Leaving out the minnows...

[HeathDavisSpeed]

There's been an interesting if some what fractious debate in the ...Daniel Vettori thread regarding what stats should be included and excluded from a player's record in order to allow comparison and standardisation.

The 'easy way' of doing this is by excluding Bangladesh. Yet, we only seem to be willing to do this if it supports the weight of the argument that we're trying to prove.

Why should we leave out Bangladesh? Yes, certain players succeed against them - being the supposed weaker team, but what about the players who fail to succeed against this apparently inferior opposition. Harbhajan, for example, has pretty poor stats vs. Bangladesh - so subtracting them from his analysis improves his results. But surely, his failure to succeed against Bangladesh shows some kind of deficiency?

Also, there is debate as to what else constitutes 'minnow'. Let's exclude Zimbabwe, but why would we want to exclude a Zimbabwe including the Flower Brothers, Neil Johnson, Campbell, Streak and the Strangs? Would we want to?

What about minnows in the past? Surely the stats in early South Africa games should be excluded to? What about New Zealand or Sri Lanka? If so, when did they stop being a minnow and start being a welcome team in the mainstream? When Hadlee started playing? When Aravinda de Silva was at the top of his game?

Then, what about excluding freak batting or bowling paradises? Surely the recent 1st India vs. Sri Lanka team inflates the batsmen's averages; whilst similarly deflating the bowler's averages? The Napier wicket in NZ has been somewhat freakish in its flatness over the last 10 years (a West Indies vs. NZ test about 6 or 7 years ago sticks in my mind as being a prime example). Is it unfair to include big scores in these games in any batting analysis?

What if a team is particularly weak in a country? Should we exclude England or New Zealand's performances in Sri Lanka if (which I can't guarantee they are) they are unrepresentatively substandard?

Even if you exclude all of this 'erroneous' information in order to acheive some utopian standardisation - you're still left with records like Vettori vs. Harbhajan where Harby has played the majority of his tests in India in very different conditions from the majority of Vettori's games in New Zealand. Think of it the other way round, a bog standard seam bowler in NZ can get super bowling stats against decent teams whilst a much better Indian or Pakistani seamer may struggle to be comparable due to playing on home surfaces.

My personal belief is that excluding Bangladesh is just focusing on the tip of the iceberg. I, personally, would not exclude any stats from an analysis of a player - its impossible to standardise records. However, the performance of a 'Vettori' could be caveated with the footnote that 40-odd wickets of his have been taken against Bangladesh.

What are your thoughts on 'The Minnows'. What would you take out of people's records given the chance?

 

[Matt79]

Some good points here. I found it a persuasive opening post on the subject.

 

[Ikki]

Yes, we should discount minnows. Some players have played them quite a bit more than others and so removing them is somewhat of a leveller. I do agree with you that if they have played enough Tests against said minnows and have conclusively failed...then these players have no case to remove minnows. And yes, Zimbabwe for all but the end of the 90s and very beginning of the 00s were minnows.

I agree with you that B/Z are the tip of the iceberg but that's what makes the discussion interesting. There are often circumstances that a player cannot control, whether they benefitted or not and they serve to act as a help/hindrance in which a comparison between said player and another can get lost. So to mention these factors serves to give more accuracy. Of course, such debates often get into tediousness but the merit is there all the same.

Still, removing minnows is a very obvious and IMO an apt act. It's like comparing a player who played often on uncovered wickets vs one that didn't. I see nothing wrong with it.

 

[HeathDavisSpeed]

I do agree with you that if they have played enough Tests against said minnows and have conclusively failed...then these players have no case to remove minnows.

But this is a problem, isn't it. If you remove Minnow Performance from Player A, but not from Player B - who, incidentally, failed against the Minnows, you're making the two incomparable on another basis.

Maybe the solution (if you wish to go down this route) is to seperate Minnow Performance out for both players into an extra stat and then show that Player B clearly has some failings compared to player A that the filtered stats don't show?

I certainly don't profess to have the solution, but selective filtering of statistics is worse to my mind than presenting the full stats with some kind of caveat or footnote.

