[Not so] Huge statistical analysis on greatest ODI batsman and bowlers ranked

[Bolo.]

I wonder how much standardization becomes flawed for players with longish careers who had different levels of form/ playing roles.

Imran for example scored about 25 runs per year in his first 7ish years. Doesn't alter his overall stats that much due to low innings numbers. But he is getting a big upward adjustment on era unless the adjustment takes into account the relative number of innings (which I am sure isnt the case here based on how high he ranks).

Probably something similar on the go for modern bats, even those without such lengthy careers.

 

[ankitj]

Interesting point that. I didn't know Imran was playing so little ODI cricket in his early years. But it also seems true for others. Very few games for Viv Richards too for example until 1983. But yeah those things can skew the outcomes. Ganguly had a break of 4 years after his debut so his benchmark period includes years when he played no cricket at all.

 

[honestbharani]

maybe break it down to years players played at least few games to make it more even? I know it is tough work but you are a bit of a coding expert, no?

 

[honestbharani]

only one spot higher off a simpler analysis, right? I think I can find enough numbers to support my hypothesis.. I am just glad it shows them that close to each other, coz I really felt it may show them farther apart.

Ankit, have you considered doing the longevity based on number of games as well as years? Some kind of average of games/year and the number of years all taken into account?

 

[ankitj]

That will favour those nations who play more ODI cricket won't it.

I am generally averse to adding voodoo variables in an analysis -- hurts intepretability of results and introduces new problems while attempting fix previous ones.

 

[ankitj]

What I like about the results of this analysis is that it rewards some of the old timers who were early pioneers of the game, likes of Abbas, Greenidge, Chappell etc. I have said before that I rate Greenidge second best ODI opener after Tendulkar and that's exactly what this result showed.

Only disappointment was Dean Jones who I would have liked to get ranked in top 10. Always rated him second best Aussie ODI batsman behind Bevan. Seeing him ranked below Sehwag feels completely wrong.

 

[honestbharani]

That will favour those nations who play more ODI cricket won't it.

I am generally averse to adding voodoo variables in an analysis -- hurts intepretability of results and introduces new problems while attempting fix previous ones.

My simple reasoning is that the more games you play, the harder it is to maintain levels of excellence. Years alone do not vouch for anything although I agree it should be a factor.

 

[honestbharani]

For instance, in your list, Ganguly benefits a lot coz he played a couple of games in the early 90s and it totally forgets the period when he could not make the side. That can be balanced somewhat by the number of games being brought into it as well.

 

[harsh.ag]

Deano was doing a Bevan at #3. That doesn't help in this exercise because, unlike Bevan, the batting average is low cause of less not outs. If you replace average with rpi, he would jump up and Bevan would go down.

 

[honestbharani]

Yeah but in ODIs RPI is also not a great measure because of the overs limit and the fact that not every batsman has the same opportunity to make runs.

 

[srbhkshk]

That will favour those nations who play more ODI cricket won't it.

I am generally averse to adding voodoo variables in an analysis -- hurts intepretability of results and introduces new problems while attempting fix previous ones.

But isn't log of years played already voodoo?

 

[stephen]

Deano was doing a Bevan at #3. That doesn't help in this exercise because, unlike Bevan, the batting average is low cause of less not outs. If you replace average with rpi, he would jump up and Bevan would go down.

How is losing your wicket equivalent to not losing your wicket? Being not out at three end when you're batting 4 or 6 is actually pretty important, particularly in an era where ball dominated bat (200 was a defensible total fort much of that era, particularly in Australia).

 

[ankitj]

Yes, we have had this average vs. RPI debate many times before. No analysis that uses RPI instead of average will be taken seriously by me ever.

 

[harsh.ag]

How is losing your wicket equivalent to not losing your wicket? Being not out at three end when you're batting 4 or 6 is actually pretty important, particularly in an era where ball dominated bat (200 was a defensible total fort much of that era, particularly in Australia).

?

Who said it's not important for lower order batsmen to not stay not out?

Just observed the basic fact that it results in a higher average than someone using the same style of play at #3.

How is scoring 37 runs per innings less important than scoring 35 (Deano vs Bevan)? Obviously you haven't said that, so I don't feel the need to ask the question.

 

[honestbharani]

Almost all of ODI batting stats are useless because it does not give us any real insight to either the opportunity that was available to the batsman, which varies by batting position and the impact it had on the match, as different stages of the game demand different styles of play etc.

 

[TheJediBrah]

Deano was doing a Bevan at #3. That doesn't help in this exercise because, unlike Bevan, the batting average is low cause of less not outs. If you replace average with rpi, he would jump up and Bevan would go down.

jfc not this again

Yes, we have had this average vs. RPI debate many times before. No analysis that uses RPI instead of average will be taken seriously by me ever.

bingo

always amazes me at the complete lack of logic people can have on this topic

 

[ankitj]

There is only one way for batting average to go up – scoring more runs.
There is only one way for batting average to go down – getting out.

Staying not out has no effect.

 

[TheJediBrah]

?

Who said it's not important for lower order batsmen to not stay not out?

Just observed the basic fact that it results in a higher average than someone using the same style of play at #3.

How is scoring 37 runs per innings less important than scoring 35 (Deano vs Bevan)? Obviously you haven't said that, so I don't feel the need to ask the question.

I'll have a crack at this though, because you do seem to be trying to think sensibly about it.

Just observed the basic fact that it results in a higher average than someone using the same style of play at #3.

This is an assumption. You're assuming that if Bevan was at 3 and Jones was at 6 then their averages would have swapped. There's no reason to think that Jones would have averaged higher playing Bevan's role than he did in his actual career. Maybe he would have only averaged 35 if he batted mostly at 6. Similarly there's no reason to think that Bevan would have averaged significantly less than he did at 6 if he batted higher. He very likely would still have averaged ~50, just with more runs and less not outs.

There's no reason to think that an average can be artificially "boosted by not outs" over any significant sample. If anything the opposite is more likely and by having more innings finished not out you've been robbed of continuing batting when you've already played yourself in. It would make more sense to say the opposite of what you're saying, that more not outs actually hurt your average (as counter-intitutive as this may seem at first glance).

How is scoring 37 runs per innings less important than scoring 35 (Deano vs Bevan)? Obviously you haven't said that, so I don't feel the need to ask the question.

This is a completely different discussion. Obviously, with 2 hypothetically identical batsmen, someone batting higher is going to have a higher RPI and seemingly "contribute more to the team". It tells you nothing that average doesn't already tell you much more accurately.

You could argue that "RPI is a more accurate indication of the runs a player scored for their team and hence the impact they had" but once again, that's not a measure of how good the player was, or how well they batted. It's just that they batted higher in the order

 

[harsh.ag]

There is only one way for batting average to go up – scoring more runs.
There is only one way for batting average to go down – getting out.

Staying not out has no effect.

The primacy of batting average over RPI is not sacrosanct. Except for a fanatic.

If Bevan was batting at #3, he would not magically have still averaged 50+. He would have been ranked lower in this statistical exercise.

 

[harsh.ag]

I'll have a crack at this though, because you do seem to be trying to think sensibly about it.

Oh thank you milord, harrummmphh

Similarly there's no reason to think that Bevan would have averaged significantly less than he did at 6 if he batted higher. He very likely would still have averaged ~50, just with more runs and less not outs.

This is fantasy fanfic.