Print

Please post once and this message will disappear! Introduce yourself, say hello, jump into a discussion...

[LosingNow]

Kban/Sudzz/Ruchir and I were discussing this on chat. (specially after Dinesh Mongia's performance yesterday). Would it not be a good statistic to display on ODI scoreboards? Strike Rates do not reveal the full picture re: the productivity/quality of a batsman in an ODI.

Interesting to see that Australia had the lowest dot ball % in the last WC...look at the table about half way down this report (sorry, I do not yet know how to copy/paste tables).
 
http://content-usa.cricinfo.com/wc2007/content/story/292912.html

Based on this table, it looks like anything close to 50% dbp is a good # (if the Aussie # can be used as a benchmark). England and WI (among the main cricket playing nations)with ~60% dbp wasted on average 10% of the balls ..i.e. ~30 balls (if innings length is assumed to be 300 balls) which accounts for a minimum of 30 runs (if at least 1 run was scored of those dot balls). 
 
--
I think dbp and SR can be useful stats to track and plan batting lineup and/or approach during the ODI.

A batsman with low dbp and high SR (say 90%+) can score equally well in singles, 2s, 3s and boundaries. He is the ideal ODI batsman. For future discussion, let's label him as Cat 1 batsman. 

A batsman with low dbp and low/medium SR (say 50-80%) can score in singles but CANNOT hit boundaries or 3s. He is a nudger/gap finder. He  is a useful "strike-rotating" batsman in ODIs. Cat 2 batsman.

A batsman with high dbp and high SR (say 90%+) scores primarily in boundaries and 3s. He is the boundary hitter/pinch hitter. He can be a good powerplay batter ..and maybe, useful final overs blaster (but his high dbp is the big question mark in this role). Cat 3 batsman.

A batsman with high dbp and low SR ..does not belong in the ODI game.. perhaps he is a good test player

----

With this classification, here are my thoughts/theory on an ideal ODI batting line up (of course, it would be ideal to have all Cat 1 batsmen in an ODI team..but that is not possible..) for a typical ODI game (not a high scoring game, not a low scoring game).

During Power-play overs : The openers, #3 (and maybe #4) batsmen should be balanced between Cat 2 and Cat 3 batsmen. Ideally, one should open with a Cat 2 and Cat 3 player..and depending on who gets out..keep that combination going for the duration.

During the middle overs : This period would be best for Cat 2 batsmen (with preferably medium SR). Cat 3 players have limited usefulness (unless it is a high-scoring game) during this period.

During the slog overs: The combination Cat 2 (NOT with low SR) and a Cat 3 batsman should ideally play during this period.

Thoughts?

(Of course, we have to add average as another stat to this mix to evaluate a batsman's usefulness in ODI.. a measure of quantity of runs rather than quality - which is what dbp% and SR are trying to capture)

--
Also, without doing the stat search, based on gut-feel I think some of the contemporary players would fit into the different categories as follows:

Cat 1 (low dbp-high SR): Hayden, Ponting, Dhoni, Yuvraj(?), Uthappa(?), Gilchrist(?), Afridi, SRT of the past, KKD (but he may not enough innings to justify this)..
Cat 2 (low dbp-low/med SR): Kaif, Clarke, Dravid, Younis Khan..
Cat 3 (high dbp-high SR): Sehwag, Jayasuriya, Ganguly, Gambhir(?)..

..and just for fun.. Cat 4 (high dbp-low SR): VVS , Jaffer
 
I am sure some of you can prove/disprove this gut-feel by looking at real #s.. what is your gut-feel?

[rajesh]

Nice interesting one .. could disprove lot of myths

[WicketView]

The dbp is a good suggestion.

Apart from the reasons of categorization (spare a thought for TL like me, and please do not call them cats) that you forwarded, there is one more utility of this number. Often a batsman's strike rate is bolstered by an over or two of a particular bowler (who may either bowl a style to his liking, or plain badly), To maintain a high dbp, one has to perform against almost all the bowlers.

On the other hand, the different modes of the game bias this estimator. With the field spread out, taking singles is a lot easier than hitting 4s or 6s (in terms of risk factor). On the other hand during the powerplays, a ball hit through the gaps enough will likely end up as a 4. For example I think this would tend to take AG towards cat 3 in the innings that he opens in ... not because of ability, but circumstances.