 

[Uppercut]

I try to at least be even-handed about removing Bangladesh from figures, which can be seen from previous arguments regarding the ability of Jacques Kallis. It's certainly not something that I do specifically to Daniel Vettori, and it could be said that you've mixed up cause and effect because Vettori's figures taking a huge hit when Bangladesh are removed is one reason I think he's overrated.

Other than that, I think your post merely summed up several reasons why statistical analysis can never be perfect. But I definitely think average without Bangladesh is a better measure of a player's ability than average with.

 

[Ikki]

But this is a problem, isn't it. If you remove Minnow Performance from Player A, but not from Player B - who, incidentally, failed against the Minnows, you're making the two incomparable on another basis.

I think you misunderstood what I said. If Players A and B have played enough tests (more than a couple) against minnows and one has failed and the other succeeded I'd take it into account. But where Player A has played them 10 times and averages 80 with the bat, for example, and Player B only played them 2 times and averages 30 then the minnows should be removed.

 

[HeathDavisSpeed]

I try to at least be even-handed about removing Bangladesh from figures, which can be seen from previous arguments regarding the ability of Jacques Kallis. It's certainly not something that I do specifically to Daniel Vettori, and it could be said that you've mixed up cause and effect because Vettori's figures taking a huge hit when Bangladesh are removed is one reason I think he's overrated.

This wasn't supposed to be about Vettori, but the issue of Minnow Stats as a whole - as it is a minefield, and exclusion or, conversely inclusion of Minnow Stats is used and abused on occasions by people with a deadset mindset just wanting to 'prove' themselves right. However, I think the Bangladesh example is pretty dangerous when you do have players out there who have failed against them (or certainly failed to exceed their average statistical mean performance)

And thanks for the clarification Ikki, that's certainly more what I would suspect - but again, I think the whole stats should be presented for player A as I'm sure that a player would be criticised if they'd failed to perform against 'The Minnows'

 

[Richard]

Personally I never take any notice of matches involving Bangladesh (or several others - we'll come to that) when I'm assessing a player's performance in Test cricket. What you say on the subject "why exclude Bangladesh?" has some merit; so I put it to you that it should not be a case of "why exclude Bangladesh?" but "why include them?" The only reason Bangladesh are classed as a Test team is because ICC say so. Most logically-thinking cricket fans agree that their promotion to Test status was a mistake - it runs pretty fluently then, thus, that matches involving them should not be Tests. Personally I recognise no Bangladesh match as a Test, because I've assessed the situation myself, disregarded what ICC say and come to the conclusion that Bangladesh are not and never have been good enough to be playing Tests.

You make a good point about picking-and-choosing - so many say "to exclude <insert team> makes <insert player>'s record worse" but have no truck including matches against such sides when it suits their case regarding another player. Personally I have none of that - in my book a team either deserves Test status or it doesn't. The individual, not ICC, has the right to decide who merits Test status, and if someone decides a team doesn't, then they're fully free to base their judgements of a player's Test credentials solely on matches involving the truly Test-standard sides, and not Bangladesh.

Regarding other sides historically... to my mind the only other cases of substandard sides in Test cricket have been South Africa up until the famous 1905/06 series (so broadly speaking "in the 19th-century" will do) and New Zealand up until the not-quite-so-famous-but-hugely-significant-for-those-in-the-know series in 1961/62. These sides have subsequently become Test-standard; hopefully, eventually, Bangladesh will do the same. The unique case in this matter is Zimbabwe: they have become substandard having previously been up-to-standard. No precedent exists here, but I've regarded the final point of Zimbabwe meriting Test (and ODI) status as the end of the 2002/03 World Cup, because mass retirements happened there, including that of the obvious standout in their days of being competetive.

Regarding sides who are competetive in one place and not elsewhere (India home\away for most of their history being the best example), well I've yet to see a side that hasn't a hope in <insert place> over a long period of time. Teams are generally competetive or not competetive, and of course there is place for considering less weight for something based on general historical trends (ie a Sri Lankan bowler's success against England in Sri Lanka might be worth a bit less than something else somewhere else due to historical English weakness there) but in my book it's no more than that - you can't be completely discounting things on a pick-and-choose basis.

