Get-Ligaen as good as Allsvenskan

[Karsten]

The comparison of national leagues is a popular topic on these boards.

Yesterday, Aftonposten, a leading Norwegian newspaper, published an article about how Swedish coaches assess the Norwegian league.

Their general view is that Get-ligaen is quite strong, competitive and underestimated internationally.

Frisk-Asker's coach, Sune Bergman who have coached in the Swedish leagues for 40 years (he should know something about this) says that Get-Ligaen is as good as the Swedish Allsvenskan (tier two). Only Leksand and Malmö Redhawks are definitively better and offer better facilities than the Norwegian teams. He also points to Comet's clear 7-3 victory over Bofors, a middle-ranking Allsvenskan team.

Bergman also think that Get-Ligaen is a better development league than the Swedish junior leagues. Rather than going to the Swedish junior leagues or signing too early with an Elitserien club, the Norwegian players should wait until they are fully developed - a view that bear strong reminscences from the Europe-NHL debate.

"We need to get the Norwegian players back to the Norwegian league", Bergman says. "It is great that Steffen Thoresen has returned to Vålerenga. Thoresen had a brief and unsuccesful encouter with Växjö Lakers. But its better for Steffen and Norwegian hockey that he returned so quickly". Steffen Thoresen should wait until he is good enough to get plenty of ice time in Elitserien. This also goes for Mats Z. Aasen, one of Norway's biggest talents. Several Elitserien clubs have attempted to buy him out of his contract with Frisk Asker this season, most lately Färjestad which made an attempt last week.
But Bergman calls this for 'bad scouting'. Mats Aasen is yet not good enough to become a top-two center and play such an important role as his playing style demands. Bergman predicts that Aasen will eventually become a top player in Elitserien, but he has to be patient.

Bergman's views are supported by Håkan Södergren, another Swede that have plenty of coaching experience from Sweden and Norway. Today, Södergren is connected with Djurgården. Södergren says that going to the Swedish junior leagues or Allsvenskan is a detour for Norwegian players. Their development is best served in Get-ligaen. Among others, Södergren points to Patrick Bovim as an example. Playing at HV71's junior team, Bovim was just one in the crowd. In Get-Ligaen, he could have played a leading role.

A third Swede to speak out is Janne Asplund, a former Tre Kronor player who is currently managing the hockey college in Bærum (NTG Bærum). Asplund believes that the Norwegian hockey collleges are better than the Swedish.

"Here [in Norway] hockey has top priority, everything else at the college is adjusted to this. In Sweden, the opposite holds true. There are more hockey colleges in Sweden (23 regional and many local), but noone compares with the Norwegian hockey colleges. Among other things, we travel abroad for six weeks every year to get international experience. Each year we travel to USA for a couple of weeks to play against US college teams, and all in all we play around 20-25 international games each year. This is unique, also by Swedish standards".

Interesting views and defintively food for sort.


http://www.aftenposten.no/nyheter/sp...cle2194394.ece

[Marc Brunengraber]

Interesting read, Karsten. Many thanks for posting it.

One thing that struck me - how can playing a few minutes per game in Norway's top professional league as a junior benefit a young player's development better than playing lots of minutes per game in Sweden's J20 league? That doesn't seem to make sense to me.

I'd bet that - on the whole - Allvenskan is slightly stronger than Get-Ligaen. I'd also argue the same for Allvenskan vs. the Danish league, though. With that said, I don't think there's any question that the top few Danish and Norwegian teams would beat middling and lower-end Allvenskan teams (and in exhibitions, I've seen the results go both ways).

[Graham]

One thing that struck me - how can playing a few minutes per game in Norway's top professional league as a junior benefit a young player's development better than playing lots of minutes per game in Sweden's J20 league? That doesn't seem to make sense to me.

I would imagine because you are also training with a top pro team rather than a J20 team.

Graham.

[Marc Brunengraber]

Given the European emphasis on development through practice as opposed to game time, that would make sense.

