Series Points - A New Way Of Ranking Test

advertisement
Series Points – A New Way of Ranking Test Cricketers
Dave Wilson, August 2008
Introduction – what’s wrong with ratings systems?
Systems such as the ICC Test Player Ratings, which is searchable for any date in Test
cricket history, is extremely useful, however it is limited in that you can either search on
batting, or bowling, but not both together. What I feel is needed is a method of
objectively measuring and comparing all players at the same time, regardless of their
main cricketing discipline, i.e. where all things are considered, batting, bowling and
fielding to give one rating which is directly comparable with everyone else.
Part of the problem is that if batting and bowling are kept separate, then we can only
compare batting with batting and bowling with bowling. If this is the case, there’s no way
of keeping both in context – whenever era-specific adjustments are made, it’s always
based on each aspect in isolation; batting variations with time are calculated based on
runs scored, without looking at what’s going on with bowling and vice-versa.
There’s another thing wrong with ratings systems – the Bradman factor. I feel that most
if not all ratings systems start with the premise that Don Bradman is far and away the
best player ever, and ratings systems which are introduced with that premise as the basis
might be tweaked until the desired result is achieved, i.e. Bradman miles ahead of the
next best. The basis of this assumption is that batting and bowling are equal. But let’s set
this aside for a minute, and accept that we don’t know if batting and bowling are equal because they may not be.
Think of it this way – there’s a built-in quality threshold for bowling which doesn’t apply
to batting; if you’re not considered to be good enough, you don’t get to bowl, whereas
everyone is required to bat. So it just might be that bowlers contribute more than their
non-bowling counterparts.
So what do you suggest?
Glad you asked. Some time ago, I presented this article on howstat.com, which
retrospectively applied the ICC World Rankings to answer the question “Which was the
best ever Test team”.
The ICC rankings are calculated using a rolling period which takes into account each of
the series results within the period in question, and is based on a Series Points total for
each team involved in a particular Test series, with the points total being derived from
the series result and taking into account the strength of opposition.
For example, when Australia (ranked first with a rating of 123) drew with England
(ranked third with a rating of 96) in the 1938 Ashes series, Australia were awarded 399
points by the system, while England were awarded 507 points (the system rewards
England more highly as they performed above their ranking to force a draw, whereas
Australia under-performed by not defeating weaker opposition). Similarly, when
Australia (102 rating) white-washed top-ranked South Africa (107 rating) 5-0 in 1931-32,
Australia received 958 SPs (one of the highest totals ever) while South Africa received
only 337 SPs.
A mean(s) to an end
I had always intended to use the figures derived from the previous article to rate
individual Test players, by apportioning the team Series Points (TSPs) between the
individuals based on their performances during the series. In this way, those most
responsible for the success of their team are awarded the most points. We can then total
those points and answer all sorts of questions regarding Test performance over the full
history of Test cricket.
To find the individual points, we must first divide the TSPs between the disciplines, i.e.
batting, bowling and fielding. This is done by comparing the runs scored and wickets
taken against the mean Test performances against comparable opposition for the time
period, which is necessary because these averages of course vary over time. Then the
batting TSPs are divided up between the batsmen based on runs scored, fielding TSPs
are divided on the basis of catches and stumpings made and bowling TSPs divided on the
basis of wickets taken and (to a lesser extent) runs given up.
Using one of the examples above, this divides Australia’s total of 399 points into 214 for
batting and 185 for fielding/bowling, whereas England’s total of 502 points divides into
273 for batting and 234 for fielding/bowling (no surprise that a series which included
England’s famous innings of 903 for seven rewards batting more than bowling/fielding).
Just what we need – another acronym
We can then divide up the discipline points, i.e. those sub-divided from the team total to
give us Batting, Fielding and Bowling Individual Series Points (ISPs – hopefully this
usage will become more popular than Internet Service Provider – no? oh well), and
divide those between the individual players based on the number of runs scored, catches
made or wickets taken by them in the series. ISPs are thus dependent on a) the total
number of TSPs, which is derived from the ICC Test team ratings, b) the earlier
determination of which side of the ball was most responsible for the team’s success, i.e.
bat or ball, and c) the individual player’s performance.
All performances are not created equal
There are some interesting aspects to this system as compared to others, not the least of
which is that it takes into account the strength of opposition, so great performances
against strong teams are rewarded more than similar performances against weaker
opposition, so those players who rise to the occasion are rewarded accordingly.
