Playoff success 101a

Part II: East Least, West Best?

Following on from Part I of this heroic series (LINK - highly recommended reading, of course), this effort attempts to see if a second hypothesis put forward by our California bureau stands up to statistical scrutiny.

Hypothesis Número B: The Western Conference is lower scoring than the Eastern Conference (and hence has a better chance of playoff success)

The Werk Wot I Did Thiz Time

• Much the same as for Part I (seriously, you have to read it now) – looking at rankings of teams by both Goals For and Goals Against, separated by conference, seeing how these compare and how they translate to playoff performance
• Again, I've looked at the last 10 regular seasons only (only 1997/98 in this sample was under the old divisional structure with Toronto in the Western Conference)
  

Obvious drawbacks

• As before, 10 years isn't a huge sample size and regular season performance doesn't automatically translate to post-season play
• Three of the four expansion teams that started in this period joined the Western Conference – this possibly distorts numbers a bit, but I'd argue that Minnesota and Nashville at least hit the ground running defensively, so didn't produce some of the ugly Goals Against figures of past expansion teams
  

Basic head-to-head comparison
The table below compares the conferences over the last 10 years in the following four categories:

• Average Goals For
• Average Goals Against
• Number of players in the top 20 points scorers (including ties, so 21 or 22 in a couple of years)
• Number of players in the top 20 goalies ranked by GAA
  

(Note that I've allocated players who played in both conferences in one year to that in which he played most games.)



Observations

• All the categories support the argument that the East is higher scoring, but only by a modest amount over the whole period. (In my mind at least, the West had a reputation of being more wide-open than the East in the mid/late-90s – possibly not supported by the numbers.)
• There isn't actually much of a difference in the comparison if you look at pre-lockout seasons only
• There is a much clearer distinction between the conferences in the last three years in each category – perhaps a reflection of much of the young offensive talent going to Eastern teams or that some teams in the East (Carolina and Buffalo, for example) adapted to (or helped create) the more offensive post-lockout style of play
  

Ranking charts
The charts below are near-reproductions of those I included in Part I – these illustrate overall league ranking by Goals For and Goals Against, the stage of the playoffs each team made (based on the key below) and this time, average rank of playoff teams and Conference Finalists in each conference. (As before, the average for each round includes all teams that made that round, not just those that got knocked out.)






Observations
There are some trends to identify here as well:

• Rankings of playoff teams are similar in each conference for both GF and GA (but slightly favouring offense in the East, defense in the West)
• However, rankings of Conference Finalists show much wider differences – on average, East Finalists are ranked 4.5 places higher in GF than West Finalists but 6.4 places lower in GA
• In addition, East Finalists rank much higher in GF than all East playoff teams, but are ranked similarly in GA – a similar pattern applies in the West but for GA
• This last point suggests that, not only is scoring favoured in the East, it is a better indicator of playoff success within that conference, while defense is a better indicator of playoff success within the West
  

Trend graphs
The following two charts (Goals For then Goals Against) are again similar to those I produced for Part I – essentially a diagrammatical representation of the ranking chart information above in terms of actual goal numbers rather than league ranking. (Note the change in colour coding in a not-entirely-successful attempt at clarity.)




Observations
Much the same as those above – the main point being the difference between East and West when looking at the Conference Finalists only (the gaps between the light and dark blue bars in each chart).

Summary
There does appear to be some distinction in scoring between the conferences, particularly in recent years (interesting to see if that continues) and how that translates to playoff success within each conference. The West have won six of the last ten Cups – very difficult, however, to argue this is just a strong reflection of defense triumphing over offense.

Coming soon
The previously advertised Part III of this series has unfortunately been subject to legal intervention – the producers of “Weekend At Bernie's” have made it clear that a writ will be served if any attempt is made to make a connection to the GM performance of Glen Sather.

Playoff success 101

Part I: Scorin' or borin'?

Welcome to part one of a short series more accurately titled “LW3H Investigates One Of Juice's Playoff Hypotheses, Makes It Unnecessarily Complex And Fails To Draw Any Firm Conclusions”.

Hypothesis Número A can be summarised as: Better defensive teams have more playoff success than higher scoring teams
The Werk Wot I Did

• Order all NHL teams by both goals scored and goals against in each of the last 10 regular seasons (back to 1997/98 – 2004/05 was excluded as it wasn't very interesting to look at)
• Look at how these numbers translated to playoff failure or success
• Construct some colourful (yes, there are two letter u's in that word) tables and charts to try and illustrate whether there are any strong trends
• I've kept this stuff pretty light on number-crunching, (a) to keep it mostly simple and observational, and (b) because I would probably shame my profession with the level of proper statistical methods that I have forgotten how to use
  

Obvious drawbacks

• 10 years isn't a huge sample size, so tough to take really definitive answers from it, but enough to give an illustration – league-wide trends in scoring have varied a lot over 20-30 years anyway
• Regular season performance doesn't automatically translate to post-season play – there are occasions where some poorer defensive teams have stepped it up significantly in the playoffs, for example
• Broadly, the best teams are stronger at both ends of the ice than average teams, so it's hard to say one aspect of the game is clearly more valuable than the other
  

Key



Ranking charts
The charts below show the overall league ranking by Goals For and Goals Against and the stage of the playoffs each team made. At the bottom of each is the average rank of the teams reaching each playoff round in each year, with the overall 10-year average in the far right column. (Note that the average for each round includes all teams that made that round, not just those that got knocked out.)