[Karsten]

I'd bet that - on the whole - Allvenskan is slightly stronger than Get-Ligaen. I'd also argue the same for Allvenskan vs. the Danish league, though. With that said, I don't think there's any question that the top few Danish and Norwegian teams would beat middling and lower-end Allvenskan teams (and in exhibitions, I've seen the results go both ways).

I know that you have some pretty rigid ideas about the relative strength of the European leagues. But since you don't follow the leagues close hand it will serve you to have an open mind.

The initial post was based on the views of coaches who have many years' experience with the Norwegian and Swedish leagues. In terms of strength, their view is that Get-ligaen is as good as Allsvenskan, not the middle or lower half as you claim. Granted, as for Norway the claim may be a little difficult to prove statistically since the Norwegian clubs don't play the Allsvenskan teams very often. But for the Danish league, we have enough material to make an estimate. First, the Danish teams very rarely play any Allsvenskan teams below the lower half. Second, with a few exceptions the Danish teams consistenly beats the upper half Allsvenskan teams bare the very top teams Leksand, Malmö and Rögle).

For close observers of the European leagues, this should really not come as a surprise. The international records of AAB and Rødovre this season should be well known. These teams are not far behind teams from the very best leagues in Europe, but their results are not lone standing. Last year, Frederikshavn (from the lower half of the league) tied HC Kladno on the road, while Sønderjyske swept their Austrian opponents on their Austrian tour. Later that year, Sønderjyske nearly eliminated Red Bulls Salzburg in the Continental Cup (lost 6-5 in a shootout in the final game at Salzburg).

This issue of strength should be seperated from the question of what is the best development league. As the initial post clearly indicates, the view of the coaches is that Get-ligaen is a better development league than Allsvenskan - not so much because the players receive more icetime as the Norwegian clubs in general have more resources. I also tend to believe that the Norwegian league is a better development league than the Danish league, partly because the young players get a better hockey education due to the close connection with the hockey colleges and partly because they get more ice time in the senior league. Many promising Danish juniors don't get any icetime in Al-Bank-ligaen. This is the most important reason why they move to the Swedish junior leagues.

[Marc Brunengraber]

LOL.....AAB and Rodovre are the two Danish teams I like best.

Aside from them, though, I've seen scores where Allvenskan teams beat Danish teams plenty of times.

No matter. When attempting to rate the strength of many of these leagues, it's an inexact science at best. Plus, many leagues are very, very close in strength.

Take Allvenskan, Denmark Oddset, Norway Get-Ligaen, Austria (Erste Bank), and Italy's Serie A. Teams from these leagues would, on average, be at least competitive with each other. Of the five leagues, does it really matter which one is rated first, and which one fifth? IMO, it does not.

Of course, with that said, I am always interested to try, regardless of the fact that, in the end, it is simply an academic exercise. :-)

[Karsten]

Aside from them, though, I've seen scores where Allvenskan teams beat Danish teams plenty of times.

This is because you don't track the evolution over time. 10 years ago, the Danish teams would typically lose the the teams from Allsvenskan. Now, they rarely lose to any upper half Allsvenskan teams except the very top of that league (Leksand, Malmö and Rögle). This season, Frederikshavn (currently #8) was the only club that lost any games to teams from Allsvenskan position 4-8. Leaving these statistical facts aside, most Danish clubs also have better facilities and much bigger player budgets than the Allsvenskan teams (apart from Leksand, Malmö and Rögle). Furthermore, the Danish clubs very rarely import (or export) players from Allsvenskan; they typically import players from the Swedish Elitserien and the Finnish SM-liiga. So all in all, it is not really hard to figure out that the Danish league is stronger than Allsvenskan.

Of course, with that said, I am always interested to try, regardless of the fact that, in the end, it is simply an academic exercise. :-)

Well, academic exercises cannot always be made from a distance; they often require thorough field trips and/or inside knowledge. To be honest, I find it quite hilarious that you continue to question the wisdom of experts that have many years' experience in the field.

