The flaw in batting averages

[Migara]

Both have similar R2 values (0.36 and 0.4) which are very healthy indeed.

When The longevity index taken as number of years, the top 20 looks like this

[B]Pos	Player			Risk Index[/B]
1	IVA Richards (WI)	35.7
2	SR Tendulkar (India)	35.4
3	Javed Miandad (Pak)	29.7
4	ST Jayasuriya (SL)	28.0
5	PA de Silva (SL)	27.1
6	BC Lara (WI)		26.7
7	ML Hayden (Aus)		25.2
8	CG Greenidge (WI)	25.2
9	Zaheer Abbas (Pak)	24.5
10	GA Gooch (Eng)		23.7
11	A Ranatunga (SL)	23.3
12	SC Ganguly (India)	22.4
13	Inzamam-ul-Haq (Pak)	22.3
14	Saeed Anwar (Pak)	22.3
15	DL Haynes (WI)		21.9
16	RT Ponting (Aus)	21.7
17	Imran Khan (Pak)	21.5
18	Saleem Malik (Pak)	21.4
19	ME Waugh (Aus)		21.4
20	MS Atapattu (SL)	20.9

 

[Migara]

When the number of innigs batted taken as the independent variable for the longevity index the top 20 looks like this.

[b]Pos	Player			Risk Index[/b]
1	SR Tendulkar (India)	52.1
2	ST Jayasuriya (SL)	40.7
3	AC Gilchrist (Aus)	38.2
4	RT Ponting (Aus)	37.8
5	Inzamam-ul-Haq (Pak)	35.9
6	BC Lara (WI)		35.6
7	SC Ganguly (India)	34.8
8	IVA Richards (WI)	33.1
9	PA de Silva (SL)	33.0
10	Saeed Anwar (Pak)	32.0
11	R Dravid (India)	31.6
12	M Azharuddin (India)	31.6
13	JH Kallis (SA)		31.1
14	Mohammad Yousuf Pak)	31.0
15	ME Waugh (Aus)		29.9
16	MG Bevan (Aus)		28.7
17	A Ranatunga (SL)	28.5
18	HH Gibbs (SA)		27.5
19	SR Waugh (Aus)		27.0
20	DL Haynes (WI)		26.9

 

[Migara]

The combined risk Indicis will give the top 20 ODI batsmen as following

[B]Pos	Player			Combined Risk Index[/B]
1	SR Tendulkar (India)	43.7
2	IVA Richards (WI)	34.4
3	ST Jayasuriya (SL)	34.3
4	BC Lara (WI)		31.2
5	PA de Silva (SL)	30.1
6	RT Ponting (Aus)	29.7
7	AC Gilchrist (Aus)	29.5
8	Inzamam-ul-Haq /Pak)	29.1
9	SC Ganguly (India)	28.6
10	Javed Miandad (Pak)	27.9
11	Saeed Anwar (Pak)	27.1
12	A Ranatunga (SL)	25.9
13	M Azharuddin (India)	25.8
14	ME Waugh (Aus)		25.7
15	JH Kallis (SA)		24.4
16	DL Haynes (WI)		24.4
17	ML Hayden (Aus)		24.3
18	Saleem Malik (Pak)	23.8
19	MS Atapattu (SL)	23.4
20	R Dravid (India)	23.2

 

[weldone]

The analysis you did is very much flawed, I must say...averages are showing a general increasing trend at first with increasing number of innings played (or increasing years) because of a simple reason - generally better players play more number of matches and for more years...In no way from this regression analysis a conclusion can be drawn that as you play more generally your average first increases in the way shown by the line and then decreases in the way shown by the line...This is very very much flawed.

However, a much better (though not completely unbiased) method will be only considering those batsmen who played for a good period or more (say 5 years), plot the data for them and then from that line gauge how performance generally varies with more number of years played (not number of innings, never). What that will ensure is that the quality of batsmen for whom data are analysed are comparable...[Or in other words, the data of Yusuf Pathan and Tendulkar are not on the same line to make it look slope upwards with number of innings played].

An even better procedure (even now not completely unbiased) will be considering only 8-10 great batsmen who have played good number of innings, then plot their averages after every innings played [ by this I mean not their overall average only; but their averages after 1 inning, 2 innings, 3 innings etc. etc. to derive a pattern]...For differnt batsmen, the pattern will be different, but a gross pattern can be arrived at...