It all boils down to whether someone is willing to accept what ICC say. I doubt many would dispute that WA vs NSW in, say, 1999/2000 was of a considerably higher standard than Bangladesh vs Zimbabwe in 2004/05. To my mind, neither match deserves Test status. Test status depends on standard and geography. Only nations or agglomerations of nations should be given Test status in my book, and only once they've proved they can regularly compete against Test sides (as West Indies, India, Sri Lanka and Zimbabwe all did before being elevated at the right time and being competetive from the start of their time as Test sides).

That is my criteria for what merits a Test match. I will then judge players, in Test terms, on that basis. If others wish to do thus that's their choice, but I see no good reason why "ICC say they're a Test team so I'm going to completely comply with that" should apply.

In ODI terms of course there are many other substandard sides. There is little complication because none of these have yet become ODI-class and their cases are exactly the same as Bangladesh in Test terms.

 

[Richard]

Other than that, I think your post merely summed up several reasons why statistical analysis can never be perfect. But I definitely think average without Bangladesh is a better measure of a player's ability than average with.

Very succinctly put. Stats are never going to be close to perfect; the object thus should be making them as useful as possible. And I agree completely that they are more use without substandard sides than with.

 

[Ikki]

And thanks for the clarification Ikki, that's certainly more what I would suspect - but again, I think the whole stats should be presented for player A as I'm sure that a player would be criticised if they'd failed to perform against 'The Minnows'

Sure, because that would be unusual and not of the norm - hence deserving of that extra attention. Whereas succeeding against them would be par for the course and obviously the player that plays them more would benefit their overall record more.

 

[HeathDavisSpeed]

Interesting thoughts Richard, and I can certainly understand them - but it makes producing stats for players so bloody hard if you're going to exclude this team and that team over different periods.

Then, you've also got to consider the difference in performance standards for changes in the game, such as, the advent of covered pitches. Is it fair to compare a spin bowler before with a spin bowler after? Probably not. But how do you then, statistically compare a 100% uncovered pitch bowler with a 100% covered pitch bowler? I would reason that you can't statistically compare them. Any comparison almost has to be done on a purely anecdotal basis.

 

[Uppercut]

This wasn't supposed to be about Vettori, but the issue of Minnow Stats as a whole - as it is a minefield, and exclusion or, conversely inclusion of Minnow Stats is used and abused on occasions by people with a deadset mindset just wanting to 'prove' themselves right. However, I think the Bangladesh example is pretty dangerous when you do have players out there who have failed against them (or certainly failed to exceed their average statistical mean performance)

And thanks for the clarification Ikki, that's certainly more what I would suspect - but again, I think the whole stats should be presented for player A as I'm sure that a player would be criticised if they'd failed to perform against 'The Minnows'

Well as I put it in another thread, each Australian FC team is better than Bangladesh are. Is the FC record of any given Australian cricketer a better indicator of his ability than his test record? Is having a poor record against Bangladesh any more of a stain on a cricketer's career than having a poor record against Victoria?

 

[HeathDavisSpeed]

It does show some deficiency though, does it not? The inability to psychologically gee themselves up for bowling to minor teams (potentially) may also serve to highlight a possible weakness against lesser batsman. Maybe that bowler is incapable of extracting the tailenders, for example.

Achieving when you are expected to achieve is still important.

 

[Pothas]

All makes for a good reason why we should ignore the stats and just enjoy the game

I would be lying if I said that I did not look at stats or even that I was not ocassionaly inclined to taking out performances against minnows when looking at them but I do not think it is something we should be overly concerned with. Cricket is about watching, not spreadsheets.

 

[aussie]

Regarding other sides historically... to my mind the only other cases of substandard sides in Test cricket have been South Africa up until the famous 1905/06 series (so broadly speaking "in the 19th-century" will do).

Although this is true. I dont in the 1920s, 30s SA where test quality though. After the war they to me weren't 100% test quality until those mid 1950s series vs AUS & ENG (although they did draw 2-2 in AUS 52/53) when they had a bowling attack of Adcock/Heine/Tayfield.

