Tuesday, June 23, 2009

Is International Cricket Divided Into The Haves And Have-Nots?

There has been some interesting discussion prompted by my post-World Cup post. In both comments and an email exchange with a friend, there was issue taken with the following passage:
Without Pakistan, the world of cricket was on the verge of a becoming a super exclusive super-club of Australia, India, England and South Africa -- who play each other pretty much twice as much as they do against everyone else. This was a victory for the underdogs, the acned and pimpled kids who never get invited to the cool-kids parties, the ones who are socially awkward and can never get the hot girls (even those that claim to like the eccentric types). This was a victory for Sri Lanka and New Zealand too, ironic since we knocked them both out, because they are in the same position we are: an afterthought in the increasingly exclusivist cricketing hierarchy.

Before you read the rest of this post, it might be instructive to go and read a couple of the critiques made of that point in the comments. Go on, I'll wait.

Using that discussion as a background, I want to answer three questions. First, is it true that international cricket is divided into an upper class and a lower class, where the upper class plays other members of the upper class a lot more than they do members of the lower class? Second, if it is true, then why is it true? Third, if it is true, then are the reasons for it being true fair, just and smart? Let's deal with each in step, in separate blog posts. Today's question is:

Is there an elite in cricket?

Let me be more concrete with what I mean by this question. It's obviously not the case that Australia, England, India and South Africa don't play anyone else in international cricket. The question to consider is: is there a noticeable bias in scheduling that leads those four teams to play each other significantly more than proportionately? Remember, perfect proportionality would dictate that every team plays every other team equally, and since there are 10 international test sides (I'm including Zimbabwe for simplicity's sake), it would mean that we should ideally expect every team to play every other team 11.11% of the time. Obviously, this won't be true to the decimal point, but how far, if it all, do the actual data diverge from the null hypothesis?

To answer that question, I used Cricinfo's statsguru to collect data on all test cricket played from January 1, 2000 to the present. I also collapsed Bangladesh and Zimbabwe into one category, so for the purposes of this analysis, BanZim is one team. In my view, this corrects for the fact that both teams really do represent the sisters of the poor, so if we combine the two, we get something approaching a "real" test team in terms of scheduling. Of course, I was then forced to assume that Zimbabwe and Bangaldesh don't play against each other, but this is no big deal, because I am not really interested in them in the first place.

Here is what I found:

You can click on the picture for a larger view. The way to read the table is from left to right. So looking at Australia's row first, you see that since January 1, 2000 they played 6 tests against BanZim, 20 against England, 20 against India and so on.

Immediately, a problem with the hypothesis manifests itself, that is, the so-called big four (Aus, Eng, Ind, SA) simply play more cricket than everyone else. This might prove to be problematic for the hypothesis. Recall that I hypothesized that the big four pick and choose to play each other more than proportionately. But could it be the case that instead of playing less than proportionate cricket against the big four, it's simply a matter of the other teams playing less cricket in total? Which of these possibilities is closer to the truth? Let us investigate.

The chart below is basically drawn from the same data. There are a couple of points to make about it, but first have a quick look at it. Again, you can click on the image for a larger view.


When you break down all the teams' big four tests vs. everyone else tests, an interesting picture emerges. Basically, except for BanZim and Sri Lanka, everyone plays more cricket against the big four than the others. This is clear from the "positive difference" column, which is basically the difference in rate of playing big four teams vs. the rate of playing everyone else. Only BanZim and Sri Lanka have negative signs in front of this number. So, at first blush, this spells a little bit of trouble for the central hypothesis, because it suggests that the big four aren't the only ones treating themselves better than everyone else; everyone is treating the big four better than everyone else. We can call this the Uncle Tom/babu explanation.

I would submit, however, that the Uncle Tom/babu explanation does not disprove my hypothesis. Why? Simple: because it could be the case that even if everyone treats the big four better than everyone else, members of the big treat each other even better. This is where the last two columns of chart 2 come in.

