You should post your ideas on the forum rather than getting people to go on your website in order to take part in the discussion. That way it's easier for cw members and you won't get done for advertising your website without asking.
I wasn't aware that "advertising" the site was an issue. I expect feedback here after people check the site - it's not really possible to explain what it does on a forum post without linking it.
Viriya, this is a nice website. My congratulations. The methodology seems to consider many pertinent factors, so bases covered there.
Given that, could inquire as to how did Lara's 153 and 277 not come out to be the top two innings? Same question for Sachin's 136 and 114? It seems your methodology gives much more weightage to volume of runs scored than anything else, as the most highly rated innings usually are the highest scoring ones in both cases.
Welcome to the forum
~ Do you think I care for you so little that betraying me would make a difference ~
Having gone through more of the site, I must say I find the Test All Round Performances list very satisfying.
The test batsmen (overall career) list however is bemusing to say the least. I hope you realize that any methodology which places George Headley at #70, Jack Hobbs at #33 and Len Hutton at #37 (let alone the rest- Mahela at #11 for god's sake!) is not a very good methodology. Perhaps you should tweak it a bit.
Some reasons (scorecard):
- Even though it was against the Aussies (with McGrath, Warne, MacGill and Gillespie in the bowling attack), this was 1999 - only McGrath was at his best. Warne was in his mid-career slump, and both MacGill and Gillespie were relatively new to international cricket so their current ratings were low. So the bowling quality was only great in hindsight when you think in terms of numbers - there was no way for the ratings to "know" that MacGill and Gillespie were going to be good or that Warne was going to come back strong a few years later.
- Lara comes into the crease at 78/3, and even though they are in trouble, it could be much worse.
- Even then, this innings is admittedly great, and it only just misses the top 100 with a rating of 2643.
I do think though that it deserves to be in the top 100 - the ratings system is still a work-in-progress so I will probably tweak the match status parameter to push it higher.
On why the system seems to factor volume of runs significantly higher than anything else - it's because the same formula is used to rate all innings and it's average is used to calculate career ratings. If I did not weight runs as highly as it is now, generally accepted great batsmen with 50+ averages would figure below some batsmen with ~40 averages. So after some though I decided that pure runs scored has to be considered more significantly.
Last edited by viriya; 27-04-2014 at 12:11 PM.
What a waste of ink.
With George Headley, it's just due to the fact that he didn't get to play enough matches - when rating careers there is no way around penalizing short careers. Otherwise, a player like Philander who has started his career with a 2-year purple patch would figure higher than I think warranted.
On Hobbs and Hutton (and to a lesser extent Hammond), I myself was bemused as to why they were figured that low when the careers were rated. After some analysis I found that the reason for this was because during their careers, there were no real great bowling attacks - in fact it was probably the worst bowling decades of test cricket. They still are rated decently - partly due to my change to consider runs scored to be a much more significant factor (as I mentioned in the previous post). Note that this list is generated competely bottom-up - every singles innings of a batsman is rated on its own merits and the batsmen overall average has no bearing on his career rating. The career rating is basically his average innings rating with a bonus for longevity.
On Mahela being #11, I realize he does not have a great record overseas (his home performances would not get the away bonus factor), but his record is stellar and he has had some great innings. I think his overseas performances are unfairly highlighted - Sri Lanka does not get to tour Eng, SA and Aus that often, so he doesn't get to "fix" his record easily either. I can agree that he might not be worthy of a #11 spot, but it would be hard to get him out of the top 20 even if I tried.
Thanks again for the feedback. Please feel free to explore the site and comment - you can also mail firstname.lastname@example.org.
- For openers, they get a default point of entry factor score, but otherwise yes they never get to come in during a crisis. But there is a flip side to this, they get to bat through the innings and maximize the Wkts at Crease factor - so I think it evens out in the long run since both factors are similarly weighted.
- Hobbs and Sutcliffe were the 2 50+ avg batsmen of their era. This is taken into account to a certain extent because this would mean that they got more of a % of the team's total runs - which is also a relatively significant factor.
I plan to implement some of your suggestions (increase match status factor weight for Lara's 153*, reduce short test batting career penalty for players like Headley, Graeme Pollock).
I don't expect a ratings system to make everyone happy - that is pretty much an impossible task , but it's meant to be a starting point for discussion - not as an answer to "who is the greatest" with no uncertainty.
Last edited by viriya; 27-04-2014 at 12:38 PM.
There is no possible rationale for Shaun Pollock to be rated above Imran Khan as a test all rounder. And how the hell can you have Shakib Al Hasan listed as a greater all rounder than Imran? And Kumble as a greater bowler than Lillee, Donald and Imran (and Akram and Holding)
I'm a bit perplexed here tbh. You have Garner ranked 45th and Davidson ranked 43rd, when both are top 20 for mine. And Kumble is 8th. WTF?
It all hurts my head...
This is utter garbage. Cricinfo should start charging people for using statsguru.
Kumble being #8 is a combination of higher batting quality during his playing days, higher wkts/match (4.8 vs 4.5 and 4), and much longer career longevity. On him being rated higher than Lillee, Donald and Imran - the rating difference between them is not very significant, but it's obviously debatable. Garner and Davidson also are affected by having relatively short careers.
On Hobbs and Hutton - I didn't mean to say they were peers, but during their main playing decades the best bowlers they faced did not have great careers (compared to say the 80s or the 90s). Arthur Mailey and Jack Gregory were the top Australian bowlers of the 1920s that Hobbs faced and for Hutton (1940s), it was Toshack, Wright and Cowie. You can use the drop-down to check decades here: cricrate | Best Test Bowling Careers. The only two bowlers with great careers during 1920-1949 were Grimmett and O'Reilly, and they played the majority of their games in the 1930s (vs Hammond mainly).
Please realize that this is a statistical exercise. I'm attempting to rate players based solely on scorecards - I might think that X is better than Y because of A, B and C, but if I can't make it plain from the numbers it's not useful.
Also, I'm not claiming that this list is some be-all and end-all of course, actionable feedback that makes sense will be used to improve on the system constantly.
There are currently 1 users browsing this thread. (0 members and 1 guests)