Visual Whatmore Effect

[Zunaid]

Tried to do some basic plotting for Bangladesh Test performances. 2 charts - one that plots cumulative Batting/Bowling averages and the other runs scored/conceded per 6 ball over.

To see a clear Whatmore effect, one needs to see both charts together. Bowling improvement is fairly obvious, the reason why this is easier to achieve was touched on by Dav himself in his interview. The batting improvement is a little bit more ambigious but is apparent when one looks at both charts. If you look at batting run rate in isolation - this has not increased (so our batsmen are still getting out for small scores). But, you will also notice that batting average has increases - so we have become bette in that respect, this even with Australian/Pak pace attacks. One thing these two charts do nor portray is overs batted per batsman.



Averages



Rates

[Edited on 24-5-2004 by Zunaid]

[Arnab]

Here's what we can do to further refine these graphs. We can normalize the graphs in terms of opposition strength.

To do that, we need to calculate the same statistics (rp6b, avg per wicket, etc.) for all our opponents during that time range.

[Zunaid]

Arnab, I assume I can count on you to volunteer?

[Arnab]

Actually, we can go even further.

We can

1. dig out the PWC ratings of all BD and opposition players for a specific test
2. Get the opposition strength factor (OSF) for that particular test by comparing the ratings.
3. Calculate the whole set of OSFs for all our tests so far.
4. Apply OSF in our calculations to normalize the graphs.

[Edited on 24-5-2004 by Arnab]

[fwullah]

So what do these charts say?

I believe the vast imoprovement in bowling has a lot to do with the Mohsin Kamal factor. Although I don't want to admit it, but Mohsin Kamal has had his influence on some of our bowlers during his tenure.

[pagol-chagol]

ZK -Thanks for the nice chart.

Arnab - You are a genious dude but the cost benefit ratio is off the chart.

[billah]

Originally posted by fwullah
So what do these charts say?

I believe the vast imoprovement in bowling has a lot to do with the Mohsin Kamal factor. Although I don't want to admit it, but Mohsin Kamal has had his influence on some of our bowlers during his tenure.

Dav, in his interview with Rabeed Imam, actuall pointed out the reason of our relative improvement in bowling. Quite a scientific approach. I don't think Mohsin Kamal had anything to do with it. We did not see this during his time. He was our worst coach, in my opinion.

[Zunaid]

A problem with normalizing to get an indication of "true" improvements is the fact that current rating schema are only useful in positioning a particular player/team on a ranking list - the delta between two scores does not provide much insight into the relative strenghts. For example, according to the ICC Test Championship rating schema Australia has 127 points and Bangladesh has 1. So, while it is obvious that Australia is better than Bangladesh, but how much better? Trying to normalize with some OSF based on current rating schemas will only give us a very subjective indication of changes in preformance metrics. The gradiant of dips/increases will be very deceptive. I wonder if there is some rating schema where the delta gives a true indication of strenght difference? Someone with time on their hands could possibly plug in some existing rating system along with head to head results and come up with some stochastic/Bayesian model to estimate true strengths. Anoyne know of such rating schema?

[Arnab]

Originally posted by Zunaid
A problem with normalizing to get an indication of "true" improvements is the fact that current rating schema are only useful in positioning a particular player/team on a ranking list - the delta between two scores does not provide much insight into the relative strenghts. For example, according to the ICC Test Championship rating schema Australia has 127 points and Bangladesh has 1. So, while it is obvious that Australia is better than Bangladesh, but how much better? Trying to normalize with some OSF based on current rating schemas will only give us a very subjective indication of changes in preformance metrics. The gradiant of dips/increases will be very deceptive. I wonder if there is some rating schema where the delta gives a true indication of strenght difference? Someone with time on their hands could possibly plug in some existing rating system along with head to head results and come up with some stochastic/Bayesian model to estimate true strengths. Anoyne know of such rating schema?

I was talking about the PriceWaterhouse Coopers ratings.

