Competitive Balance in County Cricket: Part I — By Win Percent in Finished Games

For many years now, I have wanted to measure competitive balance in English county cricket between the first official County Championship in 1890 and the last single-division Championship in 1999.

There are however many problems with measuring competitive balance in county cricket with the techniques used for baseball, basketball, gridiron, ice hockey, rugby, [Australian rules] football and even soccer leagues. These include:

1. county cricket games often — in some seasons more often than not — can end without any result
2. over the history of the County Championship, methods of point scoring have varied immensely
1. since 1968, in fact, a large part of scoring has been “bonus points” not directly related to the result of a match
2. the result is that a consistent measure of balance may not be accurate for all eras of county cricket
3. up until World War II, counties usually played schedules of substantially varying length, which makes calculating the idealised standard deviation by conventional means impossible

For these reasons, I plan to analyse competitive balance in county cricket by multiple criteria to see how they compare and how competitive balance has changed over time in the sport.

This first post will analyse competitive balance based purely upon percentage of wins in finished matches. This was the original method of calculating the Champion County, but was discarded because it was seen as encouraging unattractive cricket. Nonetheless, thinking about this idea over many years I have always thought of analysing the observed standard deviation of win percentage in finished matches as the default method by which within-season competitive balance in county cricket might be estimated.

In order to gain at least some idea of the idealised standard deviation of winning percentage in finished games, I have calculated the arithmetic mean of the number of finished games in each season, and based the idealised standard deviation on the ordinary formula of (sqrt(n))/2n. I have then calculated a competitive balance index as the ordinary formula of (ASD-ISD)/(MSD-ISD), where:

MSD = sqrt((N+1)/(12(N-1)))

with N equalling the number of teams in the competition. I have also included skewness and kurtosis for more details on the actual shape of the distribution. Extreme values have been shaded as I previously did in my 2024 study of the “Revolution of 1959”. I have also provided all-series averages for every season of single-division official County championship cricket.

CBI for Each Single-Division County Championship Season, 1890 to 1999:

Season	CBI	SKEW	KURT
1890	0.462358	-0.960141	-0.054772
1891	0.331451	+0.229151	+0.170827
1892	0.842595	-0.210308	-1.337585
1893	0.239742	+0.168580	-0.539230
1894	0.618669	+0.176070	-0.782851
1895	0.327522	+0.307711	-0.623449
1896	0.450441	+0.452382	-1.143677
1897	0.774982	+0.049695	-1.104113
1898	0.750282	+0.411154	-1.371064
1899	0.468059	+0.436129	-1.087182
1900	0.784729	-0.012455	-0.383119
1901	0.571871	-0.119775	+0.485723
1902	0.317136	+0.446170	+0.238533
1903	0.550588	-0.429954	-0.294312
1904	0.575301	+0.493857	-0.213935
1905	0.732449	-0.140498	-1.168476
1906	0.784290	+0.241416	-1.432568
1907	0.761771	+0.046237	-0.682596
1908	0.635791	+0.542957	+0.016871
1909	0.697139	-0.043934	-1.021497
1910	0.621227	-0.364317	-0.144691
1911	0.774390	-0.538290	-1.187153
1912	0.913412	+0.081078	-1.385633
1913	0.564809	+0.335478	-0.908739
1914	0.796326	-0.222787	-1.080853
1919	0.456512	+0.216030	-1.152839
1920	0.794512	-0.380103	-0.857362
1921	0.709633	+0.240165	-1.242194
1922	0.839410	-0.026496	-1.020781
1923	0.835631	0.336090	-1.018051
1924	0.839410	-0.120697	-1.175160
1925	0.858227	+0.271994	-1.036225
1926	0.791844	+0.469658	-1.004929
1927	0.667121	-0.257271	-0.864866
1928	0.878656	+0.073174	-0.473664
1929	0.666758	-0.310386	-0.995948
1930	0.690781	+0.662334	-0.771007
1931	0.454629	+0.085538	-0.322792
1932	0.793709	+0.245013	-1.078973
1933	0.733655	0.011935	-1.166727
1934	0.573070	-0.166672	-1.118039
1935	0.572791	-0.067031	-0.122990
1936	0.508035	-0.395538	-0.508390
1937	0.657703	-0.532268	-0.112059
1938	0.474863	-0.208981	+1.354882
1939	0.523154	-0.687685	+0.030232
1946	0.519580	+0.543093	-0.433125
1947	0.445312	+0.434828	+0.002714
1948	0.373594	+0.137454	-1.469418
1949	0.367723	+0.279477	-0.318211
1950	0.465111	+0.634463	-0.357330
1951	0.505072	-0.065867	-0.496338
1952	0.508765	+0.484745	-0.211689
1953	0.362378	-0.676085	-0.018059
1954	0.522281	-0.256696	-1.183995
1955	0.409663	+0.266860	-0.459407
1956	0.373128	-0.090026	-0.838410
1957	0.495334	+0.373986	-0.129682
1958	0.161731	-0.570909	-0.314868
1959	0.141932	-0.737029	-0.058184
1960	0.388917	-0.502500	-0.743616
1961	0.299882	-0.241965	-0.244088
1962	0.497791	-0.362190	-0.683229
1963	0.323291	-0.403720	-0.159527
1964	0.569966	-0.003793	-1.185038
1965	0.242032	-0.021824	-1.123755
1966	0.196613	+0.002542	-0.898257
1967	0.337281	-0.660471	+0.625506
1968	0.215358	-0.549002	-0.357759
1969	0.484943	+0.583272	+0.128426
1970	0.072694	+0.721005	+0.298446
1971	0.193056	+0.223475	-1.189572
1972	0.521367	+0.177093	-0.235752
1973	0.520214	+0.108313	-0.004182
1974	0.623300	+0.225952	-0.517476
1975	0.541885	+0.173766	-0.741075
1976	0.139265	-0.147423	-1.202972
1977	0.217728	-0.506065	+0.113360
1978	0.446960	+0.975866	-0.540541
1979	0.427807	-0.728434	+0.028456
1980	0.107447	-0.616210	+0.674108
1981	0.361980	+0.183950	-1.009637
1982	0.462704	-0.551719	+0.520057
1983	0.630372	-0.121651	-1.771580
1984	0.497179	-0.408316	-0.375583
1985	0.307034	-0.082135	-0.762481
1986	0.004652	+0.188331	-0.508590
1987	0.521703	-0.022970	-1.246956
1988	0.163785	-0.567549	+0.368864
1989	0.239509	+0.967162	-0.077891
1990	0.334461	+0.777225	-0.488989
1991	0.052972	-0.127006	-1.252023
1992	-0.097526	-0.367632	+1.399614
1993	0.238570	+0.230888	+0.209866
1994	0.179998	+0.525857	+0.763104
1995	0.462702	+0.681663	-0.761303
1996	0.636679	-0.283181	-0.378578
1997	0.337709	-0.472894	-0.460810
1998	0.556307	+0.653322	+0.184129
1999	0.180620	+0.947094	+3.056189
Average	0.48758245	+0.02190829	-0.4853056

Graph of Competitive Balance with 5- and 15-Year Means:

Competitive Balance Index by Win Percent in Finished Games in the County Cricket Championship, 1890 to 1999

Graph of Competitive Balance Index, Skewness and Kurtosis:

Competitive Balance Index by Win Percent in Finished Games, alongside Skewness and Kurtosis of Win Percent in Finished Games, in the County Cricket Championship, 1890 to 1999

Conclusions:

The first graph above clearly shows an improvement in competitive balance in the County Championship since the middle 1930s. In fact, as a fifteen-season running mean, the (ASD-ISD)/(MSD-ISD) index as defined above fell from around 0.75 for the fifteen seasons centred upon 1922 [1911 to 1929] to less than 0.27 for the fifteen seasons centred upon 1987 [1980 to 1994]. This constitutes a dramatic contrast to [Australian rules] football, soccer and basketball leagues, which have seen no improvement in competitive imbalance over the past century-and-a-quarter.