So my modification would be to use it for the same range of overs, which may also be something that you want (since you will most likely compare openers to openers; and MOB to MOBs) rather than using this statistic to figure out where a batsman should bat.

Gut feelings that are somewhat different: ( I thought Afridi would be in cat 3, Jayasuriya might be in cat 1, KKD (with more innings) in cat 2) Utthapa (not enough runs so far, RD is enigmatic ... he definitely was cat 4 earlier, now he is cat 2 and sometimes promises cat 1, but does not quite get there. Jaffer can be in cat 3  or even cat 1 if the bowler can be bribed to bowl on the legside )

[LosingNow]

 I thought Afridi would be in cat 3,

I thought about that .. because he tries to hit out each ball (whenever he is on the crease!).. I think his dbp would be low and of course, his SR is high. I think due to his inconsistency, his average would be relatively lower. From a quality point of view, he is an ideal ODI player.. the problem is consistency and that should be better reflected in average.

[WicketView]

 I thought Afridi would be in cat 3,

I thought about that .. because he tries to hit out each ball (whenever he is on the crease!).. I think his dbp would be low and of course, his SR is high. I think due to his inconsistency, his average would be relatively lower. From a quality point of view, he is an ideal ODI player.. the problem is consistency and that should be better reflected in average.

I thought that would be the case with Utthapa so far. With Afridi, who has played a lot more, I think all his tries did not connect ... often he would simply be beaten.

[KKIRANK61]

Good stats angle.........applause !!
However, the amount of runs ie the average also indicates the utility of a batsman. So a due consideration must be given to it in the formula. A similar point was floated sometimes back when a 'batsmanship factor' was defined where BF= (AVG + 0.5xSR)xEfficiency. For more details and recollection one may refer attached file shown below. In any case such stats have adv and disadv, as it revels something but hides also something. But a good cooeficient to look at.

[justforkix]

Also, without doing the stat search, based on gut-feel I think some of the contemporary players would fit into the different categories as follows:

Cat 1 (low dbp-high SR): Hayden, Ponting, Dhoni, Yuvraj(?), Uthappa(?), Gilchrist(?), Afridi, SRT of the past, KKD (but he may not enough innings to justify this)..
Cat 2 (low dbp-low/med SR): Kaif, Clarke, Dravid, Younis Khan..
Cat 3 (high dbp-high SR): Sehwag, Jayasuriya, Ganguly, Gambhir(?)..

..and just for fun.. Cat 4 (high dbp-low SR): VVS , Jaffer

Did you omit Symo and Hussey on purpose, since perhaps this discussion is about humans 

I disagree about Puppy. He definitely belongs to category 1.

KP will belong to Category 1. Kallis, Styris, Jayawardane, PC will belong to category 3.

[pieterSAN]

I like this LN....it's pretty simple. 

[sgusa]

I agree with KKK (YIKES!) that the average runs should also be factored in. No point having low dbp, high SR if you only score 30 runs everytime.

[justforkix]

I agree with KKK (YIKES!) that the average runs should also be factored in. No point having low dbp, high SR if you only score 30 runs everytime.

Yes, otherwise we may conclude that Murali and Zaheer Khan are the best ODI batsmen around

[sgusa]

I agree with KKK (YIKES!) that the average runs should also be factored in. No point having low dbp, high SR if you only score 30 runs everytime.

Yes, otherwise we may conclude that Murali and Zaheer Khan are the best ODI batsmen around

How about weighted {dbp, SR} stats ?

Take x batsmen that you are looking to compare ( note that while you could have gloabl stats here, it is more useful to look at a sample of x and see who is better, with x being some mid size sample of the X: all batsment )

xi(i=0..n)      =       Avg(xi)         
                          --------        *   { dbp, SR }
                    Sigma(Avg(xi))/n

what say ?

PS: All math done prior to first cup of coffee after waking are subject to being absolute rubbish

[openforum]

Great post LosingNow... 

