The old ICC Test Championship was a horror. I am surprised that the ICC took so long to get rid of it.
The new ICC rating scheme for Test matches is better. First, it takes into account the result of every Test match in a series (so a 5-0 win in a five-Test series is now considered better than a 3-2 win). Second, it rewards the winner of the series (the series winner is deemed to have won an additional Test; a 5-0 win is therefore like a 6-0 win, a 3-2 win is like a 4-2 win). Third, the scheme recognizes the strength of the opposition (so beating Australia is more rewarding than beating Zimbabwe). Fourth, the rating reflects current form better (because it is based on results never more than four years old with older results receiving a relatively lower weight).
The rating scheme does not, however, take into account the differences between 'home' and 'away' results. This is a major weakness. Every team -- including Australia -- plays better at 'home' than 'away'; some teams -- notably West Indies and Sri Lanka -- are practically incapable of winning 'away' matches, and rather formidable when they play at home. It seems very surprising that ICC should ignore this factor; one suspects that ICC's expert simply couldn't fit this factor into his equation!
The scheme is a natural extension of the ICC ODI formula -- which makes sense, because both the ICC formulae are now derived using the same set of rules. Both schemes are based on past results with the rather curious cut-off date of 1 August every year. While the ICC ODI scheme goes back to a maximum of three years, the new ICC Test scheme goes back to a maximum of four years. The ICC ODI scheme gives current year results a 'weight' of 1, the previous year a weight of 2/3 and results 2-3 years old a weight of 1/3; the ICC Test scheme gives a weight of 1 to the current and previous year results, and a weight of ½ for results older than two years.
Currently, the ICC Test ratings are as follows: Australia 129, South Africa 115, New Zealand 103, Sri Lanka 101, England 97 (will go up to 98 after defeating Zimbabwe 2-0), India 91, Pakistan 91, West Indies 79, Zimbabwe 59 (will go down to 58) and Bangladesh 4.
The new ratings therefore put Australia sufficiently ahead of South Africa (before the Australia-West Indies series started, Australia were briefly behind South Africa in the old ICC ratings!). But I still find it hard to accept that only 14 points separate Australia from the second best team. The gap should have been much wider. It isn't, because the ratings cannot spot the fact that Australia are winning so many 'away' Test series so strongly!
Sri Lanka too look better than what they are because of their series sweeps at home against West Indies and Zimbabwe (which are now reckoned as 4-0 wins each), and relatively fewer Test matches played abroad (where they lost practically everything).
How are these ratings calculated? Just like the ICC ODI ratings except for the additional reward for a series win. As an illustration, let's see how the ratings would change if India were to defeat New Zealand 1-0 in the two-match home Test series, to be played in October 2003.
Looking at the ICC table again, we would find that India's 91 is obtained by dividing 3914 by 43. The figure 3914 indicates the points scored by India since 1 August 1999 based on ICC's formula while 43 is India's 'match/series' total (this funny name is to remind you that the denominator is made up of all the matches played by India during the period plus any 'additional' matches deemed to have been won because of series wins). New Zealand's 103 is obtained by dividing 3615 by 35.
According to the ICC scheme, after winning the series 1-0, the series result will be 2.5-0.5 in favour of India: India 2.5 (1 for the win, 0.5 for the draw and 1 more for the series win) and NZ 0.5 (for the draw). If we assume that NZ are still at 103 and India at 91 (these values will actually change after the 1 August 2003 transition), then the two teams' points are calculated as discussed below.
With a series result of 2.5-0.5, India get 2.5 multiplied by 153 (50 points more than NZ's rating of 103) plus 0.5 multiplied by 53 (50 points less than NZ's rating). This adds up to 409 more points for India. A similar calculation [0.5 * (91+50) + 2.5 (91-50)] throws up 173 more points for New Zealand.
India's total points rise to 3914 + 409 = 4323 and NZ's to 3615 + 173 = 3788. India's 'match/series' total goes up by 3 from 43 to 46, and NZ's 'match/series' total also climbs by 3 to go from 35 to 38. India's new rating is therefore 4323 / 46 = 94 while's NZ's new rating is 3788 / 38 = 100. India, therefore, go from 91 to 94; while New Zealand drop from 103 to 100.
There's a slightly different formula if the gap between the competing teams' rating exceeds 40 (e.g. Australia with 129 and West Indies with 79 yielding a gap of 50), but we won't discuss that here.
All things considered, I believe that this new Test rating scheme is a sufficiently good effort and constitutes a quantum leap in ICC's wisdom. But ICC could still have come up with something better. For example, I have no doubt at all that the Rediff Test ratings provide a more realistic appraisal of current performance.
I am actually a little puzzled why our ratings didn't even qualify as a 'candidate' (they have been up on the Web for almost two years now). I suppose it isn't ICC's normal practice to invite proposals: they probably look around for the first capable Englishman and hand him the job. I see that same thing happening with the Duckworth/Lewis (D/L) method. I think it is patently unfair of ICC to deny V Jayadevan even the opportunity to present his alternative to the D/L method.