Woodhill's moneyballing it

[Blocky]

Absolutely. Big data in those American sports is an integral part and cricket should be just as valuable. Only problem will be budgets. If India ever cottons onto it, the rest of the world could be ****ed.

That's what WASP basically is, statistical modelling using the variables it has available to it, the underlying models give NZ Cricket much more in terms of game insight, however if the players don't execute, or can't execute, it has no way of understanding or interpreting it.

It's not even worth considering for cricket, too many different variables, again, it's not like baseball. A pitch doesn't touch the ground, must be inside a certain area, can only be hit inside a certain area against a field setting that doesn't change, etc.

 

[vic_orthdox]

I think batsmen who are good at minimising dots will be more reliable over the long run. Sure, Carlos Braithwaite smashing 4 sixes off Ben Stokes to win the world T20 is memorable but there's an alternative universe where he swung and missed a couple of times, mishit one down long on's throat and England win by 12 runs. Whereas someone like Dhoni, who was less boundary reliant, is one of the greatest finishers of all time.

The effective difference between a boundary and a dot/single is a lot bigger in effect in a shortened game where there is a limited number of balls, i.e. 120 balls to score vs 300.

The Windies have a huge advantage in that they have about eight guys in their T20 team that can take an over for 20+, not just a couple like you see in most domestic T20 games.

If you had the Windies playing against the Windies, then the number of balls scored off becomes more relevant because you have two same skill sets playing against each other, and their inherent advantage of so many top class boundary hitters is neutralised.

 

[vic_orthdox]

It's interesting stuff and I'm surprised it's taken this long for this sort of analysis to make its way into T20 cricket, particularly the IPL.

Guys like Pollard I've never rated, even in the shortest stuff. I've always thought the likes of Dhoni and Mussey were fabulous limited overs batsmen because of the way they minimised dot balls, hustled 2s and 3s where they shouldn't have and could still smash it over the ropes when the situation called for it.

Even in the shortest stuff perception is greater than reality and my gut says that a guy who minimises dot balls is more value to a team than a flashy big hitter.

A lot of it is used for matchups - trying to ensure you are bowling the right bowlers to the right batsmen. E.g. bowling spin to Tim Paine in the first six overs last two BBL seasons was far more effective than bowling pace, and he'd get behind the expected rate and get out.

Same with Chris Lynn and his SR vs spin and pace being such a huge differential.

You'll hear Ponting mention more about match-ups in the Channel 10 coverage of the BBL, something derived from his time in the IPL.

 

[Agent TBY]

Lol.. not sure why its wrong to point out that while you can use these stats as a predictive measure and maybe give a hint to the bowler when you think a batsman is likely to go for his release shot etc., cricket, as a game, has so many more variables than baseball that its absolutely stupidly silly to try and pick a team out of sabermetrics type stats. The thing about replacing Hastings with a death bowler, that kinda stuff I definitely can agree with. That makes sense, but when they start talking about picking guys who can run 3s and picking guys who wont play a slog sweep as the release shot etc., it goes pear shaped. The game is, was and will always be about skills and being able to adjust your instincts to adapt to different surfaces, atmospheric conditions and bowling/batting line ups etc. Even in T20s, you cant just call the play like they can in baseball. Its what makes cricket great.

The difference between Baseball and Cricket is that the level of statistics and underlying variables in performance when you're not just left with three strikes against a pitcher who can only find the strike zone or throw a ball is huge. I think most teams don't get the fact that sabermetrics works awesomely well due to how confined an actual at bat is from both a pitching and striking perspective.

That's what WASP basically is, statistical modelling using the variables it has available to it, the underlying models give NZ Cricket much more in terms of game insight, however if the players don't execute, or can't execute, it has no way of understanding or interpreting it.

It's not even worth considering for cricket, too many different variables, again, it's not like baseball. A pitch doesn't touch the ground, must be inside a certain area, can only be hit inside a certain area against a field setting that doesn't change, etc.

