Swervy
International Captain
i see your reasoning, but dont agree with your method.Why get rid of the players top 2 scores, why not 1, or 10, or all scores over 200 etc.It just so happens that the 2 highest scores for Lara are 2 of the 3 highest ever in tests,so gets penalised for thatviktor said:and here's one reason why:
It is common in dealing with data sets to come across xtremely large or xtremely small values within the data set. these data points, known as outliers, are generally neglected when calculating properties of the data set as they do not "show" the trend of the data.
with this premise, i conducted a "Richardian" exercise.
I neglected the highest two scores by 5 players i consider to be 5 of the best; Sachin, Dravid, Richards, Sobers and Lara (Bradman is another outlier, so out). To save me a little trouble, I assumed both the innings closed (b'men out).Then for this modified data set, I calculated the avera ge and calculated the drop in average from the original data. the results are:
Player Ori avg Mod avg
Lara 53.43 49.27
Dravid 58.09 54.76
Richards 50.23 47.72
Sobers 57.78 54.31
Sachin 57.39 55.29
while Lara's avg drops by ~4 runs, that of Richards, Dravid and Sobers drops by 3-3.5 and Sachins by ~2. to me that indicates Lara is not as consistent as the others, not by a lot mind you but still enough for the level these guys are at.
I know stats don't always tell the true story but I was hoping thus analysis removes any discrepencies in such a basis for analysis.
Please do comment,
cheers
Why not exclude all scores under 10 as well????
Would a standard deviation measurement be a better method???
Also match stuations???Chasing a target in the 4th innings,batting under pressure etc...