Bat to the FutureDave Wilson |
In the past, my match impact features have focussed on the impact of individual players. In this piece, I’d like to look at teams. For those who have no idea what I’m going on about, check out this feature and subsequent ones expounding the same theory.
I’ve given some thought to how best to do this – the obvious answer is to look at the total impact of the team in a given match, but then I also thought about looking at the teams whose players had the highest recent impact performances going into the match (the MA5 number, or moving average over five Tests). That surely would show the strongest sides?
Yet another way is to find the teams which had most players over a given MA5 number, say greater than 20%.
Looking at the issue using both the first and second methods, i.e. finding the teams with the highest match totals or total MA5 numbers for all players, I found that this tends to identify teams who participated in particularly close matches and series, such as the West Indies teams who tied with Australia in 1960-61 and which matched up against Pakistan in the winter of 1976-77, also England 2005. It was interesting that using that method uncovered two series in which both teams ranked highly – England and Pakistan in 2005/06 and England-Australia from 1961, but those series included matches which ebbed and flowed so that the win probability remained in flux.
Turning to those teams which included several high-impact players, it turns out there have been 12 throughout Test history to include five players who enjoyed an MA5 value of higher than 20% going into a particular match. Actually there are effectively nine, as two teams were basically the same line-up a couple of matches later.
The twelve teams, and the players with more than 20% average impact, are listed chronologically as follows:-
|Players with 20% MA5
|Sobers, Alexander, Kanhai, Hall, Solomon
|Davidson, Benaud, McKenzie, Lawry, Mackay
|Davidson, Benaud, McKenzie, Booth, Simpson
|Dexter, Titmus, Barrington, Trueman, Sheppard
|Mushtaq, Sadiq, Intikhab, Majid, Nazir
|Mushtaq, Sadiq, Intikhab, Majid, Shafiq Ahmed
|Richards, Greenidge, Garner, Croft, Roberts
|Richards, Greenidge, Garner, Croft, Fredericks
|Richards, Greenidge, Garner, Croft, Fredericks
|Ambrose, Lara, Bishop, Walsh, Murray
|Gilchrist, Warne, Hayden, Lehmann, Martyn
|Flintoff, GO Jones, Trescothick, Giles, Harmison
Much of the above can be explained by the way the moving averages are calculated i.e. on a five-Test moving average basis, such that the Australians of 1961 have some residual value from the great series against West Indies the previous winter, and England 2005/06 have a similar residual value from the Ashes. The same is partly true of West Indies 1976/77, following their great performances in the summer of 1976 in England.
As the system was really designed to determine high-impact individuals over the course of a career, it would seem that the application of it to teams for a match or series is not appropriate, especially considering it unearths no sides prior to 1961.
The above analysis did have me doubting the validity of the system, at least as applied to teams. I was expecting to see for example the 1980s West Indies teams show up as being high-impact, so I decided to do a more in-depth analysis of that all-conquering side. It was in performing that analysis that I discovered the aspect of this system which gave me the title of this feature.
First I looked at every match and series the West Indies was involved in during the decade of the 1980s. I examined the individual players’ MA5 values first and noted all of the West Indies team members who were rated at over 20% for each match.
The results were interesting, to say the least. Although the teams started out with sometimes three but usually at least two players rated above 20%, there was a period from 1984 in England through 1987 in Pakistan when not one single individual West Indian Test player ranked higher than 20%. Yet during that period, they won 18 Tests and lost only one. How come?
Surely that is an indictment of the impact system as a measure of teams (and possibly also players) when considering that none of the 1980s West Indian Test teams rated particularly highly using this system?
The key is, we must look at the total team impact when compared to that of the opposition. We can compare the total team impact rating to that of their opposition going into each Test throughout the decade, calling any ratings which are close enough as a draw.
Here’s what we get.
The aggregate results of the nineteen Test series involving the West Indies throughout the decade of the 1980s predicted by the impact system is
In comparison, the actual aggregate results of the nineteen Test series involving the West Indies throughout the decade of the 1980s is
If we consider a draw as half a win each, the win percentage is exactly the same, 0.7915.
So the system is actually excellent at rating teams, it’s just that all ratings are relative.
Also we can see that it’s probably more important from a team perspective to have eleven players who contribute at a reasonably high level rather than a few dominant individuals.
We can perform the same exercise for the 2000s Australians. Following the defeat in that classic series in India during 2000/01, Australia did not lose another series until the 2005 Ashes, during which time their actual aggregate results were
The predicted aggregate results of those sixteen Test series is
The above is pretty convincing, as regards the correlation of team impact totals to predicted team performance. But if you’re still not convinced…
The 1950s England side is certainly one of the most successful over a significant period of time – England were unbeaten for 14 consecutive Test series, most of which were comprised of five Tests. Here are the results of the same exercise for those teams.
The advantage of this method of comparison over a comparison based on the ICC rankings is that the team impact rating is an aggregate of the impact of each of the individual players in the team, whereas the ICC rankings are based on the performance of the team regardless of who was playing and don’t vary as much from match-to-match.
If you used the ICC rankings to predict results it would almost always predict a whitewash for one team or the other.
Anyway, if you’ll excuse me I’m off to place some bets.