There are several possible explanations for the dramatic improvement, which are not mutually exclusive:

1. standardised professional squads after the 1930s meant that no team relied on low-quality amateur players as many counties before 1930 substantially or largely did
• the exceptions or partial exceptions were:
1. Lancashire, Nottinghamshire, Surrey and Yorkshire — and to a smaller extent Kent, Sussex and Warwickshire
• these counties received sufficient industrial patronage to afford large professional staffs so would only play amateurs who could compete with their best professionals
2. Kent (again), Middlesex, Essex and Hampshire
• these counties possessed a substantial number of high-quality amateurs associated with business in London but able to devote full or partial summers to county cricket
2. standardised mass production and improved coaching of pace and seam bowlers produced more uniform quality amongst counties after the middle 1930s
• this made bowling much more consistently economical in runs conceded and also much cheaper to develop
• of course this greater uniformity at the cost of eliminating the possibility of financially self-supporting first-class cricket, which requires overwhelming predominance of spin alongside the most limited pace and seam
3. increasing breadth of search for players, which began in earnest in the 1930s and 1940s (e.g. Jack Walsh), should have reduced variance in performance
• in this context, the introduction of “special registration” for England-eligible players after World War II should have further improved competitive balance as players were no longer held by teams without need

The skewness data does not suggest a great deal of interest from cursory examination.

The kurtosis data suggests increasing (less negative) kurtosis over time, which is highly consistent with the theories of standardisation noted in 1) and 2) above.

The data suggest that standardisation has either been more radical or more effective (or both) in county cricket than in most other team sports, where similar standardisation has either had no effect on comptitive balance or even, as in [Australian rules] football, lowered it as the talent pool becomes more and more limited. Why a focus on tall, fit pace bowlers that developed after the 1948 Ashes series — and can be traced back earlier — should not lower the talent pool as a similar focus in [Australian rules] football since the 1980s has does deserve discussion as I cannot see a definitive and obvious answer.

A parody and a true cricket tale

Over recent months, I have repeatedly sang this song (or variations) to my brother:

“Modern batsmen have a pitch
Ninety percent rigged
And on that pitch they score many runs
But deserve very few
With a cross-bat here
And a cross-bat there
Here a cross
There a cross
Everywhere a cross-bat
Modern batsmen have a pitch
Ninety percent rigged”

I have often thought of writing further lines about how covered pitches plainly constitute kangaroo courts for the spin bowler, but have never sung them.

Me and my brother have fiercely debated whether covered or uncovered pitches are fairer. The facts, at least judging from English first-class bowling and batting data, firmly suggest covered pitches on average are more difficult for spin bowlers than even abnormally bad “sticky” wickets were for batsmen. Plainly put:

1. many batsmen managed to consistently cope with the most difficult uncovered pitches and achieve averages comparable to what the best batsmen under covering achieve
2. on covered pitches, zero spin bowlers achieve consistently anything like the same averages that old spin bowlers expected on uncovered pitches
• even overseas spinners — almost invariably superior to English ones under unfavourable conditions — have never consistently come close to either the averages of the old spinners nor those of the best fast bowlers

My brother consistently says that with proper coaching spin bowlers can master covered pitches properly. He also said that the effort required to bowl spin effectively outside England was accepted there and ought to have been accepted in England. However, so early as the 1956 Wisden — when concern at the declining appeal of first-class cricket was becoming serious — it was noted that the emphatic finger spin needed outside England was the most difficult method of bowling to master, and fewer and fewer were developing it in preference to easily mastered seam and pace. My brother also ignores how in many countries where “rollers” [a term used by Ashley Mallett in opposition to genuine “tweakers” in one 1996 Wisden article] were unable to turn the ball genuine finger spinners disappeared so completely as in England. This was especially true in the West Indies and South Africa, and suggests two things:

1. by 1960, the “tweaker” was already viewed as an extravagant luxury both in runs conceded and in cost of development
1. John Woodcock would note in 1963 how the “tweaker” was already viewed too expensive in runs to compete against constantly improving fast and even medium-pace bowlers
2. it was undoubtedly clear, though never discussed, that improving the “tweaker” to be less costly in runs was constantly becoming less and less financially viable as pace bowling became better and better
2. development of spin bowling even partly able to cope with fundamentally rigged conditions is possible only under very restricted social conditions:
1. in pre-war Australia, leisure was sufficiently abundant that young boys had the time to develop the abnormal effort to develop “real finger spin” that might turn on fundamentally unfair pitches
2. in the Indian subcontinent, extreme abundance of labour means that young boys could be trained to develop genuine finger spin at relatively low cost vis-à-vis the rest of the cricket world

Hence, Clive Lloyd’s decision, in Vic Marks’ word [The Wisden Illustrated History of Cricket, pages 201-202], that

“spin bowling... was a luxury, especially if four genuinely fast bowlers were available”

was entirely consistent with most captains’ experience over the preceding quarter-century as the extreme difficulty of bowling spin on well-prepared covered pitches became less and less affordable for young boys. The problem is that once spin becomes viewed as an extravagant luxury first-class cricket is financially doomed in the long term, and its problems have been noted by Huw Turberville in his recent The Final Test. Virtually all historical instances where first-class cricket was economically self-supporting correlate practically perfectly with the greatest weaknesses in fast and medium-pace bowling, and exclusive reliance upon spin of whatever quality.

