Analysis of Batting Line-ups

Analysis of Batting Line-ups

Post by Charles Le » Tue, 19 Dec 1995 04:00:00


Quote:

>The discussion a little while ago about strong Test batting
>line-ups prompted me to do a little analysis. Ideally, I would
>have looked at every Test line-up that has ever appeared, but
>since that was impractical I have made a selection that
>hopefully includes most of the strongest line-ups (before anyone
>points it out, yes this was largely a subjective process). Since
>the method I've adopted is very simple, if you think I've
>overlooked a strong batting side you can apply it yourself and
>compare the results with mine.

[Excellent research and analysis snipped]

Short of some complicated computer generated formula a la C&L
Ratings, I think John has come up with a neat if empirical method
of making a fair assessment of two quantities that can never be
proved, one against the other. He has pointed out the obvious
flaws himself, and I can see no way of getting around two of them
- i.e. the "Bradman factor" which one simply has to accept, and
the impossibility of assessing the effects of different bowling
quality and pitch conditions (I suppose one could even add the
effect of quality of fielding). However, riding on John's
coattails, I played around with his top four batting line-ups
with a view to seeing how his system could be adapted to taking
the batsmen's form *at the time* into account, rather than
relying entirely on their career averages.
In each case (Aus vs Eng 1948, Aus vs Eng 1930, Eng vs Aus 1928,
Eng vs Aus 1938 which is the order in which John's analysis
placed the teams) I worked out the median average of the top six
batsmen that John chose, for the particular series being
analysed. I then multiplied that figure by the median career
average of the six batsmen, and divided by 100 to arrive at a
convenient figure by which to grade the performance of the line
up. Interestingly, this resulted in a reversal of the top two
line ups, as well as #3 and #4 as John had them!
Here are my results alongside John's:

        John's Grading                       My Grading
1) Australia 1948     56.90        1) Australia 1930   35.86
2) Australia 1930     54.96        2) Australia 1948   34.37
3) England 1928       50.11        3) England 1938     28.45
4) England 1938       49.97        4) England 1928     27.78

[Afterwards, as a purely academic exercise, I deducted Bradman's
figures from both Australian line ups, and whereas John's grading
remained the same, my grading did a flip flop - due I suppose to
the enormous aggregate (and average) that the Don had accumulated
in 1930.]
I have not experimented with any other line ups, but I venture to
think that by interpolating the median average which the batting
line generates in a particular series, it *is* possible to cancel
out career dips or peaks that occurred outside that particular
series.
Not scientific perhaps, but an interesting empirical exercise!
Cheers!
Charles

--
"... problems that keep me awake at night - like ... how did Adam keep his
fig leaf on ... or what does one call a male ladybird?"
                                                        - Denis Norden

 
 
 

Analysis of Batting Line-ups

Post by John Hal » Wed, 20 Dec 1995 04:00:00



Quote:

>>The discussion a little while ago about strong Test batting
>>line-ups prompted me to do a little analysis. Ideally, I would
>>have looked at every Test line-up that has ever appeared, but
>>since that was impractical I have made a selection that
>>hopefully includes most of the strongest line-ups (before anyone
>>points it out, yes this was largely a subjective process). Since
>>the method I've adopted is very simple, if you think I've
>>overlooked a strong batting side you can apply it yourself and
>>compare the results with mine.

>[Excellent research and analysis snipped]

>Short of some complicated computer generated formula a la C&L
>Ratings, I think John has come up with a neat if empirical method
>of making a fair assessment of two quantities that can never be
>proved, one against the other. He has pointed out the obvious
>flaws himself, and I can see no way of getting around two of them
>- i.e. the "Bradman factor" which one simply has to accept, and
>the impossibility of assessing the effects of different bowling
>quality and pitch conditions (I suppose one could even add the
>effect of quality of fielding). However, riding on John's
>coattails, I played around with his top four batting line-ups
>with a view to seeing how his system could be adapted to taking
>the batsmen's form *at the time* into account, rather than
>relying entirely on their career averages.
>In each case (Aus vs Eng 1948, Aus vs Eng 1930, Eng vs Aus 1928,
>Eng vs Aus 1938 which is the order in which John's analysis
>placed the teams) I worked out the median average of the top six
>batsmen that John chose, for the particular series being
>analysed. I then multiplied that figure by the median career
>average of the six batsmen, and divided by 100 to arrive at a
>convenient figure by which to grade the performance of the line
>up. Interestingly, this resulted in a reversal of the top two
>line ups, as well as #3 and #4 as John had them!
>Here are my results alongside John's:

>        John's Grading                       My Grading
>1) Australia 1948     56.90        1) Australia 1930   35.86
>2) Australia 1930     54.96        2) Australia 1948   34.37
>3) England 1928       50.11        3) England 1938     28.45
>4) England 1938       49.97        4) England 1928     27.78

>[Afterwards, as a purely academic exercise, I deducted Bradman's
>figures from both Australian line ups, and whereas John's grading
>remained the same, my grading did a flip flop - due I suppose to
>the enormous aggregate (and average) that the Don had accumulated
>in 1930.]
>I have not experimented with any other line ups, but I venture to
>think that by interpolating the median average which the batting
>line generates in a particular series, it *is* possible to cancel
>out career dips or peaks that occurred outside that particular
>series.
>Not scientific perhaps, but an interesting empirical exercise!
>Cheers!
>Charles

An interesting response. I had considered using averages in the Test
series in question instead of career averages when compiling my figures.
The snag with that is that the strength of the opposition bowling and
the quality of the pitches in that particular series are likely to
influence the results to a considerable extent. For example, I suspect
that with Charles's method the 1930 Australian team leapfrogs over the
1948 one because the pitches were better for batting in 1930 than they
were in 1948.
--
There is no sincerer love than the love of food.  
 George Bernard Shaw (1856-1950)