• Welcome to the Cricket Web forums, one of the biggest forums in the world dedicated to cricket.

    You are currently viewing our boards as a guest which gives you limited access to view most discussions and access our other features. By joining our free community you will have access to post topics, respond to polls, upload content and access many other special features. Registration is fast, simple and absolutely free so please, join the Cricket Web community today!

    If you have any problems with the registration process or your account login, please contact us.

Leaving out the minnows...

Son Of Coco

Cricket Web: All-Time Legend
Well, it's never that clear unfortunately. Removing stats tend to give you a number/average that gives people a better gist of a player's records.

You standardise (if that's what you call it) if you wish to go to that route. Use the players' averages and take into account how much one player played other teams and give the identical amount to the other player. Much closer and accurate.
Except that removing stats doesn't always tend to give you a number that gives you a better gist of player's records, as Heath suggested in his first post.

And making a random guess at how many runs a player would score vs a certain team if he batted the same amount of times based on what another player did is neither 'standardising' things or making it more closer and accurate, quite the reverse in fact.



The problem if you don't do it is that people try to penalise players for scoring a lot of runs against minnows. See some of the attitudes to Hayden's 380 on here for details- it's treated as some kind of crime, as if scoring no runs would have been preferable to scoring loads.
I don't think taking out all the results against 'minnows' because of what some people try to do when their least favourite player performs against them is necessary though. I can't see how it does Hayden any great favours to exclude the runs totally from the stats rather than have people argue about whether they're good runs or not. The arguments about its worth say more about the people than the value of the stats IMO.
 

Ikki

Hall of Fame Member
Except that removing stats doesn't always tend to give you a number that gives you a better gist of player's records, as Heath suggested in his first post.
I answered his post and I think he agreed. Unless both players have played the same amount against minnows, there is no reason to remove them. Or they've both played enough to have a decent sample to suggest whether they were successful or not. But when there is one who has played a lot against them and one who has played very little then there should be some leveler.

And making a random guess at how many runs a player would score vs a certain team if he batted the same amount of times based on what another player did is neither 'standardising' things or making it more closer and accurate, quite the reverse in fact.
It's not a random guess at all. Unless you think a person's average against a certain test side is random, then all averages, minnow or not, are random. Which is not the case if there is a reasonable sample IMO.

So you just equate the number of balls the faced/bowled against certain opposition and compare. Pretty simple and I fail to see how it would do the reverse and make it less accurate.
 

Richard

Cricket Web Staff Member
If we're going to take away all the tests against minnows, what about players who have played more tests against the top 3 nations? How do we balance this up to give a fair representation of how they stand against the rest? As Heath mentioned in the opening post, once you start doing this where do you stop?
Have said it before - the difference between a substandard side (Bangladesh) and any Test-class side is a World away from the difference between a good Test side and a not-so-good one.
 

Top_Cat

Request Your Custom Title Now!
Have said it before - the difference between a substandard side (Bangladesh) and any Test-class side is a World away from the difference between a good Test side and a not-so-good one.
Peer review doesn't work if it's the same person saying the same thing multiple times.
 

Richard

Cricket Web Staff Member
Except that removing stats doesn't always tend to give you a number that gives you a better gist of player's records, as Heath suggested in his first post.
I can't really see how it doesn't. I'd say more that leaving them in essentially never gives you a better gist of a player's records. Bangladesh are not a Test-standard team, and their matches should not be classified as Tests.

Put it this way - if I managed a coup at ICC tomorrow and stripped Test status from all games involving Bangladesh, everything involving Zimbabwe from 2003 onwards, South African games up to 1905/06, Kiwi ones up to 1960/61 and that World XI rubbish, I doubt very many people would moan for very long, and in fact I assure you lots of people would celebrate. There'd be some people who'd have to adjust their views of certain players when those views have been based purely on the stats they've been spoon-fed, but that's a good thing.
 

Top_Cat

Request Your Custom Title Now!
So you just equate the number of balls the faced/bowled against certain opposition and compare. Pretty simple and I fail to see how it would do the reverse and make it less accurate.
It's to do with the predictive power of sports stats, basically. So many uncontrolled factors influence how many runs a batter will score that the predictive power of averages/runs scored on the same ground/against the same opponent is weakened significantly. Too much to use predictively at all, for mine, but that's just me.
 

Richard

Cricket Web Staff Member
You seriously believe that it was not the case that in (say) 2006, the difference between West Indies and Bangladesh was not far greater than that between Australia and (for instance) Pakistan?
 

Richard

Cricket Web Staff Member
It's to do with the predictive power of sports stats, basically. So many uncontrolled factors influence how many runs a batter will score that the predictive power of averages/runs scored on the same ground/against the same opponent is weakened significantly. Too much to use predictively at all, for mine, but that's just me.
Not just you at all; I certainly agree, know more than a few others do and suspect that the vast majority would do when they thought the matter through.

