***Official*** Tendulkar vs Ponting Thread

[ankitj]

He just seems to have bunged down that quality measure matrix by opponent/venue for each batsman, and **** knows how he uses that to come up with the effective average/runs. Such as it is, I'll remain skeptical.

I for one know it.

He has explained it. One would have to read carefully as the explanation is rather brief

 

[ankitj]

Quoting from the article:


The parameters used for the analysis are quite basic.
1. The bowling average for each opponent (in matches involving the player) is taken into consideration for home and away games.
2. The match average for all the matches is used to measure the difficulty level encountered. In matches involving Zimbabwe and Bangladesh, I do not consider the batting average of the minnows as the figure can skew the numbers badly. In these cases, the measure is purely the batting average of the other team.


The base value to measure the quality of an innings is calculated as the geometric mean (square root of the product) of the batting average (30.61) and bowling average (32.31) since Jan 1 1940. The quality index value obtained is 31.44. For each batsman, the similar values are calculated and measured with respect to the base value to obtain the accurate or effective average. For example if the batting and bowling average are 30 and 32 respectively , then the geometric mean is 30.98 and the quality factor is obtained by dividing the base value by the mean which yields 1.0147.

 

[vcs]

Ah, OK I get it. Then as Dasa pointed out earlier, you get wayy too much benefit for playing for a weak team and penalised for being part of a strong lineup.

 

[Top_Cat]

Said it a million times but the number of runs scored in a given knock is a really blunt measure of how good a player is/quality of the knock/etc, should be used to record what happened and limited to that or very broad trends or conclusions. When you start using what is already a poor measure to support some of the contentions in that article, you're venturing into very dodgy territory. Just using the runs/mean of absolutely everyone who's played Test cricket is questionable and he's a bit economical with the explanations of his methods, more detail please.

The number of runs scored and arithmetic average (he doesn't explain why he uses the geometric average either; where's the justification for it? Is the distribution exponential?) do very broadly correlate to grouping players into groups of who's better but this sort of fine-grained detail is potentially meaningless when you consider the measurement errors involved.

Massive question marks over this article, tbh. I have heaps I'd ask the writer myself if I could.

 

[honestbharani]

@PrienceEWS - Could you branch it off into a new thread. Might be worthy of good discussion.

@vcs - The methodology is sound tbh. Of course there would be some parameters unaccounted for. But this analysis at least removes one or two flaws from conventional averages.

The real issue with any of these stat analysis is that while you go on to remove 1 or 2 flaws in the "raw" stats (if we could call it that) you only end up creating even more bigger ones... I wonder what the effective averages of blokes like Sachin would be in the 90s, for instance? Lets face it, there is no THE best way to sort out stats in crcket... That's that.

 

[ankitj]

I agree with some of the flaws you guys are pointing out. I also have a pet theory of my own that bowling performances lend themselves better to statistical analysis than batting performances. Will elaborate on that some other time.

 

[Prince EWS]

I agree with some of the flaws you guys are pointing out. I also have a pet theory of my own that bowling performances lend themselves better to statistical analysis than batting performances. Will elaborate on that some other time.

If anything I think the opposite. Uppercut certainly does; I'd really like to see you two debate that. That sounds so queer I know.

 

[Teja.]

I don't think this is what you are talking about but IMHO, There is a much higher chance of a bowler bowling well and not having good figures than a batsman batting well and not having a good score.

 

[Prince EWS]

I don't think this is what you are talking about but IMHO, There is a much higher chance of a bowler bowling well and not having good figures than a batsman batting well and not having a good score.

Exactly. And he's suggesting the opposite.

 

[marc71178]

Ah, OK I get it. Then as Dasa pointed out earlier, you get wayy too much benefit for playing for a weak team and penalised for being part of a strong lineup.

So how come Dravid didn't get penalised much then?

 

[Top_Cat]

Yeah the measures for bowling would likely have a lower internal validity but I don't think it'd be by much and that's only one aspect to be aware of in using them.

 

[Sanz]

By not talking about them...

What is there to talk ?

 

[ankitj]

I don't think this is what you are talking about but IMHO, There is a much higher chance of a bowler bowling well and not having good figures than a batsman batting well and not having a good score.

OK. My reasons are mainly two:

One, a bowler gets many chances to rectify his errors, rethink his strategy etc in one innings. A batsman gets to make only one error per inning. So a batsman getting a duck has a better case for being considered unlucky than a bowler sending down 30 overs without a wicket.

