goals against average and team strength

[grandmeadowhockeyfan]

I have done some work on how goals against average, goals for average, sog for and against, and strength of schedule corelate to strength of teams. I used the rankings on minnhock.com and create a database with the above mentioned factors. goals against average was the one that seperated the teams most like the rankings produced on the web site. The second strongest correlation to the ranking was opponents shot on goal. I thought this was interesting. Any thoughts out there about this.

[O-townClown]

The truth is this is multifactorial. Last year's goals for or goals against don't mean as much if they lost almost all their players.

If I only have three things I can pretty well rank how teams will do this year.

• last year's record
  last year's SOS
  number of returners

[youngblood08]

I would say a combonation of these. least being GA or any other stat.

You should have used these,

Zip Code
Private/Public
Conference
Previous Bantam team success
Players Staying/Leaving

Rosters change too much year in year out, you can use those stats to see historically how good a program has been but these you will see how good a program will be.

[grandmeadowhockeyfan]

The reason i think goals against is important is because if you have a goalie with a save percentage of 0.900 that is returning and most of your defense returning you sure can expect a good season coming. On the other hand if you have no goalies from last year with a save percentage of 0.850 or less and have of your defense from last year is gone you will not have such a good year. You can eliminate many teams from contention based on that.

[NumberCruncher]

I appreciate the work grandmeadowhockeyfan. This could trigger some good discussion.

I understand that of the 4 or so variables you used, they had to come out in some sort order. You said GA from last year was the best of the 4, but did you get a R^2 value or anything? I'm going to assume they weren't real accurate. I agree with the others above; it would be very difficult to get reasonable predictions for the upcoming season based on one last year stat. One would have to account for graduates, transfers, coaching changes, injuries, impact of players who didn't play much or at all last year, etc.

Also, here have been studies at the NHL level that show that the home team is quite biased towards stats like hits, SOG, and takeaways, so I would recommend trying to sift through and use only road SOG for this study.

[grandmeadowhockeyfan]

If anyone is interested i want to explain my stat calculation. To start i listed all 157 teams based on the pagestat rankings. From there I went to followthepuck.com and hockeyhub.com and collected the goals for, goals against, strength of schedule, sog, opponents sog, and pagestat rank for each of the 157 teams. This info was placed on a excel spreadsheet. After that was completed I sorted the teams by each of the above statistics. After that I calculated the average of the statistic in groups of ten. So team 1 through 10 were averaged for each stat and then I did that all the way down to team 157. For example teams 1 through 157 were sorted by goals for. Each of the stats listed above were also averaged. And so down the list.

After all the stats were used as the principle stat and all the stats were averaged for groups of 20 then I calculated the difference between the stats that were not the primary stat. For example I sorted by goals for as the primary stat. Then calculated the difference between groups 1-20 to 21-40 etc. When that was done I found goals against average created the greatest difference in average of groupings in pagestat rankings. Also opponents shots on goal also created the greatest difference in average of pagestat rankings.

1-20 to 21-40 18 difference in pagestate ranking average
21-40 to 41-60 19 difference
41-60 to 61-80 15 difference
61-80 to 81-100 5 difference
81-100 to 101-120 34 difference
101-120 to 121-140 5 difference
121-140 to 141-157 16 difference

I am proposing that goals against created the greatest difference between pagestat rankings per group.

Please let me know what you think

[karl(east)]

Very, very interesting.

If someone really had the time and resources, they could go back over the past several years and tinker with a formula that includes all the factors listed in the poll, with weighted percentages to see what's most accurate.

Another factor I would strongly suggest adding to a serious effort would be some sort of measure of the Bantam A feeder program's success. This would be very difficult to do in a formula, but without looking at the numbers, I'd bet a good amount of money that the "surprise" teams according to the formula all had very strong youth programs feeding in. Not only is it important to take into account who's left a team, it's important to take measure what's coming in, too. There are obviously some problems here--not every high school has a corresponding Bantam team (most notably private schools), the star players might've been first-year bantams, the players who would be in bantams in some programs might be playing varsity out of necessity on others, and so on--but it still has to be taken into consideration, IMO.

If you want to get more specific, here are some things I might recommend:
-Instead of measuring the number of bodies returning, measure the returning percentage of last year's scoring (at least for forwards--D is more complicated).

[ilbok]

Bantam success has been discredited in most years for predicting the State Champions.

OTClown's criteria might seem to be the most reliable at first glance.

Having made it to State the previous year would seem important, too, though almost half the State Champions did not (recent decades).

Coming from a Minneapolis suburb, Catholic school, or Roseau turns out to be the most dominant indicator.

Coaching and tradition.

[grandmeadowhockeyfan]

I think that ilbock has it right. There is no formula but you can not argue that fact.

[grandmeadowhockeyfan]

Bantam success has been discredited in most years for predicting the State Champions.

OTClown's criteria might seem to be the most reliable at first glance.

Having made it to State the previous year would seem important, too, though almost half the State Champions did not (recent decades).

Coming from a Minneapolis suburb, Catholic school, or Roseau turns out to be the most dominant indicator.

Coaching and tradition.

Minneapolis suburb, catholic school, or Roseau or Duluth School. Don't forget about the Duluth schools who always have one of them come down almost every year. Duluth East is regularly comes down. Duluth Central and Duluth Denfeld are combining making them real powerful. Duluth Denfeld came within one win or coming down last year. Hermantown was a powerhouse last couple years. I can gaurantee that one of those teams will be in the tourney '11 this year. What do you Cities folks think of that.

[ilbok]

No, I was right to leave out Duluth.

[grandmeadowhockeyfan]

I agree that strength of schedule is important. That is often how i guage how far a team will make it in the playoffs. One year i charted strength of schedule to advancing in the playoffs. The teams with a 90th or worst strength of schedule never make it out of the quarterfinals. Then from there the single A schools do well, Look at Breck and Hermantown. AA schools with a lower strength of schedule around 80 or 70 are gone by the time semifinals are over.