This coefficient must be calculated based on the type of pitches?
Suppose a batsman whose combined (SR+DB%+Ave.)  puts him in cat A on flat tracks, he  might be in cat B on greentops etc.

So lets list the factors which should be taken into account for such a coefficient:
           1) Strike rate
           2) Average
           3) Dotball percentage
           4) Field restrictions (power play etc)
           5) Wicket condition
           6) Bowling attack
         List any others please...

Here is a link to a paper on estimating possible scores in notout innings and arriving at a more accurate average:
     www.jssm.org/vol5/n4/5/v5n4-5text.php

[natty]

BaP & BoP: A tale of two products


An easily understood, easily compared stat is (christened here) the Batting Product^TM

Is the Product of Strike Rate and Batting Average  (or have batting average substituted with median batting score as it is less biased than the average especially for newcomers with fewer innings)

Lets say SRT has an average of 44 and an SR of 0.85, thus SRTs career Batting Product is 44 x 0.85 = 37.4

Dhoni's Batting Product is close to 45

According to the implied theory, the impact of a batsman is (crudely) captured by this multiplicative product and probably we can pretend that it is on a ratio scale.  Using this measure, it will be obvious that Australia has many high impact batsmen and one  can (exactly) show how much they are superior on this count.  The unit of BaP is runs (as the strike rate is dimensionless).

Now a similar approach gives rise to the Bowling Product^TM

Bowling Product is  Wickets taken per 60 balls   Divided by (Runs conceded per 60 balls)/60   {it is still a product!}

The denominator is very much strike rate except that in this case a low strike rate is good (for the bowler)

The unit of BoP is wickets (as the strike rate in the denominator is dimensionless)

So both BaP and BoP are in meaningful units (runs and wickets respectively).. One can compute these not just for entire careers but for any subset of a players stats, trends etc.

[kban1]

In chat this morning, I was mentioning to LN, Suddz et al, that I had been working on an anlysis which compared our players with the Australians (the gold standard) in terms of dot ball %ge.

Hampered because the regular scorecards do not provide the breakdown between dot balls and non-dot balls, I devised a syetem of finding out a batsman's strike rate without boundaries -- which is a rough but not exact approximation of the dbp (dot ball % age), as LN refers to it

Since shorter innings do not always reflect a batsman in full flight, I chose innings > 60 runs. I also chose a time period starting from Jan 1, 2004 even though that provided me with a smaller sample for people like Hussey (debut in 2004), SRT (injury ridden), and SG (poor form, not in ODI side for 15 months).

I chose Hayden (an opener), Ponting (higher order bat), and Hussey (finisher) on the Aussie side.

On the other side, I chose SG (opener), SRT (higher order bat), and Dravid (finisher, although he has batted higher up on afew occasions including opening)

Intuitively, it seems normal that taking singles would be easier on larger grounds, so I broke up the non-boundary SR (my approximation of dot ball %) by large grounds vs average grounds in addition to the away vs home factors.

I also realized that finishers like Hussey and Dravid would have higher non boundary strike rates due to spread out fields (easier to take singles).

Obviously, the Aussies were ahead but SRT and RD were not too far behind (in fact SRT was ahead in couple of categories despite suffering from small number of data points). SG lagged Hayden by quite a bit (part, not all explainable by poor form / small sample size).

The spread sheet is attached. Feel free to modify to add new fields, new names, adjust time periods et al.

PS: I have attached the file as strikerates.xls.doc because the system wont allow me to attach an xls file. When you download, remove the .doc extension and then you can play with it.

[TheWall]

LN, maaan you are genius! Or tell me the truth - are you Buchanan or Boder posing out here?

Reading your post, it seemed to be that Aus must already have such models in place and may be utilizing them in addition to their normal criteria for evaluating players.

Basically, soon, cricket will be all Cat1, maybe some Cat2 players. Cat 3 and stuff will be acceptable from bolwers, sub-par 'allrounders'...

[feverpitch]

Nice interesting one .. could disprove lot of myths

exactly!

[atticus]

In chat this morning, I was mentioning to LN, Suddz et al, that I had been working on an anlysis which compared our players with the Australians (the gold standard) in terms of dot ball % ge.