These figures suggest there is a slight tendency for better defensive teams to get further in the playoffs:

• The 10-year average GF ranking for Cup Finalists is 9.0 compared to 6.9 for the GA ranking – average GF ranking is also worse for the other playoff rounds, but by a smaller margin
• 45 teams ranked 15th or worse in GF made the playoffs in the 10-year stretch (including four Finalists), compared to 38 ranked 15th or worse in GA (two Finalists – Carolina both times)
• Only one top-ranked team in GF (the Devils in 2000/01 obviously...) made the Final, compared to three top-ranked defensive teams – numbers are closer if you extend it to the top two or four ranked in each category, however
  

Trend graphs
The next picture shows a diagrammatical representation of the Goals For data, the red line being the league average over the period, the individual bars the average GF figure for the teams reaching each stage of the playoffs:



Observations:

• The general post-lockout scoring spike
• Playoff teams score well above the league average, as you would expect
• The GF average of the Conference Finalists (blue bars) is only significantly above that for all playoff teams (orange bars) in four or five of the 10 seasons
• 2000/01 and 2007/08 are examples of years where high-scoring teams fared particularly well in the playoffs, the snoozefest that was the 2002/03 playoffs being the clearest example of where they did not
  

The second picture is the same but for Goals Against:



Observations:

• Again, the obvious point that playoff teams have much better defenses than the league average
• The GA average of the Conference Finalists is better (below) that for all playoff teams a bit more frequently than the GF average was, but not a really strong trend there
• In pre-lockout years, there seems to be a slightly clearer bias towards stronger defensive teams going further in the playoffs than there is for stronger offensive teams in the same period (the yellow and blue bars are further below the red line in the GA graph than the same bars are bove the line in the GF graph) – again, not a definitive picture though
  

So, all in all, a lot of work to illustrate not all that much? Probably so – but some interesting information in there, I think. Maybe you can point out some things that I haven't noticed.

Stay tuned for the rest of the series:

• Part II – some West vs East type shenanigans
• Part III – Glen Sather: Still alive or “Weekend At Bernie's” style cover-up?

WPEF Rankings (1997-2008)

"What is WPEF?", I hear you ask

If you read my blog last week, commiserations. You will, hopefully, at least have some understanding about what the figures below represent.

If you missed/ignored/read but got extremely confused by that blog, I suggest taking a glance first (LINK).

As a simplified explanation, WPEF (Weighted Point Efficiency Factor) is a measure of how well a player scores points taking into account:

• Number of minutes that player gets to play
• How those minutes and the points the player gets are split between even-strength, powerplay and short-handed situations (PP points being less valuable or well-earned than ES points, which are in turn less valuable than SH points)
  

You could see WPEF as the number of points per game a player achieved if they had played exactly 20 minutes per game all at even strength.

After looking at player rankings through the mid-point of this season, the figures below cover the top individual WPEF seasons (WPEF over 0.9 - only where the player scored at least 20 points) looking at every NHL season from 1997/98. Yes, that does mean looking at over 8,500 individual player seasons. I'd post the whole lot, but that would probably not be of much interest.

Anyway to the excitement...

Point Efficiency Rankings

Which players make the most of their ice-time?

My already limited literary powers having temporarily evaporated, I have retreated back to the ever-comfortable world of cold, hard numbers.

Perhaps inspired by the return of everybody's favourite procrastinating Swede, Mats Sundin - often the subject of accusations that he was underused during his time in Toronto - I thought I'd see which NHLers produce most once ice-time is taken into account.

How does it work?
The simple thing to do would be to just take the points scored for each player and divide by total ice-time for the season. And I did do this - to make the numbers look nicer I scaled the figures to produce a Points Efficiency Factor (PEF) - effectively the number of points a player gets for every 20 minutes of ice time.

So where are the figures, dammit?
Hold on - I then did the same for points and ice-time broken down into even-strength, shorthanded and powerplay time. This produces efficiency factors called ESPEF, SHPEF and PPPEF. As you might expect, for a typical player PPPEF>ESPEF>SHPEF, since goals and points are more frequent on PPs, less frequent when shorthanded.

OK, I get it. Where are the damn numbers?
But I haven't got to the exciting bit where I lose the few people who made it this far. Why not reflect the fact that PPGs should be easier to score than ESGs which should be easier to score than SHGs? So, I decided to weight the three efficiency factors - i.e. to favour those who score more at even strength and shorthanded.

Yep, you've lost me completely. But carry on...
Roughly, goals/points on PPs are three times as frequent as those at even-strength, which are three times more frequent again as those while shorthanded. The weighting applied is as per the formula below (difficult to describe well in words, but I think it works):

WPEF (weighted PEF) = ((ESTOI x ESPEF) + 3 x (SHTOI x SHPEF) + (1/3) x (PPTOI x PPPEF)) / TOI

where

• ESTOI = even-strength minutes/gm
• SHTOI = shorthanded minutes/gm
• PPTOI = powerplay points/gm
• TOI = total minutes/gm
  

The numbers...NOW!
OK - the list below is the top 100 players ranked by WPEF. It excludes anyone who has scored fewer than 10 points so far this season - it gets a bit distorted for low points, low minutes players. You can see that players with relatively more ESPs and SHPs and those with fewer minutes per game benefit at the expense of PP hogs who play more.

Not perfect, but gives a pretty decent picture of a few players who do pick up the tougher points with limited minutes though.

Pretty sure I could've explained this one a lot better...