However, it is tempting to say that the greatest spin attacks in history — circa 1934 Australia and circa 1970 India — are fundamentally so costly to develop that first-class cricket becomes unprofitable once those costs are factored in, although attendance figures imply first-class cricket would always be profitable if the quantity and quality of spin developed in those two cases could be maintained. History shows that those standards of spin were never maintained, and the logical explanation is that maintaining from generation to generation the extreme effort involved in developing the best spin bowlers was inherently too expensive and difficult. This further supports the contention that covered pitches are, fundamentally, rigged against the spinner incomparably more than uncovered pitches are against batsmen.

Assessing the theory of a “Revolution of 1959” part II

In two previous posts (click here and here), I have argued that English county cricket shows evidence of revolutionary change in bowling statistics and records during the late 1950s. Before this shift, spin bowlers almost always headed the averages, whereas afterwards fast bowlers almost came out at the top. Before the shift, also, spin bowlers took far more wickets than they did in subsequent seasons.

Another change with the putative but quite possibly real “Revolution of 1959”, which I did not discuss in the two previous posts, is a dramatic decline in the frequency of large innings and match wicket hauls. I have long known that there have been twenty cases of a bowler taking seventeen or more wickets in a match in England, but until recently not a solitary post-World War II instance. I also knew that there were relatively many (I did not count at the time) cases of sixteen wickets in a match, but not one by a non-touring bowler in England between 1957 and 1999. A few years ago I calculated that only three seasons between 1888 and 1960 (1920, 1946 and 1950) had no case of a bowler taking fifteen or more wickets in a match, whereas between 1961 and 1988 there were a mere five such cases by non-touring bowlers in twenty-nine seasons. There was indeed no fifteen-wicket match return by an England-qualified bowler between 1969 and 1993 inclusive. Also, no season between 1888 and 1969 saw no case of a bowler taking nine or ten wickets in an innings, whereas in the 1970s and 1980s there were only a handful of cases.

To test this evidence for a “Revolution of 1959”, I have tabulated all the cases of nine or more wickets in an innings, and of fifteen or more wickets in a match, by “home” (non-touring, not necessarily England-eligible) bowlers. To be more precise, I have also compiled the instances of bowlers taking ten wickets in an innings, and sixteen and seventeen wickets in a match. The data are compiled via the Association of Cricket Statistics and Historians. High totals are shaded in gold, and record totals for a season are bolded.

