Have you looked at past seasons at all to determine whether the value of GF vs GA as a predictor of performance is relatively consistent in their comparative value?

I would refer you to the link within the article tagged "PythagenPuck". It takes you to a Win Probabilities articles done by Alan Ryder over at Hockey Analytics, which covers a variety of methods to derive winning percentage, and their relative effectiveness over time. I chose PythagenPuck because it's extremely accurate, and I'm able to replicate it's logic within spreadsheets pretty easily.

I appreciate and agree with your anyalis for NHL season performance.

However, I have done further research regarding the NHL Playoffs over the last 40 years (post expansion). Predictabily, Goal differential is a strong indicator of stanley cup championships. However Goals for is actually slightly stronger than goals against.
(average season rank of stanley cup championship 3rd Goals for 5th goals against). Also surprising is this stays consistent even durning the so -called dead puck era 1995 to 2004 when the average is 4th goals for 5th goals against.
I expect this will be even more pronounced in the new era with the advent of more scoring. first case in point Carolina which ranked 3rd last year in goals for and 19th in goals against.

It appears that contray to public belkief defence does not win championships Goal scoring and great goalies is even more important.

Dan the Stat Man

An interesting point, Dan - certainly there may be differences between playoff and regular season success. I don't know if I'd go back 40 years to discern such trends, but it's worth looking into.

I like to use stats to break apart conventioal myths -
So ,My point is, that of all the major trophies the Stanley Cup is clearly the hardest to win and "won almost always by a top team". Here in Vancouver the fans are having 'drug induced fantasies' about the canucks 'going all the way' thinking mistakenly, that a hot goalie can lead a team to the cup. It simply hasnt happened! IN FACT THERE HAS NOT BEEN ONE UNDERDOG TEAM WIN IN 40 YEARS. (You think one hot goalie on an average team would have been victorious ove r the years??)The purpose of going back so far is to just point out that through all the rule changes expansion etc etc etc one thing remains constant the stanley cup eliminates all pretenders. Dreamers like Canuck fans here cling to memories of the 82 canucks the 94 canucks the 95 panthers the00? Capitals the 04 flames 05 oilers etc they make the finals but have not yet hoisted the trophy. We here over and over how 'all you have to do is make the playoffs' and anyone can win?My research shows that the lowest goal differential to win the cup in the last 40 years is approx +.39 per game - and the average is a wopping approx .65 per game(I believe this calculates to 94 and 106 points points approx via pythag though I stand corrected on this). Also I believe that when you list the final season point totals for every year not one team has won the cup from the bottom half of the draw 9 through 16 in current format)I believe most fans dont realize this. My main point is the reg. season DOES mean ALOT. (Even though the media relentlessly says otherwise).More than any other league. This makes sense since it can take close to 27 games to win the cup!It will be interesting to see if these strong trends continue in salary cap era. The conventional view is now that we have the cap it has leveled the playing field the parity has made it so anyone can win. I am predicting otherwise I believe the bonus point shootout format as really deluded fans in Dallas and vancouver and Minessota in particular that they are on par with Ana and SJS 2) That there is a MYSTIC to old Stanley that will hold true again this year.By this research THE CUP is down to ANA, SJ,BUF, OTT, NASH, DET, CGY.(all above .39 Goal dif. )
My guess IS SJS.

dan

I'd like to take a look at your numbers, Dan - just on a hunch, I checked out the 1995 New Jersey Devils, who won the Cup coming out of the #5 seed in the Eastern Conference that year (which I believe is the only 9-16 seed to win out of the last 12 years). Their 13th-ranked offense and 5th-ranked defense combined for a goal differential of +0.31 (lower than your 0.39 value), and were certainly considered an underdog compared to teams like Calgary, Quebec and Detroit.

There is quite a bit to study here, perhaps something I'll save for the summer - what aspects of teams dictate postseason vs. regular season success...

I will send you my research just as soon as I clean it up...

I forgot to mention that I excluded
95 because it was a shortened season .(teams played only 60% of a reg. schedule ) 48 games.
This seems reasonable since my hypothesis is that the full regular season is a strong indicator of Stanley Cup winners.
48 games is simply not a large enough sample. Case in popin the Canucks who have played at a 75% clip in last 40+ games. Are they a 75% team?. In fact the numbers from 95 actually make my claim
stronger. If the Nhl only played 48 game sduring the regular season, it would be much more likely that a team with a lower GD
or an underdog would win.
I agree the summer will give us more time. i am also currently running full correlations on regular season performance and playoff performance to get a fulle picture.

Dan