Analysing wicketkeepers by byes conceded

Ananth has tried to rate wicketkeepers in his post below, but there are several problems with his analysis, most of them pointed out by commenters. Wicketkeeping is a fundamentally difficult part of cricket to analyse statistically – the usual records that people talk about are dismissals, but these are highly dependent on the bowlers. If Adam Gilchrist comes out as the best pure keeper, it is because he had a great bowling attack which tried to get edges (as opposed to, say, Wasim Akram and Waqar Younis, who often aimed for the pads or stumps).

Nevertheless, we can make a reasonable effort at assessing pure keeping ability by looking at bye rates. To some extent these are also dependent on the bowlers (if they spray the ball down leg a lot then there'll be more byes), but in general the results are much better.

The main stat I'll use here is byes per 600 balls. But this figure needs adjusting to be fair – wicket-keeping is easier in some countries than in others. I applied an adjustment factor by country, based on the overall bye rate of away keepers since World War II. (If you go back much further, you find that bye rates were much higher because keepers stood up to the stumps much more often to fast bowling.)

These were the bye rates per 600 balls for away keepers by country:
Australia: 3.24
Bangladesh: 3.25
England: 3.50
India: 4.96
New Zealand: 3.02
Pakistan: 4.97
South Africa: 2.60
Sri Lanka: 4.27 West Indies: 4.48 Zimbabwe: 1.19

I used as a reference point 3.5 byes per 600 balls. So, if a keeper conceded 2 byes in an innings in India, that would be adjusted to 2 * 3.5 / 4.96 = 1.14 byes.

Here are the results. I've given both the raw byes per 600 balls and the adjusted byes per 600 balls. Qualification: 20 Tests as wicket-keeper:

per 600 balls
name              m   balls   byes  byes  adj
PR Downton        30  29517   84    1.71  1.52
DJ Richardson     42  43222   143   1.99  2.22
APE Knott         95  101704  422   2.49  2.32
NS Tamhane        21  25659   137   3.20  2.37
IDS Smith         63  63672   277   2.61  2.63
Khaled Mashud     44  37099   152   2.46  2.69
SMH Kirmani       88  91761   506   3.31  2.70
KS More           49  54558   280   3.08  2.80
RW Taylor         57  59085   285   2.89  2.81
RD Jacobs         65  69122   294   2.55  2.83

Paul Downton had the good fortune to play some Tests for England in the 1980s, after the introduction of covered wickets and before the modern trend of picking keeper-batsmen. Khaled Mashud is the most surprising name to see here. This list is not perfect, but at least names like Knott and Taylor are near the top. The full list can be seen here.

It will become easier to assess wicket-keepers after some years of club-based Twenty20 cricket. With free player movement between sides, bowlers will bowl with various keepers over the course of a career. Then we'll be able to look at how many dismissals each keeper got off the same bowlers, and so dismissal counts will be a much more useful way of assessing pure keeping ability.

Note: There will be some errors in the byes tallies, because of keepers going off the field and being replaced. Also, I've given the bye rates to two decimal places, but I haven't checked how significant the second decimal place is.