Analysing Associate Proficiency
Russell Degnan

On CricketEurope, Roy Morgan put forward an interesting index of proficiency for measuring competitive last week. It is not without faults, as it applies averages to games where not all sides will be bowled out, and therefore probably exaggerates uncompetitiveness, but it will serve for the purposes of illustration. What I aim to do here is highlight a few reasons why I think your analysis does the associates a disservice.

Firstly, not much can be read into the results of teams that play two games. Hence, the marked weak associate results in '75, '79 and 07 are not of much more import than India's results in '75, '79 or '07 (49, 24 and 46 respectively).

Similarly, the successes that led to full member status can be over-stated. Sri Lanka followed up with 61, 49, 62 to 1992, and 53 in 1999; Zimbabwe 40, 76, 48, 50 and 30; Bangladesh 28, 49 and 42. To the extend that those teams were competitive after becoming full members the benchmark is closer to 50 than 65.

What is needed therefore is a benchmark of associate results against those achieved by test members to assess their level of "competitiveness". We can begin that by looking at what the full members have done at World Cups:

Prof. Index1975197919831987199219961999200320072011
Australia 101.3246.288.06128.3880.88131.75143.81226.28310.96127.96
Bangladesh28.1949.8842.79
England 159.82109.34155.71120.7134.4860.32103.15108.3975.987.79
India 49.124.4111.18118.4190.01124.48122.39119.1146.37114.85
New Zealand 71.14158.9284.5853.2130.6466.9175.6695.21100.55108.49
Pakistan 132.697.192.33129.2799.19135.75110.4556.73149.8151.77
South Africa126.17118.02114.72123.994.83146.26
Sri Lanka 61.0549.0162.91150.1653.88105.69138.92183.75
West Indies 185.07203.83125.14107.8585.6410290.6490.4472.4971.66
Zimbabwe 40.1975.5647.9850.7829.64

Mathematically, a proficiency index of 50 means losing a test match by an innings, but statistically, the most common test match margin is 8 wickets or a proficiency index of 60. Anything around 60 indicates that a team is losing most often, but only by an average amount (or interspersing close games with thrashings). While by and large the test teams have PIs around 100, there are plenty of instances where they've been worse than the associates.

But that kind of measure fails to get at the real problem: that "competitive" is a relative term. Three world cups have been won by teams with proficiency indexes above 200 meaning the other competing full members were only "competitive" at levels between 32 (in 2007) and 49 (in 1979). In those three cups, it is fair to say there was only one competitive team, but it would be a pointless world cup if they were the only ones invited.

If we take a historical overview, there is a symmetry between teams at different levels. The bottom three full members have had an average PI of 62 (but just 47 in the past three world cups since Bangladesh were admitted); the top three have had an average PI against them of 69 (but just 56 in the past 3 world cups because Australia was dominant for much of that period). The associates average competitiveness is only 44, but if we divide that against the bottom three members (to see how competitive they are with the level above) then the average is 73 (and 93 in the past three world cups).

What we think about competitiveness depends on our expectations. If we consider a team competitive if it can win the world cup then less than a handful of full members are ever competitive at one time; if we consider a team competitive if it conforms to an average then it depends who is in, and who out: Bangladesh and Zimbabwe would certainly be excluded, but without those two whipping-boys, so might New Zealand and the West Indies. A few poor games aside, the associate teams are as competitive as we ought to expect against the teams you'd expect them to be competitive with. If they were more competitive they'd be equal or superior which defies logic - the weaker teams ought to be weaker.

Very few teams (including several full members) will ever by consistently competitive with the strongest nations. The format of world cups, and test cricket needs to reflect that reality. At the moment it reflects a fantasy wondefully satirised at Turn and Bounce:


ALL TEAMS ARE EQUAL
BUT SOME TEAM ARE MORE EQUAL THAN OTHERS

No team is truly equal, and any system that supposes so is going to end up playing a lot of uncompetitive and ultimately meaningless games.

