New ways to rate cricketing talent.

[Neel Here]

sorry BCites for starting a new thread on this but rimon88's thread on moneyball concept had got me thinking. http://www.banglacricket.com/alochon...ad.php?t=29297

and then we had the discussion on the performers in the domestic league. we have many cricket-knowledgeable experts here as also many with mathematical/statistics skills, some with both !

what if we can come up with a rating system that can really help us find out and predict who will be the future stars of BD cricket ?

I propose that BCites use the upcoming matches against SL A team and the ranji team to form a rating system to gauge a player's performances.
people can either propose a whole system on their own or particular aspects of cricket, like bowling etc.

I have some ideas of my own but it will take some time and effort to put it in readable shape.
I'll continue only if there is enough interest in this idea.

for the moment, I am asking interested members to put up their hands by posting in this thread. if there is not enough interest, mods can simply delete this thread !

[Beamer]

Every new initiative is applauded from my end. Go ahead with the rating system that you have in mind and we will surely discuss it.

However, when domestic matches are involved, it is a bit difficult to gauge reasonable projections of a particular player since we are not able to watch him perform. Scoreboard reading don't always paint the correct picture. As will be the case when our A team faces up against The Ranji team and the SL A team.

Whatever you do have in mind, I think its important that you have different criteria to forecast any projection of an individual that is independent for different versions of the game. In other words, different standards to rate a player for three/four day matches, one dayers and 20/20.

For example, without mentioning the obvious stats ( Avg, SR ), one can focus on minor details that may help us understand whether that player is capable of different situations in a given match. If its a 20/20 game, not only his SR is important, but also, whether how many sixes or fours he has hit will be of equal importance. For one dayers, a middle order batsman must know how to accumulate runs at a brisk rate without having the luxury of infielders in the ring. Singles and two's are bread and butter for him.

There can be an infinite number of categories to rate a player if we are to be very specific. Some maybe meaningless for a casual observer, but I bet you, it will also give us a clearer picture.

Good luck..

[RazabQ]

What about Asad's WEIR? Paging Furqaan, Al Furqaan

[Ashfaq]

I liked that cricket version of money ball. But to see if a player is good enough for any perticuler situation, or which player is more suited to which job, we need something more than statguru. Normal scorecards wont contain enough information. And often the scorcard conveys wrong and misleading data. Is there any way around this problem?

Could their be any statistical representation of the true value of a cricketer? Is such a thing even possible? The best we can do is to measure the profficiency of a certain player in certain condition and change the batting lineup accordingly. Since in cricket we cant substitute, concentrating on off field variables is kinda pointless.So we should look to sort our players according to their mental strength and mindset towards cricket. That can be done by a mental evaluation. Any suggestions?

[Tigers_eye]

Any process can be statistically analyzed and continuous improvement can be implemented. This needs dedication and total commitment. More than full time jobs for multiple people. Monitoring the current process/program/procedure is the first step. None of us have that much of time. Even collectively we can't do that cause we don't have the resources. Had we had video footages/on-field observer we could do something like Rivals do for college sports recruiting.

[SGW]

Hi Guys - I've been following this thread with some interest - I'm a fan of the Money Ball methodology and thinking and have always wonderedif it could be done successfully in cricket.....cricket has so many variables that it is actually pretty hard to do..... Also as a coach and from a practical point of view - we need to make sure that any methods we use are not too time consuming or cumbersome that it becomes unusable....Anyway, as a starting point please have a look at this method that I've been using for sometime just to measure a player's actual statistical worth to the the team.....no doubt there are many more in depth methods but this one is pretty simple and seems to work reasonably well....see what you think (?)

PLAYER VALUE SYSTEM - MVP
It can be argued that cricket is a numbers game, Allan Border once used the phrase - “it doesn’t matter how, it’s how many”.

With this in mind we have developed a relatively simple system of valuing a players worth to the team using purely the results or “numbers” he produces in each match.

The player value system is an objective measurement of the players value to the team based simply on statistics alone and does not take into account other important aspects a good player will bring to the team such as brains, heart, desire to win, game sense and the many other intangibles that make up a good “team player”.

The player value system that we have developed measures each players statistical contribution during each match and allows us to see quite clearly who are the main contributors to the team in terms of pure numbers. Eight key areas are measured and weighted according to their importance to the team result.

