Contenders from Pretenders

by Toews-makes-funny-faces on Jan 15, 2012 3:47 PM CST

The ToS is fond at times of using the metric of PP% + PK% to gauge the overall special teams is well known here with a total of 100 indicating a efficient Special Teams corp.

A few other indicators can help separate Real Contending Teams from Pretending Teams, and here are three I like using.

SRS (Simple Rating System)

This can be found on Hockey-Reference.com. It is the sum of Strength of Schedule(Average Goal Differential of teams played) with average Goal Differential of the team concerned. A Positive SRS indicates an above average team, with the more positive score, the stronger the Team Play.

DRS (Differential Rating System)

This is something I came up with and you will not find it used anywhere else. It is formula that adjusts Goals For Average, Goals Against Average, and Goal Differential Average with SOS, compares each team to the rest of the league, turns each category into a percentile for each team, then sums the three scores for the DRS. A 3.00 is a perfect score and 0.00 is horrendous and 1.50 being the mean for the league. Above 1.50 indicates better then average Team Play.

GF-GA-L

Another scoring system I came up with that can be quickly figured out. It is what it says, take the Season totals of Goals For minus Goals Against minus Loses (not OTL or SOL) for the score. A positive number indicates strong team play. The higher the number the stronger the Team Play.

SYSTEM ACCURACY

Like any rating system in sports, there will be teams that are hell bent on bucking the trend, but on the whole most teams tend to follow suite, and can be used as a fairly decent gauge to if a team is overachieving, underachieving and/or a real playoff or Cup contender.

GF-GA-L

Since the 2002-2003 season every team with a final season score higher then -10 has reached the playoffs. No team with a score lower then -41 has ever reached the playoffs.

Here is the Max, Min and AVG score by seed in the NHL since the end of the 02-03 season.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
Max	82	46	38	41	50	24	18	4	-10	-16	-20	-41	-58	-62	-74
Min	4	-9	-27	-26	-45	-31	-41	-41	-57	-75	-97	-96	-111	-117	-160
AVG	47.4	16.3	1	9.81	-2.5	-9.8	-22	-22	-34	-44	-54	-68	-83	-92	-111

The 4th seed tends to be higher since it is the first non-divisional winner in seeding and as it indicates usually has a better score the the worst division winner. The progression down in scores coincides very well in Max, Min and AVG for each place 1-15.

Also since the 02-03 season this is the scoring of SC Winners, losers and those that made the Conference Finals.

52	58	60
-3	-24	-39
27.25	7.875	13.91
SC Winner	SC Runner Up	Reached the CF

As can be seen the SC Winner on Average has a score that would generally make them 1-2 seed, while the runner up 2-4 seed. Also most teams that reach the CF also have a score that would place them as from the top 4 seeds on average. But the Min score does show teams with weaker scores do at times make good runs, but no SC Winner finished with a score that would place them much below 5th seed on average (-3 to -2.5 AVG).

This data does show a VERY strong correlation of scores to teams who make deep playoff runs.

SRS

Since the 07-08 campaign only 5 of the 64 playoff teams have had a SRS of less then 0.00 (08 BOS/WAS, 09 NYR, 10 MTL/OTT).

When SRS is ranked 1-16 for the NHL, SRS had the playoff teams 55 of the 64 times (86% accuracy). The lowest SRS to ever reach the playoffs was OTT in 2010 at -0.22, also lowest rank at 21st. The highest to NOT make the playoffs was MIN in 2009 at 0.22 the highest rank to not make the playoffs was the Blackhawks in 2008 8th in the NHL at 0.16.

SC Winners and runner ups had SRS scores of

DET 08: 0.93 - PIT 0.24

PIT 09: 0.28 - DET 0.62

CHI: 10 0.77 - PHI 0.08

BOS 11: 0.56 - VAN 0.87

Teams that reached the Conference Finals since 08 had the following SRS rank (4 seasons)

1st = 2 appearances

2nd =4

3rd =3

4th =1

6th =1

12th and greater = 5

11 of the past 16 CF teams have been in the top 6 in NHL SRS. The lowest to ever make it was MTL at 20th in 2010 vs PHI.