Hampered because the regular scorecards do not provide the breakdown between dot balls and non-dot balls, I devised a syetem of finding out a batsman's strike rate without boundaries -- which is a rough but not exact approximation of the dbp (dot ball % age), as LN refers to it

Since shorter innings do not always reflect a batsman in full flight, I chose innings > 60 runs. I also chose a time period starting from Jan 1, 2004 even though that provided me with a smaller sample for people like Hussey (debut in 2004), SRT (injury ridden), and SG (poor form, not in ODI side for 15 months).

I chose Hayden (an opener), Ponting (higher order bat), and Hussey (finisher) on the Aussie side.

On the other side, I chose SG (opener), SRT (higher order bat), and Dravid (finisher, although he has batted higher up on afew occasions including opening)

Intuitively, it seems normal that taking singles would be easier on larger grounds, so I broke up the non-boundary SR (my approximation of dot ball %) by large grounds vs average grounds in addition to the away vs home factors.

I also realized that finishers like Hussey and Dravid would have higher non boundary strike rates due to spread out fields (easier to take singles).

Obviously, the Aussies were ahead but SRT and RD were not too far behind (in fact SRT was ahead in couple of categories despite suffering from small number of data points). SG lagged Hayden by quite a bit (part, not all explainable by poor form / small sample size).

The spread sheet is attached. Feel free to modify to add new fields, new names, adjust time periods et al.

PS: I have attached the file as strikerates.xls.doc because the system wont allow me to attach an xls file. When you download, remove the .doc extension and then you can play with it.

Excellent work kban. Instead of looking at innings > 60 runs, could you modify it as innings with > 75 balls - fairly long innings to see the batsmen in full flight. This will make sure you include the slow innings by batsmen. As far as your conclusions go (that SRT and RD are not too far behind the aussies), this is where the team concept comes in. Individually, Aussies are not really a million times greater than others. But every time, there are at least 1 or 2 who put their hand up for the aussies whereas for the other teams, there are just too many collective failures..

[atticus]

Great new addition to stats LN. I think, this would be a really useful analysis tool. I hope that the team managements actually use this metric even currently and it is only not available to the public. It could be something similar to the OPS in baseball.

[KKIRANK61]

Here is one more angle to the whole story of measuring batsman's performance:

I have a specially designed ‘score card’ which , although difficult to maintain, may throw some more light on how a particular batsman is fairing in batting job. One can think of keeping record in this fashion from now on,  forgetting what has happened in the past (in addition to existing statistics). We argue many times on who is performing better than whom, and thus who should be selected above whom, etc. Such score card is a detailed synthesis of an innings the batsman plays. So far we count only balls faced and runs scored. Then we work out AVG and SR as two parameters for evaluation. But there are lot more things which can be minutely observed and recorded.
See what are the possibilities in which a batsman REACTS when a ball is bowled at him.
__________________________________________________________________________________
FOR WICKET:
1. WKT+   Out to ball in a conventional manner, No extraordinary catch or runout by fielder. A regulation wicket                        1.00
2. WKT-   Out to a ball due to extraordinary catch or extraordinary runout or similar event where he is out without a serious mistake.                  0.80
( The number of times a batsman is out is multiplied by this factor. Every times when a batsman is out, it will count either 1 (pure out) or 0.7 ( ‘slightly less than out’ ). So suppose it happens on 4 out of 20 times he is out, then he is considered as out for (16+4*0.7=18. 18.8 times instead of 20 times! Some credit regained from a wicket.
( and why not because suppose one player is out to a genuine mistake and other is out to an extraordinary effort by fielder, there should be some distinguishion in the two dismissals which is currently not existing.)
__________________________________________________________________________________
OTHER THAN WICKET:                                  GRADE
1. BEAT+   Beaten to a delivery where he attempted a stroke.      0.05
2. BEAT-   Beaten to a delivery because could not read it well.      0.00
3. DEF+   Defensive stroke to ball which needs only defense.         1.00(good)
4. DEF-   Defended a ball which was scorable (eg. Overpitch, longhop, fulltoss)   0.50
5. STR++   Stroke to a ball as intended by batsman and earn runs.       1.00(good)
6. STR+   Stroked a ball as intended by batsman but no runs due to good fielding.   0.90
7. STR--   Stroked a ball but straight to a fielder.             0.70
8  STR      Stroked a ball but straight to a fielder and he misfields but no run.0.70
9. STR-   Stroked a ball but straight to a fielder and he misfields, gives run.   0.70(RNC)
10 STR-ED   Stroked a ball, an Easy catch straight to a fielder who Drops   0.25
11 STR-DD   Stroke to a ball, a Difficult catch to a fielder who Drops   0.30
12 EDGE   Edged a ball without run.                     0.20(bad)
13 EDGE   Edged a ball but got runs.                     0.20(RNC)
14 EDGE-ED   Edged a ball as an easy catch but dropped         0.10(bad)
15 EDGE-DD   Edged a ball as difficult catch and dropped         0.15(bad)
16 LEFT   No intention/ no need/ not possible to play             1.00(good)
17 INNOV   An innovative and improvised stroke ( for all STRs above )  extra 0.05