Having said all these, such an analysis shouldn't be done because it will be highly subjective...Much better will be an empirical formula where you add a component as (number of years played * a constant) to the score...

Another very very big flaw in your procedure is that you are passing the line through the origin...The line will not start from x = 0 but just after that...Or in other words, it will be an open curve, not a closed one and it's not necessary that it has to have an upward slope in the beginning..

All in all, I am sorry to say, but it was a meaningless excercise. The reason the results look better than your first excercise is that you have considered the average and sr of other players of same era against same oppositions, and not because your handling of longetivity in the way you did...

 

[Migara]

The analysis you did is very much flawed, I must say...averages are showing a general increasing trend at first with increasing number of innings played (or increasing years) because of a simple reason - generally better players play more number of matches and for more years...In no way from this regression analysis a conclusion can be drawn that as you play more generally your average first increases in the way shown by the line and then decreases in the way shown by the line...This is very very much flawed.

If you consider people who have played for long periods, they are very few. WHy most cannot play for 19 years as Tendulkar, Jayasuriya, Inzamam or De Silva? That's because after some time their performances drop and they are thrown out of the side. Even above four batsmen showed that trend, but not to a great degree. What would happen to Viv Richards if he played 350 matches?According to the downhill curve he was going through, he would have reduced his average still lower than 47. The whole idea is batsmen who can keep the average for 15 years on 40 is better than that who can keep it for 5 years. And naturally not so great players will wane off more quickly than greats.

 

[Migara]

However, a much better (though not completely unbiased) method will be only considering those batsmen who played for a good period or more (say 5 years), plot the data for them and then from that line gauge how performance generally varies with more number of years played (not number of innings, never). What that will ensure is that the quality of batsmen for whom data are analysed are comparable...[Or in other words, the data of Yusuf Pathan and Tendulkar are not on the same line to make it look slope upwards with number of innings played].

Yes,that will be a good excercise. But then you'll have to drop all the batsmen who played less than for 5 years. Like Pietersen and Hussey. Even if it is analysed in such a way, I a sure that people with longest careers will come up in front.

 

[Migara]

An even better procedure (even now not completely unbiased) will be considering only 8-10 great batsmen who have played good number of innings, then plot their averages after every innings played [ by this I mean not their overall average only; but their averages after 1 inning, 2 innings, 3 innings etc. etc. to derive a pattern]...For differnt batsmen, the pattern will be different, but a gross pattern can be arrived at...

That pattern will be for great batsmen. Not for general ones. As I said, ordinary player go down the slippery slope faster than legends. This will be a total GIGO.

 

[Migara]

Having said all these, such an analysis shouldn't be done because it will be highly subjective...Much better will be an empirical formula where you add a component as (number of years played * a constant) to the score...

That is what I have done already. Longetivity factor is that. That does not need to be a linear one.

 

[Migara]

Another very very big flaw in your procedure is that you are passing the line through the origin...The line will not start from x = 0 but just after that...Or in other words, it will be an open curve, not a closed one and it's not necessary that it has to have an upward slope in the beginning..

How could you determine the intercept? Is there any meaning to it saying in ) innigs a batsman scores -0.4 runs?
If the trend line is not forced through the origin,still it has a parabolic shape (less marked than this one), but still it shows with time performances tend to wane.

 

[Migara]

All in all, I am sorry to say, but it was a meaningless excercise. The reason the results look better than your first excercise is that you have considered the average and sr of other players of same era against same oppositions, and not because your handling of longetivity in the way you did.

You are entitled to your opinion. But the best would be to analyse it in your own method and to show results.

 

[weldone]

The whole idea is batsmen who can keep the average for 15 years on 40 is better than that who can keep it for 5 years

Exactly the point I am making...You are not getting it...Let me make it simpler...Say 300 batsmen have played for 1 year and 10 batsmen have played for 12 years...What you are doing is that putting them on the same line to show how performances wane...As a result you are comparing 4 years of Dhoni with 15 years of Richards...What I say is that performances will wane with more years even more than what your curve suggests...Simply because right tail is artificially thickened with the presence of better batsmen...

 

[weldone]

Yes,that will be a good excercise. But then you'll have to drop all the batsmen who played less than for 5 years. Like Pietersen and Hussey. Even if it is analysed in such a way, I a sure that people with longest careers will come up in front.

