The Top Ten Test Innings of All Time....Officially

[bunny]

Lets think about this for a second. What parameters would you pass? I'll only speak for the batting here: There's the: capability of the bowling attack (how strong/weak they were based on averages, s/r,etc.), the pitch on which the match was played (not sure how to quantify it but i'm sure we can somehow), the team's position when coming into bat, the partnership built, team's position when exited, the size of the ground, the type of ball used, the season in which the match was played in which country, the quality of the batting teammates, the ranking of the opposition during the knock, the # of matches played by the player upto the knock. These are what I can think just off the top of my head. What I would do from here, is to come up with about 20-or-so combinations with these things (I know a lot more are possible), and assign a particular knock that can be rated as the best to that combination, and judge everything else relative to that. Perhaps, you can offer a better idea.

The parameters you have defined make sense. But, if you have more training points it will help. So instead of judging just the best, take 15 or so really good innings of differing type (and rank them), and adjust the formula so that it's output is largely consistent with the ranking you have supplied. You should have a really good calculator.

[graphic23]

The parameters you have defined make sense. But' date=' if you have more training points it will help. So instead of judging just the best, take 15 or so really good innings of differing type (and rank them), and adjust the formula so that it's output is largely consistent with the ranking you have supplied. You should have a really good calculator.[/quote'] Thats even better. We should do that sometime (and I mean all of ICF).

[punjabi_khota]

You will need techniques called genetic algorithm or artificial annealing to do something like that.

[bunny]

Thats even better. We should do that sometime (and I mean all of ICF).

Defining the correct parameters and calculating them in the most sensible way will be the key. Once we have a good calculator it can be used to settle a lot of "greatness" debates.

[bunny]

You will need techniques called genetic algorithm or artificial annealing to do something like that.

Standard regression like techniques will work if the supplied parameters are good.

[The Outsider]

You'll need to run an optimization program with 20-30 variables - I doubt you'll get supercomputer time for this pursuit from any organization.:giggle:

[Rajiv]

oh boy, why can't it remain just a bat and ball game..oh why the pie charts

[bunny]

You'll need to run an optimization program with 20-30 variables - I doubt you'll get supercomputer time for this pursuit from any organization.:giggle:

Let's say I tweak my algorithm so that one of SRT's innings comes as the top. Surely SRT's agents will be ready to cough up some money for that, no? :)

[punjabi_khota]

Lets write a NSF proposal for supercomputing time at Los Almos. Shwetabh was voted 'most likely to be PI' in his highschool....let him lead.

[The Outsider]

Lets write a NSF proposal for supercomputing time at Los Almos. Shwetabh was voted 'most likely to be PI' in his highschool....let him lead.

I used to have supercomputing time at LLNL a couple of years back - you guys did not give me the idea earlier. Now the world of cricket is poorer for suffering through these crap ratings.

[Sachin_Rulez]

Some random BS list!! Just because it featured sehwag's 3 knocks ( based on some XYZ criteria)???