Sports stats, and cricket especially, can only be used on a "what did happen" basis, not a "what might have happened" one.

(And BTW just in case anyone is thinking of it, a chance being given and missed IS INDEED SOMETHING THAT DID HAPPEN!!!!!!!!!!!!!!!!!)
 
Last edited:

Son Of Coco

Cricket Web: All-Time Legend
It's not a random guess at all. Unless you think a person's average against a certain test side is random, then all averages, minnow or not, are random. Which is not the case if there is a reasonable sample IMO.

So you just equate the number of balls the faced/bowled against certain opposition and compare. Pretty simple and I fail to see how it would do the reverse and make it less accurate.
A person's average against all sides is a result of what actually happened up until that point. Deciding that person A would've scored X amount of runs against team he didn't play at all based on what person B did on a certain day in conditions that would certainly be different based on what theyve done against certain other teams and their respective averages is a random guess.

The idea that you can accurately guage how many runs a player will score on a given day accurately given their career average is simplistic at best. I'll be looking foward to seeing if Ricky Ponting can stay consistent and score 55.88 in both innings in Adelaide at a strike rate of 59.41.
 
Last edited:

Son Of Coco

Cricket Web: All-Time Legend
Have said it before - the difference between a substandard side (Bangladesh) and any Test-class side is a World away from the difference between a good Test side and a not-so-good one.
What about matches played on substandard wickets that produce results where both teams look poor? Or matches where 'substandard' teams perform well? Aren't you assuming a consistency that isn't really there?

Isn't it just easier to look at what a player did against the different teams they played against and make an assumption from that rather than trying to remove stats to make the overall number 'more accurate'?

For example, did anyone but the most ardent English fan think what Ravi Bopara did against a (very distinterested) WI side meant anything in the grander scheme of things? I don't think too many people think his average reflects his actual ability given they're the only side he's performed against. Do we take these results out too? I'd venture to say that if he plays enough tests his average wil be a lot closer to his ability. And if he doesn't then these tests will be viewed as an anomaly when compared with the rest of his career.

Much simper than choosing when we take away tests and when we don't.
 

Quaggas

State Captain
The idea that you can accurately guage how many runs a player will score on a given day accurately given their career average is simplistic at best. I'll be looking foward to seeing if Ricky Ponting can stay consistent and score 55.88 in both innings in Adelaide at a strike rate of 59.41.
Of course not, but presumably there is some relatively stable conditional probability distribution for 'runs scored' (or chance of getting out etc).The conditioning variables would be the opposition (ideally not just 'team A' but 'bowler X, keeper Y, captain Z' etc), pitch/grounds, etc The means model could even be auto-regressive if you believe in purple patches. Unfortunately, there are obviously not enough matches played to make a meaningful model at such fine a granularity.

Anyone written or looked at the workings of a sim?

Edit: BTW in principle 'Bowler X in 1990' would be different from 'Bowler X in 1993'.
 
Last edited:

Richard

Cricket Web Staff Member
What about matches played on substandard wickets that produce results where both teams look poor? Or matches where 'substandard' teams perform well? Aren't you assuming a consistency that isn't really there?

Isn't it just easier to look at what a player did against the different teams they played against and make an assumption from that rather than trying to remove stats to make the overall number 'more accurate'?
Not really. I've no problem with people who just try to say "stats aren't important" and claim they make their judgements without reference to stats. But a team is either substandard or up-to-standard; no picking and choosing, as I said before, it either does or does not deserve Test status. There are all sorts of teams who might occasionally perform well enough to compete with a Test side, but the one-off occasion doesn't mean they deserve Test status on a one-off basis or long-term. What irritates me is how often enough the same people who'll claim that Adam Gilchrist and Stuart MacGill's performances against Bangladesh are meaningful but Graeme Hick's Test failures mean that his performances at lower levels are irrelevant. In my book if you want to make the distinction between Test and domestic-First-Class you have to act to purify that distinction to the maximum extent possible, and I fail to see how anyone can claim that Bangladesh's inclusion among Test-playing sides does anything other than dilute the purity of Tests' superiority. In classifying Bangladesh as Test you're blurring the distinction between what is supposed to constitute Test and what is supposed to constitute First-Class-only.

As for substandard pitches, there's in my book no such thing. A pitch can be poor for batting, poor for bowling spin, poor for bowling seam or whatever; it can, very, very occasionally, be dangerous. But it cannot be substandard. Disappointing, and producing a less-than-riveting Test (ie 600 plays 600), yes. Substandard, no.
For example, did anyone but the most ardent English fan think what Ravi Bopara did against a (very distinterested) WI side meant anything in the grander scheme of things? I don't think too many people think his average reflects his actual ability given they're the only side he's performed against. Do we take these results out too? I'd venture to say that if he plays enough tests his average wil be a lot closer to his ability. And if he doesn't then these tests will be viewed as an anomaly when compared with the rest of his career.