[Marc Brunengraber]

I look at exhibition scores year by year from cross-league play, and stats from players that have played in the leagues being compared, when assessing and forming an opinion about relative strength, as well as by reading interviews with coaches, players and the like.

What makes a league 'stronger'? Having its teams beat the other league's teams more often in exhibition play? Denmark (recently) wins. Having the top team or two when lumping the teams from both leagues together? Allvenskan wins. You could come up with any number of criteria. This is a discussion that has frequently been had on the forums, with no clear answer.

It's safe to say both Allvenskan and Oddset are behind the top seven European leagues. They both fit in the "next grouping," do they not? (i.e., Erste Bank Liga, Belarus, each other, Vyshaya League, Serie A, Get-Ligaen). Again, I don't know that it matters whether Denmark is viewed as having Europe's 9th best or 12th best league....it's somewhere in that range, which makes it a very strong league. As is Allvenskan. Norway isn't quite up there with them, judging on the results I've seen and the things I've read, but it's not too far behind. It's probably slightly better than or equivalent to Mestis but at worst at the bottom end of the top 15 or 16 leagues.

There will never be a clear answer - not even when the IIHF comes out with whatever system it ultimately devises to rank the leagues, as critics will always be able to argue that too much weight is placed on one statistical anlysis versus another.

Agree or disgree, think me an idiot or not, I still found it an interesting read, and I'm glad you posted it.

[Graham]

10 years ago, the Danish teams would typically lose the the teams from Allsvenskan. Now, they rarely lose to any upper half Allsvenskan teams except the very top of that league (Leksand, Malmö and Rögle).

I watched my first Danish game in the 2nd half of the 99-00 season (Herlev v Voyens). I then watched my 2nd game about 3 years ago (Herlev v Nordsjælland) with Karsten. Both games were at the same level of the league table (i.e. the bottom half), but the increase in standard was noticeable. My comment at the time of my first game was that Herlev and Voyens would have struggled in the British 2nd tier (a 10-import league at that time). By my second game, the standard had clearly moved and the game was more akin to a British top-tier game, and the better teams at that.

My third game two nights ago was of a standard I have not seen in the UK since the early days of the old ISL. The speed and precision of the passing in the first period was exceptional, although the game did turn more clutch in the later periods when the closeness of the game became apparent.

Graham.

[RexKramer]

Last year, Frederikshavn (from the lower half of the league) tied HC Kladno on the road, while Sønderjyske swept their Austrian opponents on their Austrian tour. Later that year, Sønderjyske nearly eliminated Red Bulls Salzburg in the Continental Cup (lost 6-5 in a shootout in the final game at Salzburg).

Again this interesting topic.

Vojens sweeping their Austrian opponents is maybe not the right expression (if one wants to go into semantics ;-). There were two games, one versus Graz which ended in a tie (3:3) and one against Villach which Sönderjyske (semifinalists later on in DK) won with 5:4. Later on Graz went dead last in EBEL and VSV reached the finals where they lost to Salzburg - a little bit inconclusive. And the CC game between Red Bulls and Sönderjyske was played in France if my memory serves me right (but that's of minor importance I'd say).

Then I had a look at the this years preseason games again. I reeavaluated the pairwise league scores I have calculated with the current standings in the leagues (ALL, GetL, ALLSvenskan). It is apparent that there the NOR teams only play middle to bottom Allsv teams while the DAN teams play top to middle teams.

Pairwise comparisons:

ALL vs. Allsv: 13 games palyed (4:13 in wins, aggregate goal score 26:42)

The league versus league score I calculated (that is the average of the sqaure root of the winning margins of all games which are weighted by a coefficient determined by the position of the two teams within their own leagues) is 0.57 in favor of Allsvenskan indicating a slight superiority of Allsvenskan. With 13 games played this is one of the more robust league comparisons.