Again a strong correlation between SRS and Deep Playoff runners and playoff teams can be seen with SRS.

DRS

DRS has the same accuracy as SRS in predicting 55 of the 64 playoff teams since 2008. Since that time 9 of the 64 teams have had a DRS lower then 1.50 with Boston in 08 at 0.90 being the lowest. This makes sense since 16 of the 30 teams make the playoffs, just over half the league which would mean some with a lower then mean score would advance to the post season. The highest to NOT make the playoffs was the Blackhawks in 2008 with a score of 2.00.

SC Winners and runner ups had DRS scores of

DET 08: 3.00 - PIT 2.21

PIT 09: 2.07 - DET 2.35

CHI: 10 2.83 - PHI 1.79

BOS 11: 2.59 - VAN 2.97

Teams that reached the Conference Finals since 08 had the following DRS rank (4 seasons)

1st = 3 appearances

2nd =4

3rd =1

4th =1

5th =1

7th=1

12th and greater = 5

11 of the past 16 CF teams have been in the top 7 in NHL DRS with 7 of the 16 being 1st or 2nd. The lowest to ever make it was MTL at 21st in 2010 vs PHI.

Again a strong correlation between DRS and Deep Playoff runners and playoff teams can also be seen with DRS and with SRS.

FINAL

These are three sets of scoring systems that can be used to break Contenders from Pretenders, regardless of their Points ranking. Some good teams may be buried in the points rank my losing many close games while also winning many by large amounts. So if you want to see who the real Contenders are, I suggest using these to help clear up the picture and see who is really faking they are good and who is about to pounce and turn heads.

If you would like a spread sheet that computes all this for you automatically let me know in the comments.

Hawks Rankings as of 15JAN2012

SRS=7th at 0.31

DRS= 8th at 2.034

GF-GA-L =7th at 3 on pace for a 5.5 on the season.

ADDED:

Spreadsheet that daily calculates SRS, DRS and GF-GA-L

Poll

These ranking systems are

Awesome, thanks for pointing them out

7 votes

Meh, don't care, points are what I am concerned about

3 votes

Nope, Flukes happen too much to worry about this

2 votes

TMFF, you need to get your head examined!

5 votes

17 votes | Poll has closed

Tweet Comment 17 comments | 2 recs |

Do you like this story?

Comments

well, I voted.

but srsly, this stuff is awesome (for stats folks).. good work mate! now, if you’ll excuse me, Li’l Benny who sits next to me this period has this interesting new vintage of Paste that he want’s me to try.. apparently the ’Elmer’s ’06 vintage is showing early promise

"Kevin, You can't play Doom Metal while wearing a scarf"

by mightymike D on Jan 15, 2012 4:41 PM CST reply actions

oh, and..

quit bullying me with numbers! There’s an 86.74% chance that you already took my Lunch Money based on a “mmD gets to eat that bologna sammich” index… where I consistently come up with a negative rating despite the “crying for teacher” balancing effect.

"Kevin, You can't play Doom Metal while wearing a scarf"

by mightymike D on Jan 15, 2012 4:44 PM CST up reply actions

This is awesome work

i just wish I could understand it…

"Well now you're just talking like Foghorn Leghorn!"

by justforkicks on Jan 15, 2012 8:45 PM CST reply actions

Just three indicators

That show the strength of team play and likelihood of strong playoff runs, or possibility of making the playoffs at all.

Preemptive Palomino, lets talk political stuff elsewhere .
Recovering rec whore, and SCH's resident troll, lemming and bully.
JUNGLE DRUMS!!! JUNGLE DRUMS!!!

by Toews-makes-funny-faces on Jan 15, 2012 9:21 PM CST up reply actions

thanks, that explanation helps

i’m going to reread it a few tiems and see if i can’t get the numbers working for me

"Well now you're just talking like Foghorn Leghorn!"

by justforkicks on Jan 15, 2012 10:01 PM CST up reply actions

Spreadsheet, yes!

