>If you don't mind me butting in to this discussion,
Why not? Venky doesn't seem interested!
>I'd say that the problem with statistics is not that they are
>misleading in themselves, but that people ask them to determine
>things that they cannot. Or, rather, they are insufficiently
>precise about what the statistics represent.
I absolutely agree! As a "drop-out statistician" (just didn't have
the mathematical brain for it! Unkind mates said I just didn't have
the brain, period! :-) I recognize that the first thing a
statistician should do is to decide what he is sampling and why.
>In fact, as long as the sample size is sufficiently large, cricket
>statistics are extremely telling.
One must add a warning here ... "sufficiently large" can be a
subjective decision that may not be "statistically accurate". Which
is why when people throw around the word "statistics" it is
important to distibguish between whether they are actually talking
"statistics" or just "figures" ... which ain't necessarily the same
> Most of the instances which people cite to demonstrate a "clearly
>superior" batsman having a lower average than a "clearly inferior"
>one will involve someone either early in his career or with a
I think this is too sweeping a statement, Hugh. The only *serious*
debates on this subject that I have heard involve comparisons like
Venky's between Viv and other batsmen with higher averages
(Bradman, Barrington, Gavaskar etc.)
> We all know that a bowler can bowl brilliantly for an entire
>match without picking up a wicket, just as a mediocre bowler will
>occasionally luck in to a five wicket bag
(Hey! Is this a dig at my man Jimmy Adams? :-)
>- but over an entire, lengthy career, these things will even out.
>Of two bowlers who have each played, say, 100 Tests, the one with
>the average of 20 was, overall, the better bowler than the one
>with the average of 25 (strike rate being roughly equal). Now
>perhaps the one with the average of 25 had a beautiful, fluid
>action, and the other was an awkward, ungainly bowler. You might
>quite justifiably say you'd rather watch the more stylish bowler
>in action, but you'd simply have no justification to say he was
>the more effective bowler.
I think this is the point I was trying to make in my original post.
We seem to agree on this, but not on your next point.
>The same is true of batsmen, in fact. As I type, the radio in the
>background is carrying the commentary on the NZ-Windies Test. As
>is usually the case, the commentators were surprised that Patel,
>coming in at the fall of the seventh wicket, should be such an
>obviously stylish bat. And it is true, Patel has all the shots,
>and executes them with flair. Unfortunately, his Test average of
>about 20 is a very accurate reflection of his ability at Test
>level. Now, you might argue that he is a more "attractive" bat
>than, say, Andrew Jones - but Jones's 40+ average was an accurate
>reflection of *his* Test batting ability.
Yes ... but hold on ... we're talking about more than just flair or
pretty stroke play. The real argument is: does his statistics tell
the full story about an acknowleged top flight batsman like IVAR?
Can we use his statistics to compare him accurately with another
top flight batsman like say Barrington? I say (like Venky) ... No!
Whereas, talking about bowlers, I think differently, agreeing with
the views you outline in the previous para.
>Taking your example of Barrington vs. Richards, my guess would be
>that this is simply a matter of not being sufficiently precise in
>what we are asking the statistics to tell us, or of asking them to
>tell us things they simply cannot. In the first case, I imagine
>that if we compared their strike-rates (runs per 100 balls),
>Richards' would be dramatically superior, which would indicate
>that if you want to *win* a Test Richards is the better option,
>but if you want to *draw* you'd go for Barrington.
Now this is where I think your argument goes awry ... Compton (and
to some extent, Botham) was the Viv Richards type of batsman ...
yet I think he was equally at home "winning" tests or "playing for
a draw in a tight match". In other words, his "stats" don't mark
him down as being either a Viv Richards type or a Ken Barrington
type. Which is why I am suspicious of the value of stats as the
only criterion for assessing a batsman ... even though I buy them
as a good yardstick for rating a bowler.
Sriram Narayan also wants in.
>Apologies also for butting in, but I have been itching to pitch in
>and simply can't resist any more.
Welcome aboard! The more the merrier!
>It seems to me that the problem with averages is that it treats
>all runs as having the same value, whereas in practice this is
>simply not the case. Depending on a host of factors like quality
>of the opposition, state of the pitch, atmospheric conditions,
>match situation etc. the value of two scores of equal magnitude
can be quite different.
True ... but that's only a valid criticism if you are considering
"two scores" in isolation. A batsman's statistics involve his whole
career, and over the long course for most test batsmen's, these
conditions you mention will undoubtedly crop up in each case.
Hence, it is no point saying that Lara's 375 was not as valuable as
Waugh's double ton against the WI, for eventually the latter also
had his chance against much the same England XI, and Lara is even
now playing against a pathetic (sorry Hugh!) attack which two of
his colleagues have pillaged for double tons while he has yet to
make a 50.
>Since these factors are somewhat subjective, it is difficult if
>not impossible to quantify their effect. Even within our own time,
>it is sometimes difficult to rate a particular batsman - witness
>all the discussions about the quality of Dean Jones, Azhar or
>Mohinder Amarnath. In the case of Barrington and Richards, the
>problem is complicated by the fact that they played in slightly
Herein lies the conundrum concerning statistics: what you say above
is valid, and I agree with it as my previous statements show. Yet,
there is no gainsaying the fact that a Bradman stands taller than
all comers, contemporary or not, and paradoxically it is the
consistency of his statistics that makes his superiority undeniable
(undeniable to MOST, I hasten to add!)
>Barrington's is a strange case: when people pick teams of the '60s
>etc on rsc, nobody dares to leave him out, such is the power of
>the almighty average. And yet Fred Trueman had no trouble picking
>Tom Graveney, (who is hardly ever mentioned on rsc) over
>Barrington in an England team of the 50's and 60's.
Excellent example ... even if I do have the temerity to disagree
with Fiery Freddie!
>In fact, other than Chandrasekhar, I have never seen anyone
>nominate Barrington as the best batsman he has ever bowled to. You
>can find plenty of bowlers who thought Richards was the best
>batsman they bowled to.
I think this is a better example, and one I CAN agree with.
>Averages are good starting point when comparing players, but when
>we "know" that X is better a batsman than Y, but Y has the higher
>average, we are in a bind. We can dig further and do some fancy
>statistical analysis, but in the end discrepancy will remain and
>fuel endless discussions. There is a touch of deja-vu about all
>this. You can find some critics of the '30s arguing Trumper was a
>better batsman than Bradman.
(Can't say I've heard THAT argument!)
> Why? He was a joy to watch, was masterly on sticky wickets and
>many times threw his wicket away after reaching a 100 because he
>felt the next guy deserved a chance!
I'd say that makes him a right Wally! :-)
>Having said all this, I must admit a bias in favour of averages,
>especially if it is over 99 and even if people escorted the ball
>to the boundary in those days...-:)
In the final analysis, I have to agree with that! (Though I think
the last bit is just a leetle unfair!) Which is why I endorse what
Hugh says in a subsequent post:
>I still think that any difference that is truly significant in
>terms of "having an impact on the ultimate outcome of a game of
>cricket" should be accessible to statistical analysis. After all,
>you don't get extra points for style in cricket - you win by
>scoring more runs than the opposition and bowling them out twice.
... and also ...
>No one claims that average (bowling or batting) *by itself* is
>everything - but that doesn't mean the statistics are not an
>accurate reflection of a player's contribution to his or her
I would say a "fair" reflection rather than an "accurate" one.
So ... it's not yet time to consign Wisden to the s***heap! :-)
The opinions expressed in this post are not those of the management ...
but PLEASE don't tell her!