I have seen and worked on such ratings before. And the all-time ratings look like a joke overall. The reason is the methodology, and the final lists. The fact that you think my gripe is about specific examples like Kapil or the English batsmen of the 90s is the exact reason I didn't want to come up with examples at the first place.
This is not how you do an all-time ranking. If you see a serious effort like PEWS (I won't post a link; if you visit the list of threads started by him you'll find it there), you may realise your all-time ratings are a pedestrian effort.
LT was not a troll; he genuinely disliked the ratings to the fullest. And I support him.
I won't be so critical of your current ratings (they are better than the overall ratings, at least). But do you have examples to show how those are better than the ICC rankings? And if not, what's the point of having them?
I think you're trying to be helpful so I'll play along. What exactly in the methodology do you not like? It takes into account as many avaiable factors from a scorecard as possible to rate each innings (I'm assuming you've seen the list of factors), gets the average innings rating of a player and adds a longevity bonus scaled by era and the average rating itself (so a poor average innings player doesn't just jump up the list because he's played a lot of matches).
Is it so hard to come up with specific examples? I do think it's a work in progress, and statistical exercises will always have issues, but if you were more specific I could actually look into what you are saying.
Whether LT was a troll or not is irrelevant - for one thing he saw an early draft of the ratings so its not what you're seeing now.
I checked PEWS' started threads and couldn't find a specific ATG methodology or longevity - would appreciate it if you could post/PM a link.
The current ratings, innings ratings and career ratings are all tied together. Each innings rating affects the current rating of a player, with more recent innings given a larger weight. That current rating is a crucial part in evaluating the innings rating itself - the bowling attack's quality factor when rating a batsman's innings is based on that attack's current rating. The reason I believe my setup is better than the ICC setup is that I take the same factors + 6-7 more factors parsed from the scorecard into account (such as point of entry, match status, wickets fallen, support, closeness, team ratings etc).
AFAIK this type of comprehensive, automated, detailed-scorecard based ratings setup has not been implmented before (I've seen great ideas but with manual and limited implementation or narrow ideas with wider implementation). I realize its not perfect by any means, but I really am looking for some constructive criticism that bears in mind that this is an exercise based on data, so "x is better than y therefore this ratings is ****" is generally not helpful without more detail.
I appreciate your comments, but please be more specific - thanks.