I’d love one, and thanks for the interesting post!

by The_FFF on Jan 15, 2012 10:27 PM CST reply actions

Here you go

Spreadsheet

Preemptive Palomino, lets talk political stuff elsewhere .
Recovering rec whore, and SCH's resident troll, lemming and bully.
JUNGLE DRUMS!!! JUNGLE DRUMS!!!

by Toews-makes-funny-faces on Jan 16, 2012 10:13 AM CST up reply actions

Thanks!

Good stuff!

by The_FFF on Jan 16, 2012 12:22 PM CST up reply actions

Good stuff TMFF

Glad to see the stats discussion revived.

Re the SRS, I think it would be interesting to see:
1) If you made some adjustments for routs (eg cap GD for any game at, say, four) or for late-game empty netters, would the stat correlate even better with which teams make the playoffs? Of course, these adjustments take more work to capture.
2) I don’t really care too much about SRS at the end of the season, because by then I know who makes the playoffs. But how well does this stat as of, say, Christmas, or the all star break, correlate to playoff position after 82 games? That is, how much of second-half performance is due to a tougher or easier schedule, and how much by a team playing more poorly, or better, than in the first half?

I do have some reservations about the other two. For the DRS, I’m not sure this is telling us any more than a stat based only on GD, since GD = GF – GA. And the other one strikes me as some what circular, since it uses losses to correlate to playoff position, which is based on points.

Hey! I take penumbrage at that!

by ratbite on Jan 16, 2012 1:29 AM CST reply actions

As for DRS,

You are kinda right, but GD can vary. Which team is stronger for instance One who Averages 3.00 GFG and 2.00 GAG playing tougher teams (SOS) or one who averages 4.50 GFG but also 3.50 GAG but weaker teams. Both have a GD of 1.00 Obviously the 2nd team is very very potent, but also has very poor defense/goaltending whereas the first has strong Goals scoring and strong defense. This is what DRS will expose, the team with the 4.50&3.50 can be exposed by another higher scoring and well defending team such as the first.

For example
Team 1
2.00 75th percentile
3.00 75 percentile
1.00 90 percentile
Final DRS would be 2.40

Team 2
4.50 100th percentile (Best in league)
3.50 0 percentile (worst in league)
1.00 90 percentile
Final DRS 1.90

DRS exposes the weakness in team 2 defense compared to team 1’s combined Off and D combined.

GF-GA-L

This helps pull out teams that win convincingly compared to teams that may win the one goal games very often. On the average the team that usually wins convingly with the same record as a team that tends to just squeak out wins will be the better overall team.

Example (Indenitcal 45-19-8 records)
Team 1
256 GF 195 GA 19 Reg loses = 41 score

Team 2
230 GF 195 GA 19 Reg loses = 16 score

Remember these are also OONLY gauges to quickly compare teams overall strengths. It does not factor in injuries or other factors like that that can effect team play. Just uses the net results to compare on team or teams to another set.

Preemptive Palomino, lets talk political stuff elsewhere .
Recovering rec whore, and SCH's resident troll, lemming and bully.
JUNGLE DRUMS!!! JUNGLE DRUMS!!!

by Toews-makes-funny-faces on Jan 16, 2012 8:21 AM CST up reply actions

As for your questions about SRS

I don’t know. I use Hockey-References output and have only been keeping DTD records starting with this season. As for adjusting them to a select amount of games, I do not have anything to be able to do that but will look into it at least for next season, if not this season since I already planned on something similar to DRS.

Preemptive Palomino, lets talk political stuff elsewhere .
Recovering rec whore, and SCH's resident troll, lemming and bully.
JUNGLE DRUMS!!! JUNGLE DRUMS!!!

by Toews-makes-funny-faces on Jan 16, 2012 10:06 AM CST up reply actions

My FanPost from last season might help

I did it in February and it was “fairly” accurate at season’s end, but it didn’t do well in predicting teams that got hot or went cold. That’s the problem with SRS, it’s not very predictive.