The average level of "treating the big four better than everyone else" is 0.333; put differently, this means all international teams are will play 0.333 more tests against any one member of the big four in any given year vs. any non-member of the big four. It is against this baseline that we can truly judge levels of selectiveness. That is basically what the last column is about -- it is the quantity of "positive difference" less the quantity of "average positive difference".

When we analyze those numbers, an interesting picture emerges. Basically, Australia are enormously selective with an astounding "real difference" of 0.71. They really, really hate playing against non-big boys. At the other end of the spectrum is, predictably, BanZim with a "real difference" of -0.833. They consistently get screwed by the schedule-makers, even if you try and cheat the system by making two teams into one.

England and South Africa have negligble real differences. So they are basically average teams with respect to scheduling; they play the big boys more, yes, but no more than everyone else. Pakistan and Sri Lanka are fairly badly screwed -- they are definitely marginalized, for a variety of reasons including security (though Sri Lanka haven't really had major problems hosting cricket this decade, it should be noted).

India is fairly selective too, particularly against Sri Lanka and New Zealand (a paltry nine and seven tests respectively this decade).

The real shocker from these data is the West Indies. Their real difference is off the charts, second only to Australia. It appears that the big four should really have the West Indies in place of England, whose real difference is actually negative. I think this outlier is almost entirely due to the ridiculous amount of cricket England and the Windies have played against each other this decade -- they have played 24 tests between themselves, which is more than any other dyad (even more than immensely popular ones like Aus-Eng and Aus-Ind). So if we were to replace England as a member of the big four with the Windies, I would be willing to bet their real difference drops close to 0, because the only thing bringing up their "selective" rating is their schedule against England. Conversely, England's selective rating would shoot up, because they only thing keeping them out of the upper class is their fixture list against the Windies (I would run the data on this but I'm tired of Excel -- one of you is welcome to try and it let me know in the comments).

So given these data, we can reach the following tentative conclusions:

1. There is an elite in test cricket, but it's not the big four, it's the big two: Australia and India. These two teams really, really like scheduling cricket against other marketable teams and really, really hate scheduling cricket against the non-marketable teams.

2. Of the better teams in international cricket (i.e. everyone but BanZim), Pakistan and Sri Lanka are the teams most badly screwed. The reasons for this will be dealt with in another post, but let me just say this right now: it's not just security. Not in my opinion, anyway.

3. There is a broad middle class, consisting of England (who, on the basis of these numbers, have been excellent about keeping commitments against less marketable teams), New Zealand, West Indies, and South Africa.

Thoughts? Have I gone about this completely the wrong way? Do you find the data convincing? Could there better crude measures of this than the ones I have employed?

8 comments:

Aditya said...

I am in office and therefore pretending to work :), so I have not gone through your data in detail. However, few observations
1) Pakistan is a totally marketable team as far as India is concerned. An India-Pak series will always be a sell out.
2) The problem with playing with Pakistan (or selecting Pak players for IPL) is never Marketability. It has to do with security. You never know when something might trigger off something and the tour may have to be canceled. IPL is pure business and selecting Pak players does not make sense (atleast for the last IPL) because either India's foreign policy with Pak is not consistent or some security issue.
3) Oz cannot or does not play Pak because Players refuse to tour Pak (for whatever reason). However, why Oz does not host Pak as often as they do India does lend weight to your theory of a big boys club.

I will collate my thoughts later on this.

And congrats to you all for the WC win. You guys deserved it.

bubs said...

Ahsan: Some thoughts

1)It might have been more instructive if you had listed how many series the teams have played against each other rather than the number of matches. Until only a couple of years ago, England really loved hosting five-match series', even against the Windies. Breaking up their summer into two different series is a relatively new phenomenon.

2)The sub-continental teams love their two and three match contests. I recall England complaining that they were only being given three Tests and forced to play seven ODIs when they toured India. India only seem to prefer the four and five Test series after they beat Australia at home. Had you done this analysis in the mid-90s, India would have been closer to Pakistan and Sri Lanka, and it would have had to do as much with their preference for ODIs than their not being part of the elite.