Season	9+w/i	10w/i	15+w/m	16+w/m	17+w/m
1836	2		1
1837	2	1	1	1	1
1838
1839	1
1840	1		1
1841	1
1842	2		1
1843	1		1
1844	1		1	1	1
1845	1		1	1
1846	1
1847
1848	1	1	2	1
1849
1850	4	1	1
1851	2	1	1
1852	1		1	1	1
1853	1		1	1	1
1854
1855	1		1	1
1856
1857	1		1
1858	2
1859	1	1	2
1860	4		1
1861	2		1	1	1
1862	2	1	3
1863	2		2	1
1864	3
1865	3	2
1866
1867	1
1868	3
1869	1		1	1
1870	1	1
1871	5	1	1
1872	2	1	1
1873	2	2	1
1874	2	1
1875	3		1
1876	2		1	1	1
1877	2		3	1	1
1878	2	1	1
1879			1
1880	1
1881
1882
1883
1884
1885	2		1	1
1886	4	1
1887
1888	3	1	1	1
1889	2		1
1890	4	1	2
1891	2		2
1892	3		1
1893	2		3
1894	6	1	4
1895	10	2	8	3	1
1896	2		1
1897	2		3
1898	3		6	1
1899	6	1	3
1900	6	2	1
1901	2		2
1902	6		2
1903	2		2
1904	6		5
1905	7		2	1	1
1906	10	2	6	2
1907	8	1	5	1	1
1908	1		1	1
1909	5		1	1
1910	5		1	1
1911	6		1
1912	2		4	1
1913	3		3	1	1
1914	6	1	3	1
1919	1		1	1
1920	5
1921	8	4	2
1922	7	1	3	2	1
1923	4	1	2	1	1
1924	5		1
1925	4		2	1	1
1926	3		1	1	1
1927	6	1	3	2
1928	3		3
1929	7	3	1	1
1930	6	1	3	2
1931	6	2	4
1932	5	2	3	2	1
1933	4		2	1	1
1934	5		2
1935	3	1	2	1
1936	6	1	3
1937	5	1	3	2	1
1938	3		2
1939	9	2	3	3	1
1946	3	1
1947	5		6	1
1948	5	1	3
1949	5	2	3
1950	1
1951	2		1
1952	4		2	1
1953	4	1	2	1
1954	5		3	2
1955	7		3
1956	6	4	3	2	1
1957	3		1
1958	4		3
1959	4	1	1
1960	3		1
1961	1	1
1962	1
1963	1
1964	5	1	2
1965	3
1966	3
1967	2		1
1968	1		1
1969	1
1970
1971
1972	1
1973
1974
1975	2		1
1976
1977
1978	1
1979	1
1980
1981	1
1982	1
1983
1984
1985	1
1986	2
1987
1988	1
1989			1
1990	1
1991	1
1992
1993	2
1994	3	1	1
1995	4		1
1996	1
1997	1
1998
1999
2000	2		1	1
2001
2002	1
2003	2		1
2004
2005
2006	2
2007	1	1
2008
2009
2010	1
2011	1
2012	2
2013			1
2014	1		1
2015	1		1
2016	2
2017	2		1
2018
2019	1		1	1	1
2021	2
2022	1		1
2023			1

The table shown above gives, on the whole, more confirmation of a radical change around the late 1950s than the table of spin bowling predominance and wicket hauls in my preceding post noted at the beginning. Evidence of a dramatic reduction in the number of large wicket hauls in England at that time can be clearly seem from the following graph, which tabulates the table above into eleven-season running totals.

11-year moving totals of instance of 9 or ten, ten, fifteen or more, sixteen or more and seventeen or more wickets in a match by a non-touring bowler in England.
The figures cover seasons from 1836 to 2023, or eleven-year periods centred between 1840 and 2018.

The graph above shows:

1. low numbers of high wicket hauls up to 1880 when relatively very little first-class cricket was played
2. high numbers of high wicket hauls between about 1880 and 1960
3. low numbers of high wicket hauls after 1960 when pitches became more completely covered

The fall in frequencies of high innings and match wicket hauls during the late 1950s shown in the graph above is distinctly steep. There were forty hauls of nine or ten wickets in an innings between 1954 and 1964, but only eighteen in the overlapping period from 1961 to 1971 just seven years later. This fall is convincing evidence for a “Revolution of 1959”.

It should be noted that similar changes appear to have occurred in certain other countries. This is especially true of Australia, where very large wicket hauls have always been much rarer than in England, especially amongst spin bowlers, due to the radically different soils and consequently pitches. Ever since the close of the pluvial era from 1886 to 1894/1895 Australian pitches have proved impossible for English spinners. Before the “Revolution of 1959” these bowlers dominated county cricket — yet not one was a significant force in Australia, where the tighter grass binding reduced their gentle rolling spin to slow straight balls with zero deviation.

No spin bowler has taken ten wickets in an innings in Australia since George Giffen in 1883/1884, and no non-touring bowler has taken fifteen wickets in a match in Australia since Leslie Fleetwood-Smith in 1935/1936. In fact, no non-touring bowler took nine wickets in an innings in Australia between 1979/1980 and 2015/2016. Yet, there were quite a few cases of large match hauls in Australia during the interwar years. This suggests that the “Revolution of 1959” — whilst definitely real — was largely due to England’s efforts to counter Bradman’s great 1948 team, who demolished every prolific wicket-taking county spin bowler, and who emphasised strong, deep pace and seam attacks to minimise the ability of batsmen to hit. With bowlers bowling shorter spells, there was less and less opportunity for them to bowl enough to take large numbers of wickets — but at the same time Australian bowlers became much harder to hit than the spin-dominated interwar attacks were. England responded by phasing out most spin and turning to tight, short-of-a-length seam bowling. This helped England win more but made county cricket vastly less entertaining, turning every county into a losing proposition by the 1960s.