Although it did predict an 8th seed and about 97 points for the Hawks.

And the correlation between SRS and Points over the long term is tight.

I had put together a FanPost that I don’t believe I ever posted last season looking at SRS and playoff performance. SRS was a much better predictor of success than seeding or W-L-OTL (If I remember correctly, since the lockout, the team with the stronger SRS won the series 75-80% of the time). But even stronger was to use SRS and SOS (even though SOS is part of SRS). I’m fairly certain the team with both a stronger SRS AND SOS has never lost a playoff series since the lockout.

Nice job TMFF. The reason I’ve personally moved away from using anything based on goal differential is that Corsi and/or Fenwick are better indicators of team strength IMO (and give a lot larger sample size). Perhaps a model based on one of those, adjusted for Sv% and maybe even special teams, would be more telling?

SCH, a softer, kinder place to comment on violent sports.

by ChicagoNativeSon on Jan 17, 2012 1:58 PM CST up reply actions

I would tend to agree

On your CORSI and FENWICK portion, but they do not include a very important aspect of Hockey that does decides a lot of games, especially closely matched teams, that being the PP and PK. SRS, DRS and GF-GA-L all capture this aspect as well. CORSI and FENWICK are good at indicating possession dominance at EV, but it does not reflect the results of that dominance, being goals, and as mentioned does not reflect special teams either.

So for CORSI and Fenwick I would say would be a good gauge on who is a dominant team, whereas SRS, DRS and GF-GA-L gauge who can close the deal.

These are just gauges I use to quickly assess if a team may be over or underachieving based on where they are in the standings. If I think it, I will look at the team closer, like the PP and PK do they reflect what their GFG and GAG is or are they out of proportion. I will look at CORSI, Fenwick, SV%, SHG, SHGA etc… as well for more in depth analysis.

But on the whole, these seem to be a fairly accurate way of seeing who is the best or better teams, and who is in over their head.

I have a couple teams I think are underachieving right now
PHX: +0.08 SRS (10th); 1,896 DRS (10th) GF-GA-L -17 (12th) yet their point total is only good for 20th in the NHL.
PIT: +0.31 SRTS (8th); 2.206 DRS (6th) GF-GA-L +1 (9th), but points they are15th

I would expect both PIT to move up in the standings given their GFG-GAG ratio not being bad. PHX may move up, but they need to address their GD which is poor +1 overall, or they will drop in all three gauges.

Overachieving I have
OTT: -0.8 SRS (15th); 1.414 DRS (17th) GF-GA-L -18 (14th), but have a point total that is 7th in the NHL

I expect OTT to fall in the standings, especially if they do not address their GAG issue.

As for your SRS between teams in the playoffs, I am not sure. I may look into it, but have a couple more things on my shelf first.

BTW, your post last year set me down this path to trying to figure ways to judge a team’s strength regardless of points.

Preemptive Palomino, lets talk political stuff elsewhere .
Recovering rec whore, and SCH's resident troll, lemming and bully.
JUNGLE DRUMS!!! JUNGLE DRUMS!!!

by Toews-makes-funny-faces on Jan 17, 2012 4:00 PM CST up reply actions

Too much number: Did not read

rec’d based on historical TMFF output

Nymphin' Ain't Easy Fly Fishing
Back off man, I'm a scientist

by BigCSouthside on Jan 17, 2012 1:52 PM CST reply actions

Thanks

I always appreciate recs! lol

Preemptive Palomino, lets talk political stuff elsewhere .
Recovering rec whore, and SCH's resident troll, lemming and bully.
JUNGLE DRUMS!!! JUNGLE DRUMS!!!

by Toews-makes-funny-faces on Jan 17, 2012 4:04 PM CST up reply actions