We don't have to measure a delta actually. Say we want to measure some batting statistic for Bangladeshi batsmen. Obviously, we need to focus on the Opposition bowling strength. We know the following things:

1. Which opposition bowlers bowled in a particular test match.

Hypothetical example:

A test against Aus.

Opposition bowlers:
Gillespie
McGrath
Lee
Warne


2. The PwC rating for each of those bowlers during that test match.

Gillespie: 790
McGrath: 905
Lee: 785
Warne: 830

3. Average the bowler ratings:

Aus avg. bowling strength (in terms of pwc ratings): 827.5

4. Make a list of average opposition bowling strength (OBS) for each of the tests BD has played:

Example:

1. India: 775
2. Zim: 630
3. Zim: 645
4. Pak: 820
5. SL: 750
.
.
.
28. Zim: 650

5. Pick a base OBS (probably the OBS of the first match against IND) and normalize the rest of the OBS based on that base OBS.

6. Apply the normalized OBS on the graph.

[Edited on 25-5-2004 by Arnab]

[Zunaid]

Arnab, I believe you misunderstood me or the reverse

Let's take a hypothetical scenario, where BD plays 3 Tests against 3 different opponents, A, B and C respectively.
The batting average for BD batsmen are 22, 24, 24 respectively.
OBS being 700, 900, 500 respectively.

Assuming the first as the base, with value 1, the other two normalized OBS (NOBS) become 1.29 and 0.71

The normalized batting average scores for BD batsmen then are (actual x NOBS):

22, 30.96, 17.04

I would like to darw your attention to the dip from 30.96 to 17.04, a 45% dip which is exacltly (900-500/900) since the batting averages were the same for the last 2 Tests.

So did Bangladesh perform 45% worse?

You see, my quandary lies in the fact that the the two OBSs of 900 and 500 only tell me that 900 is much better than 500 but NOT how much better - no indication of true relative strenghts. That is why I said decreases/increases in the performance slope only gives a very subjective indication.

Now if there were a true strenght based normalized rating schema where we could have said that if team A is 1, then team B is 1.2 times better and Team C is .70 times better then the gradiant of the sloped would give a better indication of performance improvements/degradations.

[Arnab]

Hmm. Interesting points.

So did Bangladesh perform 45% worse?

If the last OBS was indeed 500, but the BD batting average was the same, then yes, BD batters indeed did 45% worse, since they scored the same amount of runs aginst a 45% weaker opposition.

The crux here is whether you think the PWC ratings for bowlers are accurate measures of bowling strengths or not. First of all, the average bowling ratings based on pwc will probably never fluctuate from 800 to 500 as your example. Unless you are comparing the Aus bowling attack with the current Zimbo attack. Even if that is the case, it will still be accurate, because I would expect the BD batsmen to avg much better against Zim than Aus.

my quandary lies in the fact that the the two OBSs of 900 and 500 only tell me that 900 is much better than 500 but NOT how much better - no indication of true relative strenghts.

Actually, since PWC rating itself is normalized against a 1 to 1000 scale, it does tell you objectively how much better a certain bowling attack from one time is than another bowling attack from another time. That's the beauty of PWC. You can compare players from different eras.

[Edited on 25-5-2004 by Arnab]

[Zunaid]

Big IF here. IF PwC does reflect true strenghts, then using PwC is fine to use. But, without being party to the underlying math behind it, from my very subjective readings of the scores, I don't quite think it is so. I agree, it is the best there is but not ideal.

[Ahmed_B]

thanx for the great vivid chart Dr. Z

what the two charts show.. could probably stated as like this:

Chart 1:
1. Bowlers are taking more wickets of opponents per innings.(very often they are getting them All-Out at reasonably low score)
2. Number of batsmen getting out per innings on BD side is only slightly reduced.

Chart 2:
1. Bowlers are giving away Much Less runs per over than earlier.
2. Batsmen havn't been able to improve their scoring rate per over by considerable margin yet.

[Edited on 25-5-2004 by crickethorizon]

[fwullah]

but how much better?