Four Day/Test Cricket
For longer version matches we are measuring;
1. Runs scored
2. Balls faced
3. Wickets taken
4. Balls bowled
5. Runs conceded
6. Catches/Run outs


The weighting system works as follows;
1. Runs scored. Runs made are measured at actual, for example if a player scores 100 runs, his value to the team is measured at 100 points.
2. Balls faced. We feel that it is important in our team that our players bat for long periods, so we have weighted balls faced at 1.66. So for example if a player bats for 1 over (6 balls) he is rewarded with 1 point (or the equivalent of 1 run per over faced).
3. Wickets taken. Are valued at 20 points for each wicket taken. For example if a bowler takes 4 wickets it is the same in value as a batter scoring 80 runs.
4. Balls bowled. We feel that a bowler is providing value to the team by simply bowling the ball (as it requires a certain amount of effort just to do that) so we have rewarded the bowler by weighting each ball bowled at .25.For example each over bowled rewards the bowler with 1.5 points.
5. Runs conceded. If a bowler is simply bowling for the sake of bowling and is ineffective and conceding runs and not taking any wickets then he is quite simply, not providing value to the team. With this is mind we have weighted runs conceded at negative .35 for every run conceded. What this means is that if a bowler concedes 4 runs per over and does not take a wicket, his value to the team is about even. If he bowls at less than 4 runs per over and not take a wicket his value to the team is higher and is rewarded accordingly. If he can also take a wicket he is then rewarded further etc.
6. Catches/run outs are valued at 10 points each.

One – Day Cricket
The weighting system is a little more complicated for one-day cricket but the same principle of providing an objective measurement of a players value to the team based purely on numbers still applies.

For One-Day cricket we are measuring;
1. Runs
2. Strike Rate
3. Wickets Taken
4. Overs Bowled
5. Economy rate (Power Players)
6. Economy rate (outside power plays)
7. Dismissals involved in (e.g catches, run outs)
8. Runs saved

We've got an excel spreadsheet that calculates all this for us and all we have to do is put in the actual numbers that the player produces etc. It seems to work pretty well and identifies who is actually providing value in terms of sheer numbers....

Anyway, just some thoughts.............good luck and best wishes to all at BanglaCricket.
Cheers
Shaun Williams

[Beamer]

Excellent Coach! That is easy enough to calculate if someone is willing to do it. Since it doesn't require the intangibles ( heart, sense, IQ etc ), one can arrive at a reasonable conclusion by simply reading the scorecard. I think the 20/20 rating can follow the one day rating system as a starter, though the points given to SR for batsmen, and economy rate for bowlers would have to assume greater significance.

[Neel Here]

'The' shaun williams ?! wow !

interestingly, some of the metrics you mentioned is identical to what I had in mind when I started this thread.
the equivalence of 5wkt and 100 runs, for example.

the way of rating bowling contribution is pretty ingenious !

some other things we may want to look at in order to feature in intangibles like match conditions (say a fast bowler bowling on a greentop) and opposition quality :

comparison with combined averages(both teams) for the first one and own team for the second.
of course there is also the need to calculate standard deviation to determine consistency.

now that this thread is definitely up and running, I'll post more quantitative measures over the next few days.

[MarufH]

Originally Posted by SGW

Anyway, just some thoughts.............good luck and best wishes to all at BanglaCricket.
Cheers
Shaun Williams

WOW.. coach... how are you!!! so good to have u here... how is things working out for you in India?

Off topic question, is Enam jnr gonna bowl against BD A?

[bujhee kom]

Originally Posted by SGW
Hi Guys - I've been following this thread with some interest - ...
...Anyway, just some thoughts.............good luck and best wishes to all at BanglaCricket.
Cheers
Shaun Williams

Deraest coach Williams,
Salaam and greetings to you from the bottom of our hearts! Welcome to Banglacricket and we are so proud to have you here among us! We miss you very dearly coach Williams and we know you are indeed a true friend and genuine well wisher of Bangladesh cricket and the people of Bangladesh! Please please do visit us more often and tell us about Maharshtra Cricket and about our BD boys of your old team! We love you coach Williams!

