least as good at shooting as they were last year, so 5% seems fine.

8 shots per game*5.0%=0.40

So now the expected goals per game:

Line 1: 2.12

Line 2: 0.80

Line 3: 0.45

Line 4: 0.20

Defense: 0.40

Total: 3.97

I shouldn’t have to tell you this, but 3.97 is quite good.

The actual goals per game in our small sample size so far:

Line 1: 2.12 (this is the same, since we used actual shooting percentage for the expected)

Line 2: 0.38 (so they have been very unlucky so far)

Line 3: 0.50 (they have been slightly lucky)

Line 4: 0.13 (slightly unlucky)

Defense: 0.25 (also slightly unlucky)

Total: 3.38

Not as high as the expected was but still high, even with more bad luck than good luck.

How dependent is this team on the first line?

Of those 3.97 goals per game lines 2 through 4 would account for 1.45 of them. Lines 2-4 plus the D would account for

1.85 of them. These numbers in a vacuum tell us nothing. Let’s take them out of said vacuum.

The 2008-09 team is well known for getting scoring from everywhere, so what did they do?

In rough terms (not nearly as precise as the stats for this year):

Line 1: 1.20 (-.92 vs expected of this year)

Line 2: 0.96 (+.15)

Line 3: 0.58 (+.13)

Line 4: 0.47 (+.26)

Defense: 0.62 (+.22)

Total: 3.82 (-.15)

So that team was much more balanced, but that was to be expected, I did not expect this year’s 2nd and 3rd lines to be so close to that year’s, actually. What this means is if the 2nd and 3rd lines this year each achieve their expected shooting percentage they just need to raise their average shots per game by 1.5 to surpass the famously good 2nd and 3rd lines from 2009.

The 1.09 GPG generated by the 4th line and the D combined that year will almost certainly not be matched by the 4th line and D this year, but they don’t need to for this team to reach 4 GPG, a number which would surely give BU the best offense in the nation.

So what I am saying is the 6-1-1 BU Terriers who are 6th in the nation in goals per game are getting unlucky on offense and I expect them to improve?


On the flip side it is unrealistic to expect O’Connor will be able to maintain is .949 save percentage throughout the entire season as luck is likely working in his favor so far, so I expect that to slightly worsen as the offense improves.

Stats are never a perfect predictor of the future, but us #nerds love #math. Hope reading that wasn’t too painful.

You can follow Burnt Boats on Twitter @Burntboats

If anyone else wants to write a Guest Post, we have no standards so if you write it, then we will post it.

No comments:

Post a Comment