GetL vs. Allsv: 6 games played (2:4 wins, aggregate goal score 20:25)

League vs. league score is 0.49 indicating a slight superiority of Allsvenskan.

To put these scores into perspective we can have a look at the GetL vs. AL comparison.

8 games (1:5 wins, 2 ties, aggregate goal score 19:34)

League vs. league score is 0.97 in favor of the Danish league indicating a pretty clear superiority of ALL.

Of course this does not fit in the picture of the comparisons with Allsvenskan as according to that one ALL and GetL should be very even. This might be a consequence of the lower number of observations involving the Norwegian teams resp. a consequence of the only ordinal nature of how I weigh the game scores based on the intra league position of the teams involved (I just split into top middle bottom assigning a coefficient of 0.9/1.1 to the score of a game between teams from neighboring classes and 0.74/1.25 for top vs. bottom games and of course a coeff of 1 for peer games). If a league however is very unbalanced this can result in a bias (in this case GetL being favored in the comparison with Allsv compared to ALL teams because their Allsv opponents are weaker than the ALL opponents from Allsv even beyond the corrective influence of the weighing procedure.

Ultimately there is only one concept that might enable us to clearly and reasonably tell wheterh one league is better then the other. This is the case whenever peers of two leagues (strongest vs. strongest, 2nd strongest vs. second strongest etc...) play each other (sufficiently often to cancel out the randomness inherent in games) the representative of one league always prevails. If this should be the case (of course only in a thought experiment as reality does not provide us with these games and also not with "everything else constsnt"...) one can safely say which one is the better league. This concept is basically equivalent to how economists compare lotteries without making any further assumptions on preferences than the basic rationality axioms. Some lotteries you can uniquely rank by this criterion (first order stochastic dominance) others you can't. Same goes for sports leagues as the criteria are arbitrary. Still a very interesting topic just by the aspects of logic and mathematics that are involved.
And I am sure it will pop up again once in a while.

At last one more thought from my side to the specific qquestion. I would always give more emphasis to hard facts (game results, player movements etc...) than to expert's assessments in making a judgement because expert statements (in the media) can very likely have a strategic background and thus might well be not sincere. But maybe that is just an economists paranoia ;-).

And before I forget it, forget about Serie A it can neither cope with EBEL nor with ALL...

[Karsten]

At last one more thought from my side to the specific qquestion. I would always give more emphasis to hard facts (game results, player movements etc...) than to expert's assessments in making a judgement because expert statements (in the media) can very likely have a strategic background and thus might well be not sincere. But maybe that is just an economists paranoia ;-).

Well, smart economist do combine quantitative data with qualitative data. From the first post in this thread it should be pretty evident that not all the experts have vested interests in Norwegian hockey. As a matter of fact, I was cautious about this when I wrote the piece.

GetL vs. Allsv: 6 games played (2:4 wins, aggregate goal score 20:25)

League vs. league score is 0.49 indicating a slight superiority of Allsvenskan.

The official list only records 5 games of which one Swedish club, Bofors, was involved in 4 games. That alone weakens the result. The result is furthermore not significant. Calculate a confidence interval around the point estimate and you will find that superiority of Get-ligaen is equally possible.
You may add more observations by adding the results from the pre-season of 2006-07 (questionable whether you can go further back in time than that unless you use discount factors). In 2006-07, Get-ligaen played Allsvenskan three times:

AIK Stockholm - Storhamar 2-4
Lillehammer - Nyköping 4-3
Storhamar - Nyköping 4-2

Nyköping and AIK were positioned in the middle of Allsvenskan in 2006. Adding these results, reinforces the conclusion that Get-ligaen is as good as Allsvenskan (if not better).