[RazabQ]

Shaun, has posted in the past here too. He's a genuinely decent fella in my books.
On to the topic at hand, I feel that Shaun's spreadsheet is not rewarding wicket taking enough. On 4 day matches/Test matches, while batting can get you a draw it's the bowling that will win it for you. I'm curious about how the weights that Shaun mentions were derived. Did you guys do some sensitivity analysis and then see which weights backed up what common sense was dictating? Also I wonder if runs scored or wickets taken or runs conceded should have a degree of difficulty or conditions parameter. A 50 scored in a flat, baked pitch after the top order has blunted the opposition is not the same as say opening on a green top.

I would think one would need an oppostion by opposition and ground by ground analysis to get there. In the moneyball example, the type of ballpark is factored in. Also for fielding, they have zones. Each zone has had the probability assigned - of a certain player hitting there, of an outfielder's effectiveness, etc.

[SGW]

I'm interested to see what else you guys might think about how we can measure players OBJECTIVELY.....it's hard to measure the intangibles......because there are so many, and still rely on "opinion" as opposed to "fact"...

Anyway, off subject for anyone who might be interested.....Enam will not play for Maharashtra in the upcoming series....his first priority is to Bangladesh cricket of course.....

As for Maharashtra, things here are going well....we have maintained our position in the Elite League of the Ranji Trophy and we currently have by far the youngest squad in the competition (average age is 21.5)....we are in the process of building a strong side for the future and part of that building process is giving our players exposure to international tours such as the upcoming series in Bangladesh. We will also be touring Australia in June/July after visiting Bangladesh...

So we're really looking forward to coming to B'desh and taking part in what we hope will be a very competitive series of matches....personally I'm also looking forward to catching up with many of the players again and seeing some old friends in Bangladesh once more......

Should be good fun

Best Regards
Shaun

[Neel Here]

A 50 scored in a flat, baked pitch after the top order has blunted the opposition is not the same as say opening on a green top.

this is precisely why I started this thread, to find if we can quantify at least some aspects of the intangibles.

I've some ideas, will post in detail later.

[SGW]

Good point about the bowlers not being rewarded enough...... we played around with that weighting quite a bit and what we realised was that in most cases at first class level and above....if a bowler actually took 5 wickets.....1. he had to bowl quite a few overs and 2. his economy rate was normally half decent.......so what that meant was that he would actually be scoring "points" in those areas as well as gaining for the actual wickets he took, which gave some extra reward over the guys who scores 100 runs.....anyway, look it's certainly not a flawless system by any stretch....but it does give us an OBJECTIVE indication of who is actually providing VALUE to the team...and it's quickand easy to punch in the numbers......sometimes we can get carried away with what we THINK is happening with certain players, but then the numbers can show us a different story (as do averages, aggregates and strike rates anyway)......the reason I posted this here was not to get a critique (which is ok) on this little system as such (I already know it has some limitations) - BUT I was hoping that some of you mathematical/probability/etc cricket loving gurus out there might be able to come up with some more ideas as to how we can do this...1. better 2.reasonably simply 3. and accurately.......
All for now guys......Shabash Bangaldesh !!!
Cheers
Shaun

[RazabQ]

Shaun, Neel:

We can use standard deviations to "objectify" some of these subjective things. For e.g., for an evaluation of grounds, look at the deviation curve of the averages of top 7 scorers each innings. Then take the 1 plus deviation off the mean. This should give you a bench mark for how a batsman should fare on this track. Perhaps that can be used to identify a grounds based parameter or multiplier for batsmen's runs scored? I'm not a stats guy and have too much work to doodle on Excel right now.

[RazabQ]

Shaun - wasn't a critique. I was trying to understand any unstated or implicit assumptions you guys had made before making any suggestions.

[Tigers_eye]

Originally Posted by SGW
I'm interested to see what else you guys might think about how we can measure players OBJECTIVELY..........

Should be good fun

Best Regards
Shaun

Dear coach,
Break down the bowling figures for one day games. Boring overs in the middle gets less value. Create normal spread. Anyone outside the spread needs more care (wrong side) and reward (right side). Bowling 0-10 overs and after 45+ overs needs more mental strenght than 20-40 overs. You can come up with a formula.

As RazabQ has already mentioned pitches must have some values and ranking. This should affect both batsman and bowlers.