For the other nations WI i guess you could say they became test quality from the 1950 tour to ENG.

IND & PAK in the mid to late 60s?

Sri Lanka probably in the mid 90s

While ZIM had a few years in the late 90s when they where.

 

[Uppercut]

Then, you've also got to consider the difference in performance standards for changes in the game, such as, the advent of covered pitches. Is it fair to compare a spin bowler before with a spin bowler after? Probably not. But how do you then, statistically compare a 100% uncovered pitch bowler with a 100% covered pitch bowler? I would reason that you can't statistically compare them. Any comparison almost has to be done on a purely anecdotal basis.

I do agree with this- comparisons across eras are so complex that it's very rarely possible to find a definitive answer.

That said, I do think it's fair to give credit for success and discredit for a lack of it. A successful uncovered-pitches era spinner is always preferable to an unsuccessful one who bowled on covered pitches. You can't give a bowler credit for wickets you think he might have taken had he been around at a time when spin bowling was easier- that's just speculation. That's the main reason I have no time for Richard's Hayden-Atherton theories.

 

[Richard]

Interesting thoughts Richard, and I can certainly understand them - but it makes producing stats for players so bloody hard if you're going to exclude this team and that team over different periods.

It does sometimes take a bit of work - but personally I'd say that it's worth it to see if you can discern a truer picture.

Take for example the case of Tony Lock. Official Test average of 25, looks damn superb. Don't include New Zealand, who were in my book not Test-standard in his day - bam, changes to 31. Suddenly you realise just how hard-done-by Johnny Wardle truly was.

Then, you've also got to consider the difference in performance standards for changes in the game, such as, the advent of covered pitches. Is it fair to compare a spin bowler before with a spin bowler after? Probably not. But how do you then, statistically compare a 100% uncovered pitch bowler with a 100% covered pitch bowler? I would reason that you can't statistically compare them.

I'd agree completely. I've said time and again that trying to compare a fingerspinner who played in the covered-wickets era and one who played in the uncovered-wickets one is completely pointless. There is no way to know how Giles or Kumble would've fared on uncovered decks - and I mean no way, at all. Likewise there's no way to know how Verity and Rhodes would've fared on covered.

Some statistical comparisons - and those regarding fingerspinners and covered wickets more than most - should simply be steered clear of, completely. IMO.

Others should be ventured into with caution and not at all without caution.

 

[Uppercut]

It does show some deficiency though, does it not? The inability to psychologically gee themselves up for bowling to minor teams (potentially) may also serve to highlight a possible weakness against lesser batsman. Maybe that bowler is incapable of extracting the tailenders, for example.

Achieving when you are expected to achieve is still important.

Yeah, it does. In the same way that Warne was often criticised for not being motivated enough for domestic games, no?

But you still consider his test average above his first-class average when you're looking for an indicator of his ability. On the whole, it's more accurate.

 

[Richard]

All makes for a good reason why we should ignore the stats and just enjoy the game

I would be lying if I said that I did not look at stats or even that I was not ocassionaly inclined to taking out performances against minnows when looking at them but I do not think it is something we should be overly concerned with. Cricket is about watching, not spreadsheets.

The way I've always seen it, you watch the game while it's being played ("there is a place for figures and it is not on the field" [Wright, 2003] - you can tell who's been writing essays ) and you look at stats, figures etc. (as you rather emotively term spreadsheets) when there's no cricket to be being watched. Me I never take any real notice of stats when a game's on, but when it's not, I love looking at them.

 

[HeathDavisSpeed]

Me I never take any real notice of stats when a game's on, but when it's not, I love looking at them.

Other than when they stick up a list along the lines of "Top 10 New Zealand Run Scorers of All Time" during a test match to show that, say, Fleming has moved up above Crowe, I'd agree with you. I'm not big on raw stats either.

To be honest, I'm not keen on huge statistical analyses at any time as you can quite often prove whatever you like by selectively abusing stats. I'm quite partial to a bit of anecdotal analysis "Player A wheeled away outside the left-handers offstump for over after over which a packed offside field" highlighting a defensive mindset, for example.