ICC team ratings relevance

[Vijay.Sharma]

I thought this should be a separate thread from the players' ratings thread.

Here's the FAQ for it -
http://static.icc-cricket.yahoo.net...1FDC31284D768163CA1CBCD_1295335858633_724.pdf

 

[Spark]

They're pretty good I feel. Reasonably accurate.

 

[Howe_zat]

Problems with the team rankings:

1) Doesn't take into account whether victories are home or away.

2) Doesn't take into account the margin of victories.

3) It uses the same time period for every team, which leads to unfairness as the "current" international sides have not been together for the same length of time. For example, Pakistan have changed their opening pair, two middle order batsmen, captain, keeper and opening bowlers within the last year; while India have made hardly any permanent changes to their side from three years ago.

4) The ODI rankings suffer from all of the above while also regarding near-pointless matches and dead rubbers, when the best players are often rested and debutants are tried out, with the same importance as World Cup knockouts.

I think the rankings are mostly accurate as they are based on results, which is obviously the most important thing. But when two sides are ranking closely, within a few points of each other, you can usually use a combination of these points to argue against them.

 

[Spark]

ODI rankings are pretty dodgy yeah. Test rankings - or at least the Test point scores are a good general guide though. You should also mention that how often a team plays can have a big effect, though.

 

[Vijay.Sharma]

Problems with the team rankings:

2) Doesn't take into account the margin of victories.

I think it does. Check out the answer to FAQ 1, second para.

3) It uses the same time period for every team, which leads to unfairness as the "current" international sides have not been together for the same length of time. For example, Pakistan have changed their opening pair, two middle order batsmen, captain, keeper and opening bowlers within the last year; while India have made hardly any permanent changes to their side from three years ago.

Here also it does. The rating points are based on 3-4 years with greater weightage given to the more recent ones.

I think the rankings are mostly accurate as they are based on results, which is obviously the most important thing. But when two sides are ranking closely, within a few points of each other, you can usually use a combination of these points to argue against them.

It takes care of this too. The rating points are calculated relatively that is the opposition's rating points are part of the algorithm used to calculate your team's rating points. Of course the farther the two teams are the more apparent it is. I don't see it in the FAQ but they have actually published the exact formula with an example of how the rating points are calculated - maybe cricinfo or icc-cricket.net

 

[Howe_zat]

I think it does. Check out the answer to FAQ 1, second para.

It doesn't. I'm not talking about the series score, but the margin of victory in matches.

Here also it does. The rating points are based on 3-4 years with greater weightage given to the more recent ones.

I know about the weightings. My point is that it treats each team equally with said weightings. Obviously that's the only fair way to do it, but it doesn't give an accurate portrayal of what's happened. When ranking the current teams, you should give more weight to performances by the players who are still in the team and less to those who aren't. There's no way to do this mathematically.

It takes care of this too. The rating points are calculated relatively that is the opposition's rating points are part of the algorithm used to calculate your team's rating points. Of course the farther the two teams are the more apparent it is. I don't see it in the FAQ but they have actually published the exact formula with an example of how the rating points are calculated - maybe cricinfo or icc-cricket.net

That wasn't a separate point. I was saying that you can use my previous points to argue against a few points' difference between teams.

 

[marc71178]

I think it does. Check out the answer to FAQ 1, second para.

That's series score, not match result though.

 

[Magrat Garlick]

Problems with the team rankings:

1) Doesn't take into account whether victories are home or away.

2) Doesn't take into account the margin of victories.

3) It uses the same time period for every team, which leads to unfairness as the "current" international sides have not been together for the same length of time. For example, Pakistan have changed their opening pair, two middle order batsmen, captain, keeper and opening bowlers within the last year; while India have made hardly any permanent changes to their side from three years ago.

4) The ODI rankings suffer from all of the above while also regarding near-pointless matches and dead rubbers, when the best players are often rested and debutants are tried out, with the same importance as World Cup knockouts.

I think the rankings are mostly accurate as they are based on results, which is obviously the most important thing. But when two sides are ranking closely, within a few points of each other, you can usually use a combination of these points to argue against them.

5) Matches change arbitrary value on 1 August, instead of a sliding decrease through the year.

 

[OMM!]

Does any ranking/league system in any sport include points for how good a win it was?!

Man Utd don't get 6 points for winning 6-0. Federer doesn't get more points for winning 6-0, 6-0, 6-0 in a Grand Slam. It's all based on just winning.

 

[cricpk]

I do not like this Ranking System as teams do not get to play equal number of games.

 

[marc71178]

That's why the points are averaged over matches played.

 

[zaremba]

Problems with the team rankings:

1) Doesn't take into account whether victories are home or away.

Why should that be taken into account?

If you give a greater weighting to away victories, then a team which always wins away but loses at home will rank higher than a team which always wins at home and loses away. Seems to me that both teams have equal merit and equal (but opposite) weaknesses. Winning away from home is great, but it doesn't outweigh losing at home.