For T20, extra balls are criminal. Front foot noballs must be punished.

For test/4 dayers: Anyone bowling 8+ over spells on the dot should get a reward. I love the ball faced equation in to play. Batsmen need to be categorized in to few groups (A) Top orders (1-4), (B) middle order (5-7), (C)tail (8-11). Not outs with the side getting allout values more than notout with declarations. A - notout should be valued more than B and C notouts. B should be > C.

++++
Subjective: Make videos and seek a leader. Vocal and performance wise. Alert, Agile etc. Then while comparing them make sure to get one as your "chosen one" who has an off-field interest. Be it hobby, social work, or a good family man who does activity with them.

My two cents.

[RazabQ]

Good to see TE espousing the standard dev approach too. If I have time, I'll doodle around to see what kind of web services CI or CricArchive provide. Else you can just create a local instance of all the stats on your hard-drive.

[Tigers_eye]

Originally Posted by RazabQ
...Else you can just create a local instance of all the stats on your hard-drive.

There is no other alternative than this. Not only these type of analysis will give a better objective picture but can also alert the weak points of a person, position, team, mgmt. So the question becomes how easily one can get the raw stats?

[RazabQ]

I have to believe if not free than cheaply ... also wife & web service could work ...

[al Furqaan]

aight guys, i'm here...

good thread neel, i am definitely the biggest fan of numbers here or among the biggest. my thoughts:

1) i'm still a fan of my WEIR system that i devised about a year and half back
2) i am also a fan of standard deviations in regards to measuring consistency
3) there is no way to rate talent, only productivity. potential and talent are often hidden and come out late or not at all.

************************************

as of now, i have rating systems in place for tests and ODIs for both batting and bowling. i would like to include wicket keeping as well, and also fielding (catches and run outs) but that may be a bit more difficult. i have now accepted the importance of T20 cricket, and before the world cup i will have ratings up for that, inshallah.

i haven't had much time to update my WEIR, but i plan on doing it.

i have also been compiling a stats database on excel since summer of 2007, however that project has also stagnated. i do fully intend to resume work on it as soon as next week.

i am also excited about the prospect of working with other BC stalwarts, but I haven't seen much interest in that regard. what i need are honest opinions on my WEIR system, because asides from that, BC has no other rating system in place.

as of now i felt that incorporating strike rates in Test batting are not completely irrelevent and so i have waited it at 50% the value it has in ODIs. i guess in T20Is i will give a double value or 150% value.

[al Furqaan]

Originally Posted by RazabQ
Good to see TE espousing the standard dev approach too. If I have time, I'll doodle around to see what kind of web services CI or CricArchive provide. Else you can just create a local instance of all the stats on your hard-drive.

i was doing this on excel, albeit manually. i'm not sure if this is what you are taking about. i've made significant progress. i'd estimate that i have maybe 30-40% of all our top 20 players stats compiled in Tests and ODIs. and thats taken me not that long, though i started the project 1.5 years ago.

[Neel Here]

furqaan bhai link to WEIR please ?

[al Furqaan]

not to be a downer, but i think we're getting a little bit ahead of ourselves, unless someone has some magic program which makes the data collection relatively easy. incorporating pitch and venue is great, but it seems time consuming and, at this point, a minor detail. and thats not including that certain venues can produce different types of pitches.

[Neel Here]

i was doing this on excel, albeit manually. i'm not sure if this is what you are taking about. i've made significant progress. i'd estimate that i have maybe 30-40% of all our top 20 players stats compiled in Tests and ODIs. and thats taken me not that long, though i started the project 1.5 years ago.

better write the thing in a code and start putting the whole data in a data file.

after the first hassles it will simple become plug and play !

We can use standard deviations to "objectify" some of these subjective things.

yes razab, I mentioned SD in my second post.

comparison with combined averages(both teams) for the first one and own team for the second.
of course there is also the need to calculate standard deviation to determine consistency.

but that is for determining performance over a period of time.

for the moment however, let me propose a system to quantify limited overs batting performance for a single match, taking into account quality of opposition and overall ground conditions.

I have used the simple concept of weightage to assign 'importance' to some parameters.
the system is kept as simple as possible for ease of use.

the following posts would hopefully explain everything.