Two, the pressure factor affects the bowlers much less. We find ourselves discussing great knocks under pressure all the time while rarely discussing spells under pressure. Therefore "not all runs against the same bowlers are equal" is a stronger argument than "not all wickets of same batsmen are equal" argument.

Of course those arguments have merit on average and in general. Exceptions may be pointed out but they will just be exceptions in IMO.

 

[weldone]

OK. My reasons are mainly two:

One, a bowler gets many chances to rectify his errors, rethink his strategy etc in one innings. A batsman gets to make only one error per inning. So a batsman getting a duck has a better case for being considered unlucky than a bowler sending down 30 overs without a wicket.

Two, the pressure factor affects the bowlers much less. We find ourselves discussing great knocks under pressure all the time while rarely discussing spells under pressure. Therefore "not all runs against the same bowlers are equal" is a stronger argument than "not all wickets of same batsmen are equal" argument.

Of course those arguments have merit on average and in general. Exceptions may be pointed out but they will just be exceptions in IMO.

The first point is a very valid one, from statistical point of view.

When we are talking about a batsman's (let's say Sachin's) batting average, it is the average of 250 data points or so. But when we are talking about a bowler's (say Muralitharan's) average, we are talking about the average of many more (in this case, 700) data points. So, the second one is a more reliable measure than the first one.

 

[kingjulian]

This is an analysis based on their batting in noughties. So i'm not really surprised Sachin Tendulkar has figured as low as he has. Because for 3 years he was clearly a pale shadow of himself.
He was India's best batsman up to 2004, but till mid 2008, there were at least two batsmen in the Indian team who were batting better than him.

I'm more surprised by Ponting going lower in the analysis.


Also here is one more stato analysing the numbers and trying to enlighten us.

Intellections along the course of my journey: Test Cricket's Virtuoso Batsmen

 

[ankitj]

Also here is one more stato analysing the numbers and trying to enlighten us.

Intellections along the course of my journey: Test Cricket's Virtuoso Batsmen

Analyses like these choose arbitrary parameters and assign arbitrary weightages and are therefore much less logically sound than the kind earlier posted IMO.

 

[vcs]

ok. My reasons are mainly two:

One, a bowler gets many chances to rectify his errors, rethink his strategy etc in one innings. A batsman gets to make only one error per inning. So a batsman getting a duck has a better case for being considered unlucky than a bowler sending down 30 overs without a wicket.

Two, the pressure factor affects the bowlers much less. We find ourselves discussing great knocks under pressure all the time while rarely discussing spells under pressure. Therefore "not all runs against the same bowlers are equal" is a stronger argument than "not all wickets of same batsmen are equal" argument.

Of course those arguments have merit on average and in general. Exceptions may be pointed out but they will just be exceptions in imo.

the first point is a very valid one, from statistical point of view.

When we are talking about a batsman's (let's say sachin's) batting average, it is the average of 250 data points or so. But when we are talking about a bowler's (say muralitharan's) average, we are talking about the average of many more (in this case, 700) data points. So, the second one is a more reliable measure than the first one.

awta.

 

[Top_Cat]

haha, love analyses like that. All the transpositions in the world will never be able to escape the original problem of poor measures. Based on data currently collected, it's an intractable problem, tbh.

Got a few approaches I'd personally try but using runs/averages would not be one of them if I wanted to be serious about it.

 

[MrIncredible]

Analyses like these choose arbitrary parameters and assign arbitrary weightages and are therefore much less logically sound than the kind earlier posted IMO.

After reading thru it I came to the same conclusion.

Analysis also makes a serious error by rating for example Gavaskar for performances vs

WI (because they were supposedly forgeting they were the best attack) forgetting that

Sunil didnt always come up against the best WI attack. (Ditto Tendulkar and others).

 

[honestbharani]

This is an analysis based on their batting in noughties. So i'm not really surprised Sachin Tendulkar has figured as low as he has. Because for 3 years he was clearly a pale shadow of himself.
He was India's best batsman up to 2004, but till mid 2008, there were at least two batsmen in the Indian team who were batting better than him.

I'm more surprised by Ponting going lower in the analysis.


Also here is one more stato analysing the numbers and trying to enlighten us.

Intellections along the course of my journey: Test Cricket's Virtuoso Batsmen

Something about that article sounds so so much familiar..