Much simper than choosing when we take away tests and when we don't.
Bopara's case merely shows the virtue of the first-chance average. He was dropped in each of those innings' and had the chances been taken he'd have scored nothing-much in any, disinterested WI or not. The disinterest, however, was symptomatic not of the fact that WI were substandard, but that the series in England simply should not have been being played at all.
 

Son Of Coco

Cricket Web: All-Time Legend
Not really. I've no problem with people who just try to say "stats aren't important" and claim they make their judgements without reference to stats. But a team is either substandard or up-to-standard; no picking and choosing, as I said before, it either does or does not deserve Test status. There are all sorts of teams who might occasionally perform well enough to compete with a Test side, but the one-off occasion doesn't mean they deserve Test status on a one-off basis or long-term. What irritates me is how often enough the same people who'll claim that Adam Gilchrist and Stuart MacGill's performances against Bangladesh are meaningful but Graeme Hick's Test failures mean that his performances at lower levels are irrelevant. In my book if you want to make the distinction between Test and domestic-First-Class you have to act to purify that distinction to the maximum extent possible, and I fail to see how anyone can claim that Bangladesh's inclusion among Test-playing sides does anything other than dilute the purity of Tests' superiority. In classifying Bangladesh as Test you're blurring the distinction between what is supposed to constitute Test and what is supposed to constitute First-Class-only.

As for substandard pitches, there's in my book no such thing. A pitch can be poor for batting, poor for bowling spin, poor for bowling seam or whatever; it can, very, very occasionally, be dangerous. But it cannot be substandard. Disappointing, and producing a less-than-riveting Test (ie 600 plays 600), yes. Substandard, no.

Bopara's case merely shows the virtue of the first-chance average. He was dropped in each of those innings' and had the chances been taken he'd have scored nothing-much in any, disinterested WI or not. The disinterest, however, was symptomatic not of the fact that WI were substandard, but that the series in England simply should not have been being played at all.
I wasn't saying I make judgements without reference to stats, but that looking at the overall average is often misleading (especially when dealing with players who haven't played many tests). Rather than trying to 'standardise' that according to whatever suits why not just break them down and see what they reveal?

Bopara's case reveals more than the virtue of the first-chance average. The virtue of looking past his overall average and seeing what he really did/who he played against/the situation stands out bit too. You don't have to remove the stats to do that.
 

Son Of Coco

Cricket Web: All-Time Legend
Of course not, but presumably there is some relatively stable conditional probability distribution for 'runs scored' (or chance of getting out etc).The conditioning variables would be the opposition (ideally not just 'team A' but 'bowler X, keeper Y, captain Z' etc), pitch/grounds, etc The means model could even be auto-regressive if you believe in purple patches. Unfortunately, there are obviously not enough matches played to make a meaningful model at such fine a granularity.

Anyone written or looked at the workings of a sim?

Edit: BTW in principle 'Bowler X in 1990' would be different from 'Bowler X in 1993'.
If you could make a sim that took all the variables into account and it even got close to being accurate over a period of time I'd be amazed.

After a certain amount of games you can look at a guy's average and make assumptions on the likelihood of him scoring runs based on prior performances, the pitch, strength of opposition etc. That doesn't make it any less likely that Bowler A, B, C, D or E won't bowl an absolute jaffer and uproot his middle stump first ball. A guy averaging 50 is obviously more likely to score more runs than a guy averaging 1.50, doesn't mean it'll happen though.

The original post I was replying to seemed to be suggesting you could simply take what one guy has done and make assumptions based on that for another guy on a different day using their averages, and do so accurately.

I'd respectfully suggest that would be total waste of time.
 

GIMH

Norwood's on Fire
For example, did anyone but the most ardent English fan think what Ravi Bopara did against a (very distinterested) WI side meant anything in the grander scheme of things? I don't think too many people think his average reflects his actual ability given they're the only side he's performed against. Do we take these results out too? I'd venture to say that if he plays enough tests his average wil be a lot closer to his ability. And if he doesn't then these tests will be viewed as an anomaly when compared with the rest of his career.
.
His average is 33, I reckon that probably does just about reflect his ability tbh
 

Uppercut

Request Your Custom Title Now!
I wasn't saying I make judgements without reference to stats, but that looking at the overall average is often misleading (especially when dealing with players who haven't played many tests). Rather than trying to 'standardise' that according to whatever suits why not just break them down and see what they reveal?

Bopara's case reveals more than the virtue of the first-chance average. The virtue of looking past his overall average and seeing what he really did/who he played against/the situation stands out bit too. You don't have to remove the stats to do that.
Well, to put the same question to you as I did everyone else, would you also include Australian FC stats? That's a higher standard than a game against Bangladesh.
 

Top