3) A lot of this comes down to crowds. England and Australia can still guarantee large crowds for Tests. Test cricket attendance in Pakistan was destroyed the mind-numbing five-test series we had against India in the 80s and never recovered.

4) I would much rather see the best teams play against each other more frequently. Australia, South Africa and India should never have less than five match series. England is the only sucky team playing way more matches than they should, but they get a pass for being the only country willing to keep longer tours alive.

5) If you did a similar analysis for ODIs, you would find that India and Sri Lanka and Pakistan (before their isolation) make up the elite. Australia would have made the list a few years ago, until they scrapped their insanely long tri-nation tournament (is there really any point to teams playing each other four times and then having a best-of-three final?). This suggests to me that the teams from the subcontinent had actively decided that ODIs are more important than Tests, with India changing its mind after they became a good Test team.

Rohith said...

Ahsan

With Regards to numbers of 2nd chart :
1) Indian difference is +ve cos of emergence of rivalry with Aus and the Eng factor (note that Eng has substantial indian diaspora), but still 0.2 is really not much.
2)if WI is brot in, the data still would not differ much for Ind as Ind always plays 4-5 tests in WI, and that is also where ind has played better progressively in the last decade culminating in a series victory in the last one.
3) If India Pak data is sought over 3 year periods from 2000 to 2009, i am sure you would find high density in the period that there was a thaw in relations, where we played each other almost every six months(Dosti series etc). So that puts to rest the opinion that India discriminated in superficial ways.
4) Aus is biggest surprise as they were the best team in the decade and really had no one to look upto; but then you may give them the consideration that being the best in the world, they would have liked to played the best contenders(to attract the best crowds); That is probably a part of the bargain when you are the best. To conclude, they maybe considered to be the only elite as they had complete discretion of who they wanted to play against, but that would be harsh in my opinion.
5) Also, when you talk about the elites as a group, they are not necessarily united in that sense cos they have had face-offs in many issues (Hair controversy (Pak as well as Murali), Bhajji controversy). Infact Aisan boards have quite a clout in that way.

Bubs
Points 1 2 and 4 I agree and alluded to as much in the comments of other post. WRT data used for analysis (total number of matches), it is at best a symptom for the premise of elitism, but for reasons of the same, more data mining might be needed, one aspect of which i had touched upon in point 5 of the comment in other post.
Point 3 suggests business aspect of it all which is not necessarily a "discriminatory class attribute". After all, har board ko apna choolah jalaye rakhna hai.
Point 5 is very interesting and another example of that mite well be India's thrust and focus on IPL after establishment of credentials in T20WC 2007 win. Again this aspect will continously evolve and will yield different results over different tracts of time.

PS :- WRT India's focus on IPL, it is beginning to get worrisome with talks of 2 IPLs in a year making the rounds....that would be something of an overkill :((

Ahsan said...

Bubs:

1. Actually, I think precisely the opposite is true. I think counting number of series seriously masks the bias/selectivity factor. Think of how many bullshit 2-test "series" Pakistan has played in the last six or seven years. The same is true for other "small" teams. Eng, Aus, and Ind regularly play 4-5 test series. That's where the main difference comes up for me.

2. Many, many of India series are now four or more. And they never play fewer than three in a series now. Small teams do so regularly.

3. Yes, this is true. The lack of crowds for test cricket in Pakistan, amongst other places, is really disappointing. This is why I love watching cricket on TV when it's taking place in places like England and Australia, because they actually give a shit over there.

4. This is a problematic argument in my view, and I will deal with it properly in a future post.

5. Yes, the privileging of ODIs is quite pissing off. I'm hoping that with the rise of T20, ODIs die off and more tests are scheduled as a result.

Rohith:

1. It's all relative. 0.2 is a big difference when the only other sides above it are OZ and the Windies (which is an outlier as I explained).