About a (one) hundred year.
Explanation: Look at the time period of Australia's test status and Bangladesh's test status.

About the chart - I have only one thing to say regarding our batsman - "slow and steady wins the race". And that is what the chart of batting average suggests.

[fwullah]

P.S. I have complete confidence on our batsman.

1. Shahriar Hossain - did not improve his batting techniques from what I have last seen him in the late 90s, so overall, he hasn't improved. However, he has proved it that against (lower ranked world class teams) ICC Associate teams, he is a basher.

2. Hannan Sarkar - He is the most successful opener for Bangladesh in recent times, both in ODIs and Test cricket, however, he is no Hayden or Anwar, only a Bangladeshi who is hailed from the pathetic batting pitches of Bangladesh. So, we should be proud of him and nurture him carefully, not drop him on each and every chance that we can get, to drop him.

3. Habibul Bashar - his fate still hangs in the balance, being the captain and all, as long as he is producing those big knocks once in every 3 to 5 matches and Bangladesh wins, I don't care about his bad shot selection.

4. Rajin Saleh - here's the blunder - I didn't see him in his debut test series on TV as I was away, but I've heard all the magnificient stories of him over the Internet, and once watching him in action, live on TV, it seemed to me that those who haven't watched him, don't know about his weakness(es), and so were thinking him as someone 'god'like, which he's not. It is my personal opinion that he hasn't done too badly for a new kid, even in this recent WI tour.

5. Mohammad Ashraful - lets wait and see what he has to offer in this series.

6. Foisal Hossain - we must remember during the test series that he was dropped on 17 in the 3-day practice match, before going on to make 43. So lets not expect too much from him, and he may flourish like all debutants.

7. Alok Kapali - (Currently) A confused kid, who doesn't know if he's a specialist batter, or a specialist bowler or a specialist all-rounder, or not. The sooner he gets to decide what he is, the better. And by that time, he will get his form back, too.

Well, if we don't take each player on match-by-match basis, and look at their batting average, then there is nothing for us to worry about, they're just trying to do what they can do - be on a batting average of around 20, and if they're particularly in form, they'll be around 25 and if they're not, they'll be around 18.

There is just 1 more thing that I must add. Our all-rounders did a very good job out there with the bat in the middle on the 3rd ODI (in particular, Mushfique-Rana). You gotta learn to play proper cricket shots and then try to hit bowlers out off the ground.

So, in the conclusion, I would like to repeat that I have complete confidence in our batsman, and in Whatmore, too.

If Dav can't make our batsman bat better, then no one can.

[pagol-chagol]

thanks for the good summary, fwullah.

[Tokai]

mind if i add a few lines about those players in my own way?

Shahrier Hossain: Hopeless
Hannan Sarkar: with proper training, which i'm sure he's getting from dav, and with some more experience he can be our next ataher ali. i agree with u that we should not drop him.

Javed Omar: hopeless for this series. but, can accompany hannan in the future tests. we've to keep in mind that he's just back from injury. he needs more time to get back to cricket. this series was a early call for him in the nation team, by any means.
habibul bashar: the whimsical kid. trys really hard to through his wicket, but sometimes the opponents are rude to him and does not take his wicket as soon as he wants to give it up. those are the days for bangladesh.
Rajin Saleh: sure has faults, but still too young to be faultless. and he trys hard to be the best. a bright aspect, but again need proper direction.
Md asharaful: the two ducks are not his fault in the last 2 odi, i believe. let him play at his natural position, and he'll rock.
foisal: he'll do good in this series, but most likely will dissappear like mushfique-england, rajin-paki and so on. unless rana can be a model for him, he rocked in zim, and did better in wi.
rana: only true inform alrounder.
pilot: will do good in the team is under pressure, in other words when u need him most.
i think this is where the batting stops. not to hopeful, i must say.

[bhobishshot]

I think my brain is wired differently than onters. When I was reading fwallah's comment, I felt like she was pointing out how unreliable BD batters are. But, somehow the conclusion assumed a lot of confidence in them. ? !

[Arnab]