Makes perfect sense to start an analysis from the beginning or the end of the year when the analysis is done in general -- choosing a calendar year is an accepted norm in the generic case.
In the case of the quoted discussion, this made even more sense because the underlying statement was the drop in quality of bowling somewhere around the turn of the millenium. One could argue that this happened in 2000 or 2001 but it happened around the that approximate time. Since we were comparing batsmen's performance across a set of bowlers, it makes sense to use a calendar year cut off for the purposes of analysis --just so that there is a common starting point. I picked 1/1/00 for all the batsmen under discussion -- Dravid, RP, Kallis et al because I was using the decade of the 2000's as a cut off and what better start for a decade than the beginning of a calendar year.
Exceptions to the generic case happen when choosing a year end cut off means that you choose part of an ongoing series that takes place over the winter, in which case, you modify the dates to include / exclude the series because a series is a continuos set of matches and when measuring form, its perhaps appropriate to include the entire series or exclude the entire series.
However, the discussion on this thread were about the timing of Ganguly being "dropped" from the squad and whether such was justified or not. This is a specific instance of a player being dropped, not a generic case of performance evaluation where the calendar cut off works
When a player is dropped, the minimum period his performance is evaluated is a period of 12 months starting from the point of the drop --this has been long established in this DG via prior analysis, simply because the rolling 12 months provides a better and more accurate indicator of performance. Choosing a calendar year cut off when a player is being dropped in March / June / Sep / Nov or anywhere in between makes little sense from a consistency standpoint.
One chooses a metric that provides the best means for analysis -- one does not apply a single formula to all situations.
Now back to the original post --
It measured Ganguly from 1/1/08 to 12/31/08. But why ? Ganguly retired on 11/6/08. So why not measure his performance till 11/6/08 ?.
In fact given the context of the discussion -- which was whether his dropping was justified or not -- why not measure his performance from the point at which DBV decided to drop him --the end of the SL series ? Since the SL series ended on 8/8/08, it provided too few data points for a proper determination of form if the generic calendar year cut off were used, hence my choice of 12 months prior.
However, using a metric which actually analyzes the events at hand does not sit well with the objective of pointed demagoguery, hence choosing an appropriate metric can be sidestepped by choosing a calendar year cut off.
Not only, that, it also makes sense to extend the timeline beyond the Nov 8th retirement date, only because it makes someone else look better. hence the term --convenient choosing of dates.
Sample the quote while comparing ganguly's results:
Here is his Daddy over the same period
Rahul Dravid
unfiltered 1996-2009 137 11256 270 53.60 28 1 1/18 39.00 0 188 0 Profile
filtered 2008-2009 21 1552 177 44.34 4 - - - - 28 0
wow. even a player going through a tough phase scores better than him! wow!
If you read the quote carefully, you will notice that ganguly's performance over the period Jan 1, 2008 until Nov 8, 2008 (when he retired) is compared to the other person's performance over 2 whole years --YES, the comparison for the other person spans from the beginning of 2008 up until the current date in 2009. This includes 13 months when Ganguly has played no tests (since he has retired).
This is called cherry picking numbers --numbers that benefit the point of view of the person making the argument, no matter how unproven it is.
If one looks closely, then one can see what the comparison results are when the appropriate timelines are used.
a)
When measured from Jan 1 2008 (adjusted backwards to include melbourne test staring 12/26/07) to Dec 31, 2008, this person (who was used as a comparison to show how "bad" ganguly was) has a record of
16 tests, 30 innings, 2 not outs, 826 runs, avg of 29.50 with 2 100's and 4 50's
ganguly didnt play against England, so even without playing the 2 extra tests:
14 tests, 27 innings, 3 not outs, 866 runs, avg of 36.08, 1 100, 5 50's
b)
While still using the calendar year cut off (modified to include the Melbourne test against Aus staring on 26th Dec, 2007) but extending only till Nov 8, 2008 (the day SG retired), this person's numbers are:
14 tests, 26 innings, 2 not outs, 683 runs, avg of 28.45, 1 100 and 4 50's
Ganguly's numbers for the exact same period:
14 tests, 27 innings, 3 not outs, 866 runs, avg of 36.08, 1 100, 5 50's
c)
When the calendar yr cut off (modified to include melbourne) is maintained, but the ending date is placed at the Sl series, then this person's numbers are:
10 test, 19 innings, 2 not outs, 563 runs, avg of 33.11, 1 100 and 3 50's
SG over the same period:
10 tests 19 innings, 1 not outs, 542 runs, avg of 30.11, 0 100 and 4 50's.
d)
Without the calendar yr cut off -- using trailing 12 months from the end of the Aussie series (Ganguly's retirement):
This person's numbers are
17 tests, 32 innings, 3 not outs, 874 runs, avg of 30.13, 1 100 and 5 50's.
Ganguly's numbers are:
17 tests, 33 innings, 3 not outs, 1400 runs, avg of 46.66, 3 100's and 6 50's
e)
Finally, using a trailing 12 month from the end of the SL series (because its after this series that Ganguly was "dropped") but adjusting bacwards to include the entire series against Eng (including complete series)
this person's numbers are:
16 tests, 31 innings, 4 not outs, 880 runs, avg of 32.59 with 1 100 and 5 50's
SG's numbers are:
16 tests, 31 innings, 2 not outs, 1325 runs, avg of 45.68 with 2 100's and 7 50's.
Unfortunately, in each of the above scenarios (bar 1 in which SG's average was marginally lower), there happens to be no comparison worth speaking of.
yet the need exists to show him in poor light.
When an appropriate metric --relevant to the context of the discussion-- does not work, then the need for different time periods and data that show Ganguly in poor light arises. When even the combination of numerous different time periods dont work, then the only remaining alternative is to present Ganguly's performance from 1/1/08 to 11/06/08 while presenting the comparison figure's data from 1/1/08 to 12/1/09 --an ENTIRE EXTRA YEAR of performance.
Cherry picking of numbers and dates in order to prove a point is a poor attempt to mislead others. A person who indulges in that in cares nothing about the topic or the integrity of the debate --he is interested in misleading and winning --by hook or by crook, thereby opening himself up to the charge of arguing for convenience.
AuthorTopic: Ganguly penalised for coming late in Ranji match - not a joke! (Read 8914 times)
March 17, 2006, 07:05:03 AM