Sehwag, the closest thing to Bradman

[hendrix]

Neither method is appropriate.

Just create an average of runs scored per innings, not per wicket. That solves both issues.

 

[viriya]

Neither method is appropriate.

Just create an average of runs scored per innings, not per wicket. That solves both issues.

Top 5 becomes (ignoring <30 innings):
Player Innings Avg50+Score
DG Bradman*(Aus) 42 150
WR Hammond*(Eng) 46 115
BC Lara*(WI) 82 113
V Sehwag*(ICC/India) 55 113
MS Atapattu*(SL) 33 113

Not significantly different.

 

[Spark]

I'm trying to find the average 50+ score a batsman can get.. if I include not outs every 50* score would be unfair on the batsmen. Only considering the out innings shows the actual high score potential of the player.

Wha? I'm interested in what you're trying to do but this be some cray cray logic here. If you omit the not out innings then you're artificially reducing the apparent highest score potential. I mean, under this analysis you'd not be counting Lara's 401* - which would kind of invalidate the whole exercise given what you're trying to measure, surely?

FTR I've long been interested in a very similar measure - average score in innings > 25, as a measure of conversion, because I had a hunch that certain players (well, Clarke, basically) were especially hard to get out once set. Might want to look into it

 

[viriya]

Wha? I'm interested in what you're trying to do but this be some cray cray logic here. If you omit the not out innings then you're artificially reducing the apparent highest score potential. I mean, under this analysis you'd not be counting Lara's 401* - which would kind of invalidate the whole exercise given what you're trying to measure, surely?

Check the posts after.

 

[Spark]

I don't see how they address my point. You still have to count not out scores just the same because you don't know if 50* could have been 350 had there been sufficient time.

There's a sense that you're manipulating the method to get the result you want, tbh. If Chanderpaul slots in at #8 when you don't mess around with the rules, then good luck to him.

 

[Spark]

Batting records | Test matches | Cricinfo Statsguru | ESPN Cricinfo

This is the sort of thing I was talking about. Obviously this is very rough but >80 seemed to represent pretty good conversion and >90 was excellent. >100 is slightly scary IMO. Of course, on its own it's obviously limited but it seems to correlate pretty well with that obvious measure of conversion - 50s:100s ratio, with the exception of Chanderpaul which I suspect I can explain.

 

[viriya]

I don't see how they address my point. You still have to count not out scores just the same because you don't know if 50* could have been 350 had there been sufficient time.

There's a sense that you're manipulating the method to get the result you want, tbh. If Chanderpaul slots in at #8 when you don't mess around with the rules, then good luck to him.

Chanderpaul has a highest score of 203* with just 2 double hundreds in 263 innings - if he is at #8 in a highest average 50+ score list then I don't think it's crazy to think he shouldn't be there. I'm not suggesting that Chanderpaul couldn't have made big hundreds - it's just that he didn't. If I'm trying to find out the batsmen who have really cashed in after getting to 50 Chanderpaul doesn't belong in that list.

Not outs skew the whole analysis without giving any insight - if you consider only dismissed innings or average runs/innings (with not outs) the results are similar as shown above..

I feel like the title of the thread made people believe that I thought Sehwag is somehow the second coming of Bradman - only hendrix seems to have understood my point.

 

[Spark]

Chanderpaul has a highest score of 203* with just 2 double hundreds in 263 innings - if he is at #8 in a highest average 50+ score list then I don't think it's crazy to think he shouldn't be there. I'm not suggesting that Chanderpaul couldn't have made big hundreds - it's just that he didn't. If I'm trying to find out the batsmen who have really cashed in after getting to 50 Chanderpaul doesn't belong in that list.

Not outs skew the whole analysis without giving any insight - if you either consider only dismissed innings or average runs/innings (with not outs) the results are similar..

I feel like the title of the thread made people believe that I thought Sehwag is somehow the second coming of Bradman - only hendrix seems to have understood my point.

But you can't guarantee that he wouldn't have made plenty of big scores given that he seemed not to lose his wicket all that often after getting set, no? There are obviously clear caveats that you have to place on his results given that he's always batted kind of low in a weak order with a weak tail, but to just arbitrarily excise all innings where the batsmen had the good fortune of not nicking off is a bit... weird. Far better to just look at the "raw" results and then try to interpret rather than forcing that into your statistical machine itself.

