Quote:
Originally Posted by Miraz
I am not really convinced that this AC - (Average in terms of Consistency = [runs - (HS + LS)]/(matches - 2) is a good statistical tool to judge consistency.
Too much reliance on two innings scores, (highest score and lowest score), doesn't put enough emphasize on other scores within the same time period.
|
I see your point. Here's mine: I followed the idea of "dispersion" over a certain period. Therefore I dropped the top and bottom scores which could have been "flukes" to arrive at a more consistent average. That way, no single "flash in the pan" or "bad day at the office" can inflate or deflate an API too much.
There are margins of error no doubt, and admittedly, AC can make someone's average look worse it perhaps is, but that's only due to real-world inconsistencies being taken into account.
Then again, in the unlikely matter of the batsman consistently having self-similar type of innings without such events, AC will be closer to API.
This is nothing new. Similar sort of scoring with particular variations are done quite extensively in Olympic events like gymnastics and diving for similar reasons, and the idea of both negatively and positively biased judges influencing the score.
This is just one way of looking at it. It is by no means the ONLY way ...