Additionally, we are able to compare batsmen and bowlers (and all-rounders and for that
matter fielders) directly using the same basis, which obviously cannot be done with
traditional methods, either by using batting and bowling averages or even through the
ICC player rankings, which separate batting and bowling rankings. This is because this
system generates a single number representing the individual player’s performance in
the series which can be directly compared, so we can assess whose contribution was
more significant in a particular series victory, e.g. Don Bradman’s 447 runs or Clarrie
Grimmett’s 33 wickets in the 1930-31 series against the West Indies (slight edge to
Grimmett, 101 ISPs to 91).
There are lots of questions we can pose to this system, and derive objective answers to
those questions, for instance “Who has had the most major contribution in an Ashes Test
series?”, or “Who was the most successful West Indian Test cricketer of all time?”, or
even objectively identify the Player of the Series for Test series’ from before that
particular accolade was awarded.
I couldn’t resist – how good was The Don?
It’s my intention to eventually rank all Test players throughout history using this
method, but I just couldn’t help wanting to see once and for all just how much better Don
Bradman was than his contemporaries, so I first calculated the rankings for the decade of
the 1930s, when The Don was in his heyday.
I included the two winter tours of England to New Zealand and West Indies during the
season 1929-30, as all of the Test matches were played in 1930. Here is a list of the Test
series’ played during the years 1930-39, with the Player of the Series identified for each:Test Series [No. of Tests]
1929-30 England in New Zealand [4]
1929-30 England in West Indies [4]
1930 Australia in England [5]
1930-31 England in South Africa [5]
1930-31 West Indies in Australia [5]
1931 New Zealand in England [3]
1931-32 South Africa in Australia [5]
1931-32 South Africa in New Zealand [2]
1932 India in England [1]
1932-33 England in Australia [5]
1932-33 England in New Zealand [2]
1933 West Indies in England [3]
1933-34 England in India [3]
1934 Australia in England [5]
1934-35 England in West Indies [4]
1935 South Africa in England [5]
1935-36 Australia in South Africa [5]
1936 India in England [3]
1936-37 England in Australia [5]
1937 New Zealand in England [3]
1938 Australia in England
1938-39 England in South Africa [5]
1939 West Indies in England [3]
Player of the Series
KS Duleepsinhji (Eng)
GA Headley (WI)
DG Bradman (Aus)
B Mitchell (SA)
CV Grimmett (Aus)
GOB Allen (Eng)
DG Bradman (Aus)
Q McMillan (SA)
DR Jardine (Eng)
H Larwood (Eng)
WR Hammond (Eng)
LEG Ames (Eng)
H Verity (Eng)
DG Bradman (Aus)
GA Headley (WI)
B Mitchell (SA)
CV Grimmett (Aus)
WR Hammond (Eng)
DG Bradman (Aus)
J Cowie (NZ)
WA Brown (Aus)
WR Hammond (Eng)
L Hutton (Eng)
Total SPs
93
76
190
121
101
60
196
60
23
170
63
71
85
125
115
118
176
62
133
60
70
115
106
A couple of things stand out here – first, we can see names we would expect to be
prominent for the time period, e.g. Bradman, Headley, Hammond are confirmed as
dominant forces with multiple Player of the Series awards (it’s a necessity of any good
rating system that it should recognise the obvious). However, other less high-profile
players also come to the fore, e.g. South Africa’s Bruce Mitchell, who dominated the
South Africa team of the ’30s. It can be seen that Bradman in particular had some hugely
significant performances in 1930s series’, and although we already knew that his
performances were dominating, we can now quantify the importance of his batting
performances directly against e.g. the bowling of Grimmett and O’Reilly.
Turning to the decade as a whole, here is a list of the top-ranked players in terms of total
SPs for the Tests played in the 1930s:Total SPs
1085
837
639
603
556
521
512
471
369
350
Player
WR Hammond (Eng)
DG Bradman (Aus)
CV Grimmett (Aus)
SJ McCabe (Aus)
H Verity (Eng)
LEG Ames (Eng)
B Mitchell (SA)
WJ O’Reilly (Aus)
GOB Allen (Eng)
M Leyland (Eng)
Series
16
7
6
8
13
13
6
6
9
10
Matches
60
33
28
39
40
46
27
26
22
34
Hammond is way ahead of the next highest, Bradman, but this is less surprising
considering that Hammond played in almost twice as many matches as Bradman during
the period, with the England team being involved in 72 matches as compared to 39 for
the Australians during the 1930s. Other nations played in even fewer games, so ranking
based purely on the decade totals gives us only part of the picture.