Of course this does not fit in the picture of the comparisons with Allsvenskan as according to that one ALL and GetL should be very even. This might be a consequence of the lower number of observations involving the Norwegian teams resp. a consequence of the only ordinal nature of how I weigh the game scores based on the intra league position of the teams involved (I just split into top middle bottom assigning a coefficient of 0.9/1.1 to the score of a game between teams from neighboring classes and 0.74/1.25 for top vs. bottom games and of course a coeff of 1 for peer games). If a league however is very unbalanced this can result in a bias (in this case GetL being favored in the comparison with Allsv compared to ALL teams because their Allsv opponents are weaker than the ALL opponents from Allsv even beyond the corrective influence of the weighing procedure.

This is indeed the case. The Danish teams only play the upper half teams of Allsvenskan, and primarily the very top teams of Allsvenskan, i.e. teams that are Elitserien contenders (Leksand, Malmö and Rögle) This is an important and interesting observation neglected in your approach. because it fails to take qualitative considerations into account. The reason why the Danish teams primarly play the very top teams of Allsvenskan is explained by the fact that these are the Swedish teams the Danish teams want to measure against. The Danish teams are not quite there yet. Allthough the games are usually pretty tight, the Danish teams usually lose to the top 3 of Allsvenskan (as I have repeatedly stated). I fact, Leksand, Malmö and Rögle account for 11 of the 13 Swedish victories in your record. Växjö account for the rest of the victories (against Frederikshavn from the lower half of the Danish league).

Another thing that could lead to obscure results in your approach is the fact that Allsvenskan does not have as much depth as the Danish and Norwegian league. There is really not much movement in the positions in Allsvenskan from year to year.

This is not meant as a harsh criticism of your approach. No approach is perfect, but you do need to take a qualitative look at your data before you draw conclusions. With the right adjustments, you can easily reach the conclusions that the Danish league is better than top 4-8 in Allsvenskan and thus better than Allsvenskan in general; that Get-ligaen is more or less on par with Allsvenskan as a whole, and that the Danish league is better than Get-ligaen. Consistent conclusions which have much validity both from a quantitative and qualitative point of view.

[RexKramer]

Well, smart economist do combine quantitative data with qualitative data. From the first post in this thread it should be pretty evident that not all the experts have vested interests in Norwegian hockey. As a matter of fact, I was cautious about this when I wrote the piece.

And very, very smart economists use such data only when t has been elicited in an incentive compatible way . and no doubt about you being careful it ws just a basic statement.

The official list only records 5 games of which one Swedish club, Bofors, was involved in 4 games. That alone weakens the result. The result is furthermore not significant. Calculate a confidence interval around the point estimate and you will find that superiority of Get-ligaen is equally possible.
You may add more observations by adding the results from the pre-season of 2006-07 (questionable whether you can go further back in time than that unless you use discount factors). In 2006-07, Get-ligaen played Allsvenskan three times:

AIK Stockholm - Storhamar 2-4
Lillehammer - Nyköping 4-3
Storhamar - Nyköping 4-2

Nyköping and AIK were positioned in the middle of Allsvenskan in 2006. Adding these results, reinforces the conclusion that Get-ligaen is as good as Allsvenskan (if not better).

Clearly the lack of observations is the major problem in every quantitative approach. Having a high concentration of teams on any side of the comparison is a burden as well, you are right. I am well aware of this problem but there's just nothing we can do. Limited by data...

And I have no problem with your criticism it is a well founded one in general. But significance and confidence intervalls is by now something much ahead of what I have been doing which is nothing more than producing a kind of descriptive statistic for the data (which does the right things I am inclined to say).

Where I want to head to is developing a theory that formalizes the stochastic process producing the game results. What is really needed to get somewhere with all this reasoning is characterizing a function that maps from the unknown and unobservable true "playing strengths values" (assuming such exist, are constant over time and that the implicit assumption of transitivity brought by this construction makes sense - which is problematic I'd say) to a (discrete) probability distribution over game scores...
Then one could infer with parametrical methods from the observations to playing strength which is what this is all about.
My problem is that my background is game theory and microeconomics (so I think I can cope with buildinga model that describes the process) and I don't know too much about statistics. But I am really thinking about turning this into a research project.