 

[viriya]

But you can't guarantee that he wouldn't have made plenty of big scores given that he seemed not to lose his wicket all that often after getting set, no? There are obviously clear caveats that you have to place on his results given that he's always batted kind of low in a weak order with a weak tail, but to just arbitrarily excise all innings where the batsmen had the good fortune of not nicking off is a bit... weird. Far better to just look at the "raw" results and then try to interpret rather than forcing that into your statistical machine itself.

Ignoring not outs or looking at runs/innings is "forcing into a statistical machine"? It's a simple exercise.

Chanderpaul averages 35 when batting at #1-4 over 72 innings with only 3 hundreds. He has not shown that given a longer stay in the crease he is capable of making big scores. Being a master of survival is not the same as being able to make huge scores.

 

[Spark]

I agree but that's the sort of analysis you do after you've done your statistical whathaveyou, you don't try to make your statistical exercise reproduce the result you want it to just because, well, you want it to. That's just fudging.

Just include it as a normal, runs/dismissal average afterwards IMO. Then if you see any weird names on there that shouldn't be - like Chanderpaul - then go back, look at his record more closely, and try to explain it then.

 

[viriya]

I agree but that's the sort of analysis you do after you've done your statistical whathaveyou, you don't try to make your statistical exercise reproduce the result you want it to just because, well, you want it to. That's just fudging.

Just include it as a normal, runs/dismissal average afterwards IMO. Then if you see any weird names on there that shouldn't be - like Chanderpaul - then go back, look at his record more closely, and try to explain it then.

This explanation was obvious to me so I didn't feel the need to draw it out. You're suggesting that I checked the with not out numbers and then "fudged" the numbers to get Sehwag on top somehow. I did no such thing - the first instinct when calculating avg50+Scores was to only consider out innings because of the obvious issues with players like Chanderpaul.

Just looking at average is a lazy exercise - its a very deceiving statistic - just because it's the most popular way of judging a player doesn't mean you're required to start from there if it doesn't make sense.

 

[Spark]

All statistics are deceiving. The key is to use the one with the best - which, in most cases, means the simplest - assumptions so you can most clearly see when the statistic doesn't seem to match expectations and why.

 

[indiaholic]

Hey Spark, found that list very interesting. Could you let me know how to convert the average into a runs per innings number?

 

[Spark]

Hey Spark, found that list very interesting. Could you let me know how to convert the average into a runs per innings number?

Uh... take the number of runs and divide by the number of innings? Both numbers are listed there.

 

[viriya]

All statistics are deceiving. The key is to use the one with the best - which, in most cases, means the simplest - assumptions so you can most clearly see when the statistic doesn't seem to match expectations and why.

Isn't runs/innings simpler than average where not outs magically become non-existent innings?

 

[Spark]

Isn't runs/innings simpler than average where not outs magically become non-existent innings?

Eh? They don't become non-existent. They're just counted as not out i.e. without dismissal - because you don't know how many the batsman would have ended up with had they continued to bat, so it's incorrect to treat them all as if the batsman would have gotten out the very next ball - which is effectively what runs/innings does.

People seem to have this weird prejudice against not outs. I really don't get it.

 

[Spark]

http://stats.espncricinfo.com/ci/en...5;runsval1=runs;template=results;type=batting

Here's that list expanded to all players, but with a 2000 rather than 1000 run cutoff (for obvious reasons).

Want to especially look at Ponting there - 91 is bloody impressive, which makes you wonder just what might it have been had he not spent the last three-four years of his career racking up 70s...

 

[viriya]

Eh? They don't become non-existent. They're just counted as not out i.e. without dismissal - because you don't know how many the batsman would have ended up with had they continued to bat, so it's incorrect to treat them all as if the batsman would have gotten out the very next ball - which is effectively what runs/innings does.

People seem to have this weird prejudice against not outs. I really don't get it.

Non-existent innings as in the innings doesn't count when calculating average. I've run runs/innings including not out innings and excluding them with similar results above. But this conversation is going nowhere.