Yes, cricket is not like baseball, but it's much closer to it than it is to basketball, which is a more fluid game where analytics still makes an impact.

Putting aside the subject of the article, who might just be using it as a ploy to big up his brand, you would be foolish to disregard the advantages you could gain by zeroing in and acting on the right kind of data (this is important).

Obviously it will never be the be all and end all, and data mining in cricket is still in a very nascent stage, but saying it'll never work is a definite mistake.

 

[Blocky]

Again, if you understand what statistical models are actually capable of doing versus what the hype of "Big Data" is, then you know why Cricket is ultimately near impossible to game the same way that Basketball or Baseball are.

Basketball, court conditions never change, you have three outcomes of having the ball - shoot, pass, turnover.

Cricket, you've got pitch conditions, condition of the ball, match conditions that vary widely, concentration that needs to span over multiple hours with you in the hot seat, etcetera.. It's just not a game that much can be inferred from, video analysis and scrutiny of particular flaws in a players game is far more important, being able to assess plans based on how someone plays the ball, rather than the results are more important

 

[Agent TBY]

And all a batsman does is play or leave.

Of course cricket is different, every sport is. You don't think basketball experts said the sport was too different for analytics to infer anything from it? And yet it did.

Like I said, it'll be a process to find the right kind of relevant data that will actually prove useful in cricket, but someone will find them. And vic_orthdox just gave us some examples of exactly that kind of analysis already being used.

 

[Blocky]

A batsman can play multiple shots to the same delivery, the delivery can bounce/swing/seam in multiple ways from the same starting position, the field can change markedly meaning that a shot that would normally be good becomes poor, the pitch detoriates over time meaning the ball does different things, the ball detoriates over time meaning it acts differently.

Vic_Orthodox looked at match ups, that's not statistical modelling, that's saying "Kevin Pietersen struggles against Left Arm Spin based on his dismissal record" - statistical modelling is "Kevin Pietersen is susceptible to being dismissed in this particular over, if you bowl four balls on the off side before shooting one at the stumps, based on the pitch condition, the ball condition, the bowler, where Pietersen is in his innings, what Pietersen did in the last over"

 

[vic_orthdox]

 

[indiaholic]

Vic any examples of how it has helped with field placings? That was an area I felt would be the obvious candidate with players having gps trackers on them and stuff..

 

[Blocky]

I still say you'll get a lot more out of guys who have a feel for the game, looking at the match situation and forming plans based on what they're seeing a long time before you'll get any value out of predictive modelling and artificial intelligence.

I'm actually surprised that more teams don't have known tacticians sitting in their midst, solely reviewing video footage and formulating plans for the bowlers and field settings based on what's currently happening out there. There has already been a lot of investment by NZ Cricket and English Cricket in statistical modelling and it hasn't really bore fruit yet.

 

[vic_orthdox]

I'm actually surprised that more teams don't have known tacticians sitting in their midst, solely reviewing video footage and formulating plans for the bowlers and field settings based on what's currently happening out there. There has already been a lot of investment by NZ Cricket and English Cricket in statistical modelling and it hasn't really bore fruit yet.

This was basically David Saker's remit/strength as a bowling coach for England.

 

[Blocky]

This was basically David Saker's remit/strength as a bowling coach for England.

As much as teams talk about the need for a good captain, ultimately I'd much rather captains of today became good man managers and got the best out of their players (i.e Vaughan and his ability to get the best out of Pietersen ) because I think all the talk about their tactics is bullshit, I think ultimately tactics should be done off field considering they've got far more information and can read the game far more.,

 

[honestbharani]

Yes, cricket is not like baseball, but it's much closer to it than it is to basketball, which is a more fluid game where analytics still makes an impact.

Putting aside the subject of the article, who might just be using it as a ploy to big up his brand, you would be foolish to disregard the advantages you could gain by zeroing in and acting on the right kind of data (this is important).

Obviously it will never be the be all and end all, and data mining in cricket is still in a very nascent stage, but saying it'll never work is a definite mistake.