Not necessarily; I am almost certain that Richards will take the first place and by a considerable margin followed by Sachin...

 

[weldone]

That pattern will be for great batsmen. Not for general ones. As I said, ordinary player go down the slippery slope faster than legends. This will be a total GIGO.

That pattern will be for batsmen of similar calibre...You can't say Dhoni averages 45 after 4 years and DeSilva 35 after 15 years hence average decreases by 10 in 11 years...You have to compare Dhoni with Dhoni and DeSilva with DeSilva to know how averages wane...Hope this time I've made it simpler...

 

[Prince EWS]

That pattern will be for batsmen of similar calibre...You can't say Dhoni averages 45 after 4 years and DeSilva 35 after 15 years hence average decreases by 10 in 11 years...You have to compare Dhoni with Dhoni and DeSilva with DeSilva to know how averages wane...Hope this time I've made it simpler...

Indeed. Migara's comparing the global average of batsmen after four years with the average of a select few after fifteen years. The data is distorted by the fact that Matthew Elliott for example didn't played for fifteen years - De Silva's average after fifteen years should be compared to what he averaged after four, not what the average of everyone averaged after four.

Wow, I just realised I explained it the same way you did. It really makes a lot more sense in one's head.

 

[weldone]

That is what I have done already. Longetivity factor is that. That does not need to be a linear one.

That's why I said 'empirical' ... However, that doesn't need to pass through (0,0) either ... The initial huge upward motion of the line is what I am worried about...Just see the difference between the point after 1 inning and after 20 innings in your line and you'll understand that it is improbable as well as impossible...According to your line a very very low average after 1 or 2 innings is very normal (as low as 1 or 5) which is illogical...The line comes like that only because you have considered many crap players who played 1-10 ODIs.

 

[weldone]

How could you determine the intercept? Is there any meaning to it saying in ) innigs a batsman scores -0.4 runs?
If the trend line is not forced through the origin,still it has a parabolic shape (less marked than this one), but still it shows with time performances tend to wane.

The intercept should be positive and not negative...The slope should almost always be negative because it is of more credit to play for more years than it is to play for less years with same performance...

 

[weldone]

You are entitled to your opinion. But the best would be to analyse it in your own method and to show results.

I have already given my method with slight empirical adjustment...I don't have enough time for calculation right now; but when I get time I might do it...However, as you wanted results I performed the same analysis with test batsmen (not considering sr of course) some months ago and top 5 positions were taken by Bradman, Sobers, Hobbs, Hammond and Tendulkar respectively...

 

[Migara]

Exactly the point I am making...You are not getting it...Let me make it simpler...Say 300 batsmen have played for 1 year and 10 batsmen have played for 12 years...What you are doing is that putting them on the same line to show how performances wane...As a result you are comparing 4 years of Dhoni with 15 years of Richards...What I say is that performances will wane with more years even more than what your curve suggests...Simply because right tail is artificially thickened with the presence of better batsmen...

Then, the point I highlight will be firther strengthened. I always used to think a legend in ODIs should have a long long career. As you suggest, if the wane is still more for a typical batsman, adjusted averages of People like Miandad, Tendulkar, Jayasuriya, De Silva and Inzi will further jump up, and further getting them up. That's why I used the number of innigs also to dampen the effect.

 

[Migara]

Not necessarily; I am almost certain that Richards will take the first place and by a considerable margin followed by Sachin...

 

[Prince EWS]

That's why I said 'empirical' ... However, that doesn't need to pass through (0,0) either ... The initial huge upward motion of the line is what I am worried about...Just see the difference between the point after 1 inning and after 20 innings in your line and you'll understand that it is improbable as well as impossible...According to your line a very very low average after 1 or 2 innings is very normal (as low as 1 or 5) which is illogical...The line comes like that only because you have considered many crap players who played 1-10 ODIs.

Indeed. The conclusion assumes that everyone plays the same exorbitant number of innings. To correctly measure how a player improves or declines over time, only the data of the players who have actually survived over that time should be considered.

To use a crude example, if there are only two batsmen in history - one who plays 100 games and scores 20 in all of them, and one who plays 50 games and scores 40 in all of them - the global average after 50 games is 30, and the global average after 51 games is 20. It'd be silly to draw the conclusion that players typically decline extremely sharply at game 51, and the Migara's data uses the same flawed logic on a bigger scale.