RNC = RUNS NOT TO BE COUNTED

Explanation:
Ideally speaking, suppose a batsman faces ‘n’ balls, (and he is ideal) so all these balls will be either DEF+ or STR++ or LEFT category as stated above. But nobody is perfect and hence there is variation. As per the amount of incorrectness in reacting to a ball, the action is given a grading out of 1. Thus eg. Edging a ball is a rather disqualifying attempt (hence 0.20) and scoring on that is a ‘shame!’ hence runs not to be counted ie. Separate record of such runs should be kept. Thus edging a ball for a catch is a ‘crime’ and being dropped is a ‘shamefull escape’ hence 0.1 grading for that. Likewise defending a ball which is a scorable like fulltoss, longhop, overpitch etc is also a disqualification (so  0.5) but still not as bad as edging which is 0.2 Being  beaten all end up amounts to 0.00!! and why not?
Thus every ball is not just counted like today but categorized in this manner such that at the end when a batsman faces say 100 balls and if this total of grading comes to 80 that means his EFFICIENCY is 80%.
The current measurement of performance is in the form of AVG which measures a runs scoring ability but not the manner in which it is scored. There is SR also which measures quickness in scoring. To this now the efficiency factor can be added. Moreover there can be a thought of combining all these three ie. AVG, SR and now EFF to arrive at some sort of quotient.

eg. BATSMANSHIP FACTOR (BF) = ( AVG + 0.5 x SR ) x EFF

Thus somebody having more conventional figures like AVG = 35, SR = 76 and EFF = 75 has BF as
BF = (35 + 0.5 x 76 ) x 0.75 = 54.75
Another more conventional but with more luck factor will have less EFF value say 0.5 instead of 0.8 above. Thus will have BF = 36.5
A slogger with AVG = 25, SR = 100 and EFF = 40 ( being a slogger will be low on efficiency scale) has BF = (25 + 0.5 x 100) x 0.4 = 30
A genius with AVG = 45, SR = 85 and EFF = 85  will have BF = 74.8

Note: Here avg is calculated using  (1) excluding lucky runs as stated above (2) with revised ‘number of times out’ as stated above where out with bad luck is counted not 1 but 0.8.

A score card may be designed where such a detailed response of a batsman may be marked and a statistics can be built which might be useful to analyze who is performing how much as compared to other.
 It may be noted that it can still not be a 100% measurement of performance since factors such as opponent’s quality, pitch condition, match situation etc, are also some of the vital factors which play some role in performance. Further efforts can be taken to include this and some other possibilities as well !!

[atticus]

kban,

I found 1 slight error in your stats. The formula for column "balls with 0 boundaries" is wrong. It should be
=+C2-(D2+E2). In your excel, it was  =+C2-D2+E2. Basically, all 6's were not considered boundaries. It really does not affect a lot of stats (because the number of 6's are very less). So, no big deal.

[kban1]

atticus:

Thanks.

Also thanks for pointing out the error.

You are right, it didnt affect the numbers much except for Hussey because Hussey has smaller number of innings peppered with a lot of sixes.

I am reposting the adjusted spreadsheet as an attachment to the original post.

[sgusa]

"balls without boundaries" - sounds like the title of a porno