Edit:

I appreciate of course that, all other things being equal, winning away will be a greater achievement than winning at home. In that sense you might be thought to "earn" more points by winning away. But the purpose of ranking points is not to compare the merit of the various victories secured by a particular team; it's to compare the merits of different teams as against each other (and over a fairly long period). And on the premise that all teams play an approximately equal number of home and away matches then it will all equal out and, for the reason I've given above, it's right that home and away victories count the same.

The flaw in my theory is that the premise that all teams play an equal number of home and away matches is not, currently, accurate, because Pakistan play no proper home matches. If they achieved an exactly equivalent record as, say, Sri Lanka, then Pakistan are indeed arguably the better team because they have not had the benefit of home advantage. I'm not sure that this anomalous and (hopefully) temporary consideration should lead to the whole system being changed, though.

 

[PhoenixFire]

Does any ranking/league system in any sport include points for how good a win it was?!

Man Utd don't get 6 points for winning 6-0. Federer doesn't get more points for winning 6-0, 6-0, 6-0 in a Grand Slam. It's all based on just winning.

Fair point, I don't see a good reason as to why it shouldn't be taken into account. Seems like it would add a lot more depth to the rankings.

 

[Howe_zat]

Why should that be taken into account?

If you give a greater weighting to away victories, then a team which always wins away but loses at home will rank higher than a team which always wins at home and loses away. Seems to me that both teams have equal merit and equal (but opposite) weaknesses. Winning away from home is great, but it doesn't outweigh losing at home.

Edit:

I appreciate of course that, all other things being equal, winning away will be a greater achievement than winning at home. In that sense you might be thought to "earn" more points by winning away. But the purpose of ranking points is not to compare the merit of the various victories secured by a particular team; it's to compare the merits of different teams as against each other (and over a fairly long period). And on the premise that all teams play an approximately equal number of home and away matches then it will all equal out and, for the reason I've given above, it's right that home and away victories count the same.

The flaw in my theory is that the premise that all teams play an equal number of home and away matches is not, currently, accurate, because Pakistan play no proper home matches. If they achieved an exactly equivalent record as, say, Sri Lanka, then Pakistan are indeed arguably the better team because they have not had the benefit of home advantage. I'm not sure that this anomalous and (hopefully) temporary consideration should lead to the whole system being changed, though.

That's an excellent post, and yes, my thinking is that it's the sort of thing that ought to be taken into account because there is a discrepancy between the number of home matches per side. It's not just Pakistan either, as England and Australia both play a lot more home matches than away.

 

[Howe_zat]

Does any ranking/league system in any sport include points for how good a win it was?!

Man Utd don't get 6 points for winning 6-0. Federer doesn't get more points for winning 6-0, 6-0, 6-0 in a Grand Slam. It's all based on just winning.

The rankings system is not a tournament, it's a way of judging which team is the best.

 

[Vijay.Sharma]

It doesn't. I'm not talking about the series score, but the margin of victory in matches.

OK maybe I misunderstood the FAQ.

I know about the weightings. My point is that it treats each team equally with said weightings. Obviously that's the only fair way to do it, but it doesn't give an accurate portrayal of what's happened. When ranking the current teams, you should give more weight to performances by the players who are still in the team and less to those who aren't. There's no way to do this mathematically.

Since the rating points at any given point in team is based on the relative strength of the opposition (as measured by their rating points), I think this item is implicitly already addressed. If you calculate a ratings point of India v/s SL in 2008 when SL were say 115 points and India were 118. Now in 2011 even though SL is 100 and India is 115, the inlfuence of the series of 2008 will still be as per 118 v/s 115 (of course it will be adjusted for its recent-ness).

 

[Vijay.Sharma]

5) Matches change arbitrary value on 1 August, instead of a sliding decrease through the year.

Is that so...I haven't seen their algorithm to be sure of that. But I guess they will never share it

 

[Vijay.Sharma]

I do not like this Ranking System as teams do not get to play equal number of games.

Like Marc sez, number of games has no influence whatsoever on the ratings points.

 

[Vijay.Sharma]

The rankings system is not a tournament, it's a way of judging which team is the best.

Yeah, but then that criteria actually can backfire badly. Consider the following case -
Aus v/s Zimbabwe - Zimbabwe bat first and hence have slightly easier conditions and the score at the end of the game would be Zim 368/ao, Aus 525/ao, Zim 214/ao, Aus 58/1

SL v/s Ban - SL bat first and the scores are SL 685/3 decl, Ban 314/ao, Ban 167/ao

How do you reconcile such events?

Also in many Tests...in fact in more Tests than we may like to believe...it boils down to who got the better of the conditions. So while a team may bat in perfect overcast conditions and lose 6 wickets, suddenly the skies become clear and they muster along to a respectable total. But the second team now has the advantage of the conditions until the next morning and can make merry. Of course it is part and parcel of test cricket but still the magnitude of victory is very tricky to include in a formula.

It becomes even more difficult in case of rain affected games and draws.

 

[cricpk]

Like Marc sez, number of games has no influence whatsoever on the ratings points.

It does. Imagine India played 4 tests against England. Srilanka Played 3 of them. Now if India had also played 3 they would have been number 2 at the moment. I mean it matters in longer run, Specially when you get more points beating a higher ranked side.