2. Yes, but it would be different for England, which is what my point was.

3. India absolutely discriminates. They've played a combined 16 tests against SL and NZ in the same time frame, which is LESS than their number of tests against OZ and just about the same as vs. Eng. And correct me if I'm wrong, but I think India remain the only team in world cricket to not have hosted Bangladesh for a test yet.

4. This is a problematic argument (similar to Bubs' point number 4) which I will treat in a future post.

5. Yeah I agree they're not united by any means (don't think anyone will claim the OZ and Ind teams and boards like each other). But it's still damaging to intl cricket in my view.

Rohith said...

Ahsan,

Yes, SL and NZ point is striking.

I am not serving this as an excuse but NZ really are not a force in cricket today. They even find it difficult to pay off their cricketers which has caused lot many to go away from the game or switch to ICL. Rugby is a much preferred sport than cricket. I am saying all this only to point to the fact that the other country's board also has to have many things going for itself. And there are no "stars" coming out of there. Which makes it a vicious circle. But others countries do play them, yes. (Note:- Aus NZ rivalry has kept their bilaterals going i think)

I think for Ban and SL, the factor at play is the familiarity of the conditions. It really doesnt make for any "challenging" cricket(for either country) in these permutations. India tours Ban cos crowds there will be pulled cos of India but reverse will not happen. For SL, I think the attitude of BOTH boards towards each other needs to be known. To be sure, Pak also has similar playing conditions as India, but you know how it is with IndPak :)

I think with each comment I am convinced that this is more a case of business proposition, which I dont see as being too wrong as long as there is a bare minimum that is being done to avoid complete neglect of some countries. Which is why the ICC code for 2 bilateral tours in 4 yrs. It needs to be enforced ofcourse.

The second thing coming across is that the chances of a country not being neglected is dependent on 2 things 1) latent demand for game 2) quality of players.

In continuation to the note on NZ, the case of SA is particularly interesting.For SA it has turned out to be a happy coincidence that the great talent buildup in 90s resulted in a fantastic team this decade (consistently no 2 i believe, until India came along), which has further translated into it being a tournament organising hub(2003 WC, 2006 champs trophy,2007 t20, 2009 IPL) which again has had the domino effect on the games popularity. And see the result; at first sight, you did consider them to be among the elite :). They have been able to generate that kind of attention.

And that is why i think its all the more imp that Pak atleast now agree to host tournaments at neutral venues. Keep the talent, game and demand alive and strive to be in the conscience of the world.

Indophile said...

Ahsan
The real problem is of Australian reluctance to play in Pakistan. I think I may be wrong but Pakistan has visited Australia two times since 1999 and they are scheduled to visit them this year. Meanwhile India has visited Australia three time in this duration so that's same amount of test series in a decade against Australia in Australia for both India and Pakistan. But the problem arises when Australia refuses to play Pakistan in Pakistan.

With regards to India-Pakistan I think both of them has played insane amount of cricket against each other from 2004 to 2007.
After this period there was a tour scheduled at the end of 2008 which was cancelled after Mumbai attacks.So I am not really sure about India or Australia not playing against Pakistan if you discard the security factor which bothers Aussies.

Rohith said...

Also, another factor governing preferences is the ICC ratings system.

The no of points given to a team for a series win is more if the result of the series is different from what it was the last time around at the same location(victory/defeat wise as well as final tally, 4-0 or 3-1 or 2-1), in addition to the dependence on relative rating of the opposition. So if we have Ind vs Aus @ Ind, Aus will be keenly interested in the tour so they can improve their position in the ratings(assuming they lost the prev one), reverse also being true. In such a case, the only hope for teams down the order is to have great home records and try at the same time to get some breakthru on the AWAY front. If u observe, that is precisely what has happened with India. Home was always strong, away has been a revelation(competitive series with aus eng wi and wins as well)

Rohith said...

I seem to have made a mistake in the criteria, there isnt anything wrt last series' result as per below link (though I had heard abt it on some show in ESPN), but relative rating of opposition is surely one.
http://en.wikipedia.org/wiki/ICC_ODI_Championship