But cricket is not just fluid. It is impacted by external factors to a degree that is miles ahead of baseball or basketball. That was the whole point. Going for replacements based on the players' role in the team rather than his actual skillset bucket is good. Match ups are ok, but still very much prone to the human element but going beyond that is just asking for trouble. You are much better off focussing on improving the actual skills that you will need in the game and ensuring the players get more game nous through their various coaches, like bowlers being able to tell when a batsman is going for a big shot based on his movements and style at the crease rather than an algorithm calculating the average balls to his big shots from his last 25 innings against left arm spinners in T20s held in Bangalore. That is the whole point. You really cannot "moneyball" cricket, at least not based on statistical modelling alone.

 

[honestbharani]

This makes much more sense than Moneyball-ing cricket. Obviously what cricket needs is more Sherlock Holmes and less Billy Beanes.

 

[vcs]

But cricket is not just fluid. It is impacted by external factors to a degree that is miles ahead of baseball or basketball. That was the whole point. Going for replacements based on the players' role in the team rather than his actual skillset bucket is good. Match ups are ok, but still very much prone to the human element but going beyond that is just asking for trouble. You are much better off focussing on improving the actual skills that you will need in the game and ensuring the players get more game nous through their various coaches, like bowlers being able to tell when a batsman is going for a big shot based on his movements and style at the crease rather than an algorithm calculating the average balls to his big shots from his last 25 innings against left arm spinners in T20s held in Bangalore. That is the whole point. You really cannot "moneyball" cricket, at least not based on statistical modelling alone.

Nicely put. This is what I feel as a cricket fan as well. Would be depressing if the sport turned into a paint by numbers exercise. The technical aspects of batting and bowling are what draws us to the game, and keep us hooked.

 

[Pratters]

Using data intelligently is good. However, one shouldn't use it blindly without understanding what the data means. For instance, if a player averages 55 in England, it means that he can play swing well, in all probability.

 

[vic_orthdox]

The "moneyball" aspect of cricket comes along with franchises, shorter form cricket where the situations arise more commonly so you can make comparisons. And also where you have greater flexibility to change your squad, transfers trades etc. rather than where you have to choose from what your nation has to offer as a player pool.

E.g. we need a keeper who can perform well at number 7, as our top order is settled - no use getting James Pierson who has only performed at the top of the order but has better stats, we'll get Peter Nevill instead.

The most it will come into international cricket is in squad selection for World T20, trying to ensure that a lot of options are provided to the captain/coach on tour to help ensure they can counter the best the other teams can provide; no left arm orthodox at Chris Gayle, but you might have it against McCullum for NZ.

 

[vic_orthdox]

But cricket is not just fluid. It is impacted by external factors to a degree that is miles ahead of baseball or basketball. That was the whole point. Going for replacements based on the players' role in the team rather than his actual skillset bucket is good. Match ups are ok, but still very much prone to the human element but going beyond that is just asking for trouble. You are much better off focussing on improving the actual skills that you will need in the game and ensuring the players get more game nous through their various coaches, like bowlers being able to tell when a batsman is going for a big shot based on his movements and style at the crease rather than an algorithm calculating the average balls to his big shots from his last 25 innings against left arm spinners in T20s held in Bangalore. That is the whole point. You really cannot "moneyball" cricket, at least not based on statistical modelling alone.

You can't "moneyball" any sport on statistical modelling alone. It's about working out how to solve a problem, using what numbers tell you - go with what history has said will work in the past, we all have memory biases. Of course that's not perfect, but you can live with the decision easier knowing that you got beaten by a person playing better on the night.

And it's no different to other sports - a 25% three point shooter in basketball hitting 67% for the night, for example. If he pulls that out then fair enough, well played.

 

[Blocky]

I wouldn't even call it Moneyball - I laugh at every single IPL auction when an unknown Australian gets picked up for big money, when conversely they could get a proven T20 performer across multiple leagues from NZ, South Africa or elsewhere.

 

[dontcloseyoureyes]

People don't understand what 'moneyball' actually is.