I decided that a fairer way to rank players, and one which makes it easier to evaluate
dominance as opposed to opportunity, is based on the number of points per 5-Test series
(effectively points-per-game, but multiplied by five to give us a figure we can measure
against good 5-Test series performances) ; so for example if a player participated in 20
matches, we divide the total SPs by four to give his points-per-5-Test series. This is a
more objective method of ranking players based upon their opportunities, and allows for
the fact that England played in 19 series’ during the decade whereas for example India
played in only three.
Below is the revised list based on points per 5-Test series (PP5), with a minimum
qualifying number of Tests applied which is different for each country based on total
number of Tests played, e.g. England is 10, whereas India is only three :PP5
126.8
114.1
99.3
94.8
93.8
91.8
91.2
90.6
90.4
84.1
83.9
82.0
79.3
77.3
74.5
Player
DG Bradman (Aus)
CV Grimmett (Aus)
LN Constantine (WI)
B Mitchell (SA)
H Ironmonger (Aus)
GA Headley (WI)
L Hutton (Eng)
WJ O’Reilly (Aus)
WR Hammond (Eng)
KS Duleepsinjhi (Eng)
GOB Allen (Eng)
EA Martindale (WI)
HG Vivian (NZ)
SJ McCabe (Aus)
CS Dempster (NZ)
Matches
33
28
15
27
12
19
13
26
60
11
22
10
7
39
10
72.2
71.4
70.5
70.4
70.3
RWV Robins (Eng)
L Amar Singh (Ind)
Q McMillan (SA)
WE Bowes (Eng)
WH Ponsford (Aus)
18
18
11
14
20
Again we can glean some interesting points from this ranking. As we might have
expected, Don Bradman was the Player of the Decade, and as expected he’s quite a bit
ahead of the next man, Clarrie Crimmett – this to me is the most significant aspect of the
ranking system, in that it allows us to directly compare batting and bowling
performances side-by-side.
Learie Constantine is often cited as being one of the greatest to ever play the game (see
for example CLR James Beyond A Boundary), however traditional statistics have not
tended to support that claim – here we can see that, at least using this ranking method,
Constantine is right up there with the greatest. This high ranking also illustrates
Constantine’s consistency, as he didn’t show up as Player of the Series in any of the West
Indies 1930s series, although that was partly due to not playing in all of the tests (he
played in 15 Tests compared to 19 for Headley).
Bert Ironmonger is a surprise at number five, considering he was already 45 when he
made his Test debut, but he had great success against the West Indies in 1930-31 and
also South Africa in 1931-32, taking 31 wickets in four matches - and all this by a man
who had lost the forefinger of his bowling hand!
At number twelve is the West Indian fast bowler Manny Martindale, who with Learie
Constantine had used leg theory against England in the summer of 1933, shortly after the
“bodyline” tour, and was West Indies’ leading Test wicket-taker when England toured
there in 1934-35.
It was necessary to have a qualifying number of games, so that a couple of players who
shine brightly but briefly would not skew the list, however at least one of the nonqualifiers deserves a special mention. Harold Larwood played in only nine Tests
(infamously including the “Bodyline” Ashes series), but his number of points-per-5 Tests
would have placed him third between Grimmett and Constantine.
So there you have it
In conclusion, a new player rating system is presented, which allows us to compare
players directly regardless of their main discipline, i.e. batting or bowling. Additionally,
the ranking takes into account fielding prowess, and rewards good performances against
stronger opposition more than those against weaker opposition. Many and varied are the
cricket questions which can be answered objectively using this system:-
Who was the best ever Test spinner (Laker, Verity, Grimmett, O’Reilly, etc.)?
Which was the best ever all-rounder (e.g. Miller, Sobers, Hadlee, Botham, etc.)
Which was the most perfect all-rounder (i.e. ISPs split most equally between
batting and bowling)?
Who was the best fielder (highest number/percentage of ISPs for fielding)?
Who was the best wicket-keeper?
Formulating the rankings takes a good deal of time, so I intend to present them on a perdecade basis with a final, combined ranking once all Test series have been accounted for.
Then we’ll see if The Don was, as most cricket followers suspect, the best ever Test
player.
Download