Idle Summers 31st March, 2011 12:30:32   [#] [0 comments] 

Random notes on the World Cup
Russell Degnan

D-Day Mark Two

Group B may well be decided today, with games between Ireland and the West Indies and England and Bangladesh. If the two big-name teams win, only an Ireland vs South Africa upset and some luck can keep the group interesting. Readers harking back to D-day mark 1, tipped to be the two deciding gams in each group between New Zealand and Zimbabwe, and West Indies and Bangladesh, will note the possibility of this being another shocker.

The current qualification odds:
RSA 96%, Ind 100%, Eng 94%, WI 66% Ban 18% Ire 26% Ned 0%

Zimbabwe's losing streak has effectively ended any interest in Group A. The remaining possibilities for an upset involve one of three scenarios:

  • Australia losing to Pakistan, Kenya and Canada (estimated probability: 1 in 30,000)
  • Pakistan losing to Zimbabwe and Australia, Zimbabwe beating Kenya, which is not out of the question
  • New Zealand being mauled by Canada, losing to Sri Lanka, and Canada beating Australia.

Aus 100%, Sri 100% Pak 95% NZ 99% Zim 6% Can 0% Ken 0%


Post-match update

Both good matches, but not quite the best result with Ireland losing. They only have themselves to blame, and I've not been impressed at any point in this tournament with Porterfield's use of his bowlers. I like their cricket, but they lack the edge and lost two winnable games against Bangladesh and the West Indies. That's ought to be enough to send a side home (unless you are England).

Changed probabilities don't take into account NRR, so realistically, the West Indies are 95%+, Bangladesh 22%, and Ireland 8%. England face a must-win against the West Indies to progress - though how a team that has won 2 of 5 games ought to progress is a question best left to the organisers.

RSA 95%, Ind 100%, Eng 78%, WI 75% Ban 39% Ire 13% Ned 0%


Mismatches, but not always mismatched teams

Scyld Berry noted the other day that most of the games have been very one-sided; what he didn't mention (apart from wanting a calculator for his birthday), was that of the six games where more than 5 wickets were lost, or the margin was less than 80 runs, half have involved the pilloried associates: Bangladesh vs Ireland, England vs Ireland, Pakistan vs Canada; and of the chases that went into the last 5 overs: Kenya vs Canada, England vs Netherlands, and India vs Ireland. Roughly the same proportion of good games as there have been games involving associates. Something that was also true four years ago.


Repeated myths

Two myths persist about the format, amongst the constant stir for something better: that the tournament is shortened by removing teams; and that the number of mismatches increases as more teams are added. I've covered the first already; the second is counter-intuitive, but simply explained. If the number of matches is held constant, then the number of games each full member plays against each team increases as the groups expand. If our concern is matches between the big-8 and the rest, then the number of those games at different sized world cups is 16 for 10, 12 and 16 team world cups, 20 for a 24 team world cup, and 24 for a 14, 20 or 32 team edition. The quality of the associates reduces somewhat as you expand, notwithstanding that most teams ranked around 16-25 are pretty evenly matched, but any qualification tournament risks excluding quite good teams as well (Afghanistan and Scotland). It is almost always better to have too many than too few.

It is a measure of the insanity of the 14 team format that the ICC managed to increase the number of scheduled mismatches, reduce the value of individual wins, and make the world cup more exclusive. The real reason for reducing the number of teams is to have certain teams play more games. A glorified league, but not a world cup.


Just when you think a ODI captain couldn't be more defensive

Many commentators have noted the tendency to use spinners in the early overs, but few have picked up on the reason they are proving so valuable. With only two fielders out in the first 10 overs, a pace bowler needs to deploy a fine-leg and third-man, or face a barrage of boundaries from flicks, glances and scoops. A spinner gives up at most 2 runs from late-cuts and paddles, allowing the fielders to be deployed deep in front of square.

In other words, fielding restrictions are best negated by ignoring the attacking potential of the new ball; a good reason to remove them as well.


Finally...

There is still another nine days of group games to come.

Idle Summers 11th March, 2011 07:51:29   [#] [5 comments]