A bit of fun - standardised ODI performances

[Top_Cat]

I'm quite suprised with Fairbrother's and Basit Ali's high positions.

Prince 2008 games haven't been factored in yet have they?

I'm not too surprised, with regards Fairbrother. He was one of England's best ODI batsmen for a while, always chipping in with good finishing knocks. Good to see, in context, his worth.

 

[Prince EWS]

I'm quite suprised with Fairbrother's and Basit Ali's high positions.

Prince 2008 games haven't been factored in yet have they?

Yep they have.

 

[Migara]

No standardization for number of matches / time period played?

 

[NUFAN]

No standardization for number of matches / time period played?

Time period play has a huge standardization. I think you should re read the first post which advises you that each year is taken into account and it clearly shows the adjusted SR for each batsmen.

Number of matches, min: 30 innings.

 

[weldone]

Not a bad initiative...Specially I agree with the decision of considering data of the top 7 only...

But I have some reservations about this approach...Firstly, again no consideration (weightage) assigned to longetivity. Achieving something after 30 matches (or 2 years) is different from achieving the same thing after 300 matches (or 20 years)...

Secondly, Zimbabwe for a long part in the 90s was a fighting team unlike Zimbabwe or Bangladesh of these days. If you are not considering Zimbabwe of the 90s, then you can't consider Sri Lanka in their early days...

Another thing which is not a complaint but a suggestion for improvement is that for normalizing data of one batsman, data against bowlers of his team shouldn't be considered. For example, the compiled data is against all the bowlers (even the West Indian bowlers). This data is used to normalize Viv Richards' data, though he never faced Marshall, Holding, Garner etc. I suggest when normalizing Richards' data, only data of other batsmen against non-West Indian bowlers should be considered. Same for all other batsmen.

Anyways, for a first-time attempt, this was OK.

 

[Prince EWS]

Firstly, again no consideration (weightage) assigned to longetivity. Achieving something after 30 matches (or 2 years) is different from achieving the same thing after 300 matches (or 20 years)

Indeed, possibly something I could implement later, particularly if I could find such data easily. As it stands it merely standardises the performances across time; it merely attempts to minimise the largest, most easily identifiable variable in ODI batting stats - that is the era in which the players played and the expectations of scoring at the time.

Another thing which is not a complaint but a suggestion for improvement is that for normalizing data of one batsman, data against bowlers of his team shouldn't be considered. For example, the compiled data is against all the bowlers (even the West Indian bowlers). This data is used to normalize Viv Richards' data, though he never faced Marshall, Holding, Garner etc. I suggest when normalizing Richards' data, only data of other batsmen against non-West Indian bowlers should be considered. Same for all other batsmen.

That's a great suggestion actually. It would require a lot more data collection and more intense programming but in reality it would probably be worth it in the interests of consistent data.

Regarding Zimbabwe, I agree in a way - for a while they were quite competitive indeed. Just as competitive as early SL teams and arguably even as much as the current WI mob, but I thought including certain years teams were good enough in and not others would be cherry-picking the data just a little too much. In the end I don't think excluding all matches against Zimbabwe is going to distort the data too much if at all, especially at the top end; certainly less than it would if I included all matches against them.

 

[Prince EWS]

Time period play has a huge standardization. I think you should re read the first post which advises you that each year is taken into account and it clearly shows the adjusted SR for each batsmen.

Indeed. If he's talking about which specific years players played in, that's the main thing that actually is being considered.

I think he's talking about the length of one's career though. As weldone suggested - the standardisation of longevity. I can actually think of a way to do this fairly accurately and may implement it later. It's important to not break too far away from the raw data though, IMO.

 

[Aamir Alauddin]

[B]				Avg	S.Avg	SR	S.SR	S.Per[/B]

1	IVA Richards (WI)	48.36	49.47	90.59	103.24	51.07
2	Zaheer Abbas (Pak)	47.85	49.55	85.04	97.06	48.10
3	MEK Hussey (Aus)	54.71	49.60	85.63	81.36	40.35
4	MG Bevan (Aus)		53.89	52.18	73.62	75.68	39.49
5	CH Lloyd (WI)		39.54	41.51	81.22	93.91	38.98
6	GS Chappell (Aus)	40.19	42.65	75.71	89.07	37.99
7	KP Pietersen (Eng)	48.46	43.76	87.28	82.79	36.23
8	DM Jones (Aus)		44.62	44.51	72.56	81.05	36.08
9	MS Dhoni (India)	46.38	41.90	90.57	85.79	35.94
10	A Symonds (Aus)		41.06	38.16	93.64	91.63	34.97

I have always been a great fan of Zaheer Abbas. He seems to be the most under rated player if we look at his performance and the way people rate him in these threads.

 

[pasag]

Prince has turned the thread topic into an article for CW, feel free to leave a comment there.

http://www.cricketweb.net/blog/features/41.php

 

[weldone]

Indeed, possibly something I could implement later, particularly if I could find such data easily. As it stands it merely standardises the performances across time; it merely attempts to minimise the largest, most easily identifiable variable in ODI batting stats - that is the era in which the players played and the expectations of scoring at the time.



That's a great suggestion actually. It would require a lot more data collection and more intense programming but in reality it would probably be worth it in the interests of consistent data.

Regarding Zimbabwe, I agree in a way - for a while they were quite competitive indeed. Just as competitive as early SL teams and arguably even as much as the current WI mob, but I thought including certain years teams were good enough in and not others would be cherry-picking the data just a little too much. In the end I don't think excluding all matches against Zimbabwe is going to distort the data too much if at all, especially at the top end; certainly less than it would if I included all matches against them.

Good that you understood the shortcomings of your good initiative...Yes, the issue of not considering the data of bowlers of the same team requires a lot more effort for data collection...

Once I did a similar study for test batsmen...For including the effect of longetivity, I used an empirical formula where I simply added (no. of years played * 1.1) to the points and it led to a believable ranking...The first 5 positions were taken by
Bradman,
Sobers,
Hobbs,
Hammond and
Tendulkar
respectively...I believe in your case also, if you add (1.1 * no. of years played) to the existing points, it'll lead to a more believable result...

The other adjustment (of not considering bowlers of one's own team) may come later as that is more cumbersome...

 

[Zinzan]

Top effort Prince, Looking forward to seeing the bowling results