Woodguy has an outstanding bunch of info in today's Lowetide on using relative GF% to assess dmen - and as some of you know - I think using GF% with lots of context is one of the best tools out there to assess the net contribution of what players create AND what they give up.
Lowetide.ca | Peace frog
Lowetide.ca | Peace frog
Lowetide.ca | Peace frog
Adam Larsson is the stud dman on this team, and good god, if you gave him an equally good 2 way partner who has a bit more offense in his game they would be a legit top tier pair in the league. Some real food for thought on what we should do with the defense - and see below for the start of the conversation.
-------------------------
If you were to look at a single metric that tells you how a player changes the goal share for his team when he is on the ice, that metric is Relative Teammate GF%.
Its not a “player on”/player off” metric like normal Relative metrics, rather its the aggregation of the WOWYS.
I think its a very interesting and informative stat.
Like all stats, its needs a lot of context.
In this case, very good players on bad teams will always look very good as their team mates’ GF% will be better when on with the better players.
You also have to be careful about Dmen. A Dman’s results are also that of his partner, and I don’t know how to get Relative Team mate GF% for pairs of players.
You also need large sample sizes. One season is not enough to account for PDO heaters etc.
You also need to discount the results of 3rd pairing Dmen a bit on when they play on a team when the coach hard matches Dpairs (which is usually all teams)
When a top 6 forward plays with a 3rd pairing Dman its usually vs lesser comp and a top 6 forward will have a better GF% vs lesser comp.
All that being said, here is the EDM results of 5v5 Relative Teammate GF% using the last two years as the sample:
(all traded players only have their EDM results in the sample)
Player RelT GF%
CONNOR.MCDAVID 15.4
PATRICK.MAROON 4.8
ANTON.SLEPYSHEV 4.3
JESSE.PULJUJARVI 4.2
JORDAN.EBERLE 3.8
LEON.DRAISAITL 0.7
ZACK.KASSIAN 0.3
RYAN.STROME 0.3
MILAN.LUCIC 0.2
RYAN.NUGENT-HOPKINS -0.6
MIKE.CAMMALLERI -1.0
JUJHAR.KHAIRA -4.3
ERIC.GRYBA -5.6
DRAKE.CAGGIULA -7.1
BENOIT.POULIOT -7.2
MARK.LETESTU -7.4
MATT.BENNING 7.4
DARNELL.NURSE 6.4
ADAM.LARSSON 3.7
KRIS.RUSSELL -2.1
OSCAR.KLEFBOM -2.1
ANDREJ.SEKERA -4.3
BRANDON.DAVIDSON -5.9
Some thoughts on these results:
1) McDavid is from another dimension.
2) RNH had tough matchups for most of last year and some of this year, but had better help this year (esp. with McDavid….duh)
RNH’s RelTGF% from year to year:
16/17 -9.24
17/18 +9.20
Wild. Maybe leave RNH on 97’s wing eh?
3) Sekera is similar to RNH, but in reverse. His injury was a real *****:
16/17 +5.38
17/18 -28.1
4) Adam Larsson is the best Dman on the team.
Look at Nurse’s and Klefbom’s GF% splits with him over the last two years:
16/17
Klef w/ Larsson 59.9% GF
Klef w/o Larsson 43.3% GF
Larsson w/o Klef 57.6% GF
17/18
Klef w/ Larsson 39.1% GF
Klef w/o Larsson 44.93 GF
Larsson w/o Klef 53.4% GF
That’s crazy.
Klefbom might be the player to trade to get the 2RD.
His best season was last year and he was still below 45% GF away from Larsson. Not good.
I like Klef, but I think its becoming clear that he gives up more than he creates.
To be fair to Klef, he was hurt this year, but last year he wasn’t and his injury history is growing…..
Might be the exact right guy to cash in this summer.
Nurse’s splits with Larsson this year:
Nurse w/ Larsson 56.9%
Nurse w/o Larsson 53.7%
Larsson w/o Nurse 39.5% (see results with Klef above)
5) The Drake Caggiula Experience should probably end.
16/17 -1.99
17/18 -10.0
6) Benning results are always way better than the visual
7) Despite not scoring at all, Lucic didn’t drag down GF% too much for his team mates
16/17 +1.68
17/18 -0.52
8) Russell rode the PDO pony hard last year, not so much this year:
16/17 +2.47
17/18 -4.98
Interesting stuff.
Not the burning bush, but interesting nonetheless
I prefer to use actual goal share when the sample is large enough rather than Expected Goal metrics.
I find expected goal metrics over state the value of shot volume vis a vis goals.
There is skill is scoring goals and stopping goals that shot metrics don’t capture.
They’re great for what they are, but they have their limitations.
They are still the best we have for predicting future goal share, but I don’t think they describe play as well as goal share in 2 season + samples.
They’re great in samples smaller than a season though.