Blog 8: Player tiers based on LS-GAA

In this blog I will try to put players into tiers based on their LS-GAA. To start with I will look at the forwards. The thought process is that there’s an elite tier with a small, but very impactful group of players. I have made 5 tiers – Tier 1: Elite players, tier 2: 1st line players, tier 3: 2nd line players, tier 4: bottom 6 players and tier 5: fringe/non NHL’ers.

Here’s the tiers if we look at this season only.

Forward tiers based on LS-GAA_60 in 2019/2020 (TOI>500):

RankAgeCap hitTOILS-GAALS-GAA_60
1-1023.7$4,816,269108811.570.638
11-9325.5$4,179,26310235.580.327
94-18626.4$3,701,6689751.600.098
187-30026.4$3,063,993924-1.43-0.093
301-39926.2$2,443,344770-4.83-0.376

We do indeed see a top tier with a much higher LS-GAA per 60 than the rest of the league. We also see the bottom tier being really bad, but with only one year worth of data this is to be expected. Players can have single seasons that differs quite a bit from their true talent – good or bad.

Therefore, I have also looked at the forward performance over the last 3 years. The age and cap hit is from this season, but the time on ice LS-GAA is the total over the last 3 seasons.

Forward tiers based on LS-GAA_60 from 2017/2018 to 2019/2020 (TOI>900):

RankAgeCap hitTOILS-GAALS-GAA_60
1-1024.6$6,102,900359731.300.522
11-9326.0$5,034,625352015.170.259
94-18626.4$3,742,76730424.190.083
187-30026.9$2,947,6462787-3.19-0.069
301-39726.7$2,167,0272323-11.01-0.284

Now we see the extremities being smaller. We also see a cap hit that better fits the right tier. I think using data sets of 3 years is the best way to go, so I have done that for defenders and goaltenders as well. I have used the same rankings in terms of percentages, so the elite tier for example is the top 2.5% like it was for the forwards.

Defender tiers based on LS-GAA_60 from 2017/2018 to 2019/2020 (TOI>1200):

RankAgeCap hitTOILS-GAALS-GAA_60
1-526.6$5,717,500461026.860.350
6-4625.7$3,427,703386112.760.198
47-9326.4$3,277,40836755.520.090
94-15028.2$4,004,6493871-2.37-0.037
151-19928.7$2,813,2173136-11.88-0.227

Defenders have less impact than forwards if we look at the performance per 60 minutes, but the total LS-GAA contributions in each tier are almost identical. It’s interesting that the cap hits and TOI are similar in tiers 2, 3 and 4. This indicates that GMs and coaches are bad at evaluating defenders (or the model is bad at evaluating defenders). We saw the same thing in blog 7 by the way.

Let’s also look at the goalies.

Goaltender tiers based on LS-GAA_60 from 2017/2018 to 2019/2020 (TOI>1200):

RankAgeCap hitTOILS-GAALS-GAA_60
1-227.5$5,325,000737346.380.377
3-1529.5$2,731,474545423.630.260
16-3129.7$4,290,258697011.620.100
32-5028.5$3,371,9526299-2.39-0.023
51-6628.7$2,245,4434539-19.92-0.263

The difference between tier 1 and 2 is not as big for goaltenders as it was for the skaters, but this is primarily because John Gibson is having a bad year. Otherwise he would have been miles ahead of the pack.

It’s difficult to conclude too much from looking at these tiers, but it does seem like there’s an elite tier consisting of a few players with a huge impact. These are the players worth paying big bucks. Their positive impact will always outperform their paycheck.

I have also looked at player performance based on contract type:

SigningAgeCap hitCountLS-GAA
ELC21.0$852,550661.96
RFA25.2$3,597,9033163.39
UFA30.4$3,871,712280-0.89
Total27.0$3,440,0086621.44

So as a general rule you should not sign UFAs. They are overpaid and perform below average, although obviously there’s exemptions to the rule.

Here is the tier distribution based on contract type, and it’s clearly the same story – an overrepresentation of bad players in the UFA group.

SigningTier 1Tier 2Tier 3Tier 4Tier 5
ELC314191614
RFA1084807567
UFA439579981
Total17137156190162

And finally here are the distribution amongst the teams.

TeamTier 1Tier 2Tier 3Tier 4Tier 5
BOS35626
T.B38544
STL18727
COL07673
PHI14753
VGK14934
WSH051151
CAR14794
PIT08492
NSH17353
TOR18363
DAL06744
MTL05584
WPG05648
NYI04587
NYR05338
EDM033106
MIN14655
VAN13878
ARI134122
CHI13485
FLA05572
CGY06457
CBJ02844
L.A03194
S.J04181
N.J03375
BUF016513
ANA12775
OTT01079
DET012515

That’s it for now. In the next blog I will initiate the work on a predictive model based on LS-GAA.

Conclusion:

  • Dividing players into tiers wasn’t quite as informative as I had hoped. If you have a truly elite player on your team, then you should do everything you can to retain him. Having a lot of middle tier players can also be a winning model though as long as you don’t have a lot of really bad players.
  • Signing UFAs to long term deals is very rarely a good idea.
  • There’s a big mismatch between cap hit and LS-GAA when it comes to defenders.

Stay safe and remember to be kind

All raw data from http://www.evolving-hockey.com and http://www.capfriendly.com

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: