Skip to content

Advancing cricket statistics

October 11, 2013

I read a very interesting piece on cricket stats this week and my large number of regular readers will know this is a subject very close to my heart. It was clear from the feedback on that article that people fell roughly into three groups – those that hate statistics, those that think the existing statistics do the job fine and those that think the existing statistics do as much concealing as they do revealing. It is pretty obvious what group I fall in.

Without doubt traditional cricket stats like averages and strike rates only tell part of the story. It can be an important part of the story but they are completely free of context and therefore it can be very easy to draw the wrong conclusion on them if you don’t dig a bit deeper.

One thing I really hate is the suggestion that things like dropped catches or poor umpiring decisions balance out in the end because I don’t think anyone has ever done the work to know if they do or not. It is just lazy speculation and I believe that it is probably completely untrue.

Any developments to cricket statistics aren’t about replacing more traditional statistics they are just designed to complement existing statistics by adding a further layer of context to them whether that context is luck, match state, quality of opponent or any number of things you need context to be better informed.

With that in mind here  are a few concepts for statistics that I would love to see someone implement:

Dot balls in T20

Seriously how hard is that one? The primary limiting factor for the batting team is not wickets, it is deliveries. About 80% of completed innings come to completion because the overs have run out, not the wickets. Therefore every dot ball is a wasted opportunity to score. When you also consider that the team with the least dot balls win about 80% of matches (most boundaries is about 85% which isn’t significantly more) it really surprises me that wasting deliveries through not attempting to score isn’t a bigger talking point. If you wanted to jazz up dot balls I recommend my Opportunity Score because I am into self promotion.

Contribution Scores/Value Runs

In my T20 formulae article I looked at Contribution Scores which is fairly similar to Andrew’s Value Runs (his blog is here) in that it takes a batsman’s context free performance and tries to add the context of the match circumstances to give you more idea of what the batsman has contributed to his team’s performance. Not every 56 from 37 is created equally but in the eyes of traditional stats they are the same thing and that ought to change.

Wicket value

Bowling can be quite difficult to measure because bowlers are subjected to a lot of external factors, like what the batsman does and what the fielders do to help out.

One measure that certainly would be easy and useful is to provide some information about the quality of the batsman that have been dismissed. A very simple method would be to assign a value to each position in the batting order with a higher number to represent the top order and a lower number to represent then add the numbers of the batsman dismissed by a bowler together to give a wicket value. Say two bowlers take 3-45 but one dismissed an opener, the number four and the number six for a wicket value of 34 while the other bowler dismissed nine, ten and eleven for a wicket value of 13. The main issue with this is first it assumes that Bangladesh’s batting order is as good as South Africa’s and it also assumes that a team’s best batsman are always in the same spot in the line-up.

A more complicated method would be to provide a combined average of the players dismissed as a wicket value. Better batsman have higher averages so chances are if you dismiss the top order you would have a higher wicket value than the bowler that dismissed the tail-enders. This method also accounts for the fact that taking 3-45 against South Africa’s top order is a greater achievement than the same analysis against Bangladesh. The main flaw in this method is you would need to provide some sort of weighting to the combining of the averages to avoid strange results and then weighting in itself can cause some strange results…. still I am sure it isn’t something that a smart person couldn’t address.

Bowler/fielder and bowler/batter wicket attribution

If Ross Taylor chops a delivery on to his stumps off James Pattinson does Pattinson deserve the credit? He might deserve some for bowling a tight line or length but at the same time it could just be a crap delivery combined with an even more crap shot combined with the good luck of the ball finding the stump. It certainly isn’t as good a delivery as the ball that hits the top of the off stump but history records it just the same.

It couldn’t be that hard to attribute the dismissal to the bowler or the batter can it? Well I suppose it can if the bowler has bowled to a pressure plan and picked up the wicket with the wider delivery, but at the same time you could argue that the batsman shouldn’t have fallen for the trap so it is still on him.

The same goes for looking at a fielder’s contribution to a wicket. If a fielder does something unbelievable does it really seem right that it just goes down as caught Mason bowled Nethula? Not really. But on the flipside…..

Opportunities generated

How fair is it for a bowler to get an edge only to see a regulation catch spilled by first slip? The bowler has played his part and been let down by his fielder but this fact will forever be lost. When you are using stats to measure performance it is completely misleading to judge a bowler based on a factor he has no control over.

Make no mistake there are bowlers that are lucky *cough* James Pattinson and those that are unlucky *cough* Trent Boult wouldn’t it be great to be able to quickly measure part of the impact of that luck? You could do some wonderful things with opportunities generated and wicket attribution compared to actual wickets to measure what luck a bowler actually has.

The chances generated adjusted average

This is basically the same as opportunities generated but focusing on the batsman. I mentioned this about Hamish Rutherford’s debut innings, by a traditional measure is he scored 171 and was dismissed once, but he also offered chances on 52 and 64 that weren’t taken. Those catches being dropped had nothing to do with Rutherford but he effectively gets all the benefit. Under the chances generated adjusted average we would add 171 runs and that he should have been dismissed 3 times rather than just once to give him an average for the innings of 57.

Wicketkeeper value

As raised by the ODT’s Adrian Seconi last year in a Notes from Slip column, no position in cricket is as under-represented as the wicketkeeper by traditional statistics. Considering their overall contribution to the game they get dismissals and byes. That is it. The day after it happens nobody but the keeper remembers whether the four byes went between his legs or was a terrible delivery that should have been called wide everyone else just sees four byes as a black mark on the keeper. There are regulation catches and there are tough catches but they are all the same in the scorebook. How hard would it be to generate a ratio of chances to dismissals and further incorporate wicket attribution and bye attribution into that? It would certainly give an understanding of who the better glovemen are.

That is seven areas that traditional stats could be improved by complimenting them with new statistics. The surprising thing is that most of these things aren’t very complicated and only require a small bit of additional information than currently exists and yet the improved information they provide could be really significant to our understanding. All it would really take is someone to do some leg work. If I didn’t work fulltime I would be that someone, although if I didn’t work I would probably spend a lot of time watching TV and playing games because I am not a high motivation guy, but I digress.

Really this is just a starting point. These are all too basic to even qualify as being advanced stats. The rise of statistical analysis in US sports has certainly made its way into what some teams are doing even if that progress is slow. England employ a statistician with a maths degree from Cambridge who is also a qualified cricket coach to crunch numbers and use hawkeye to do delivery mapping and sooner or later what they are doing will seep out to other teams and then down to a wider audience.

Sky gave a crack at some advanced statistics last summer when they introduced a win prediction measure into their coverage, I think it was called WARM. So sooner or later I think a lot of this analysis will be mainstream to the point that avid cricket followers know all about them.

In the meantime these type of discussions certainly help the open minded among us to further our understanding of the greatest game so I thank Andrew for getting me started.

RTC’s Prospect List

With Ish Sodhi’s test debut in progress it is time to release my to be annual prospect list of the five best prospects under 23 yet to play for New Zealand. First, I swear I formed this list in April. Second, my assessments are based on reading articles and analysing what I have just told you are context free stats and I probably shouldn’t draw the conclusions that I do but I have anyway. I have seen a few of these players with my own eyes but I am not a gut feel selector so that doesn’t play into the list much if at all.

Ish Sodhi (20) – right arm leg spin

Just the fact that he is a leg spinner gets me pretty excited. His current first class stats might not be especially flattering (27 wickets at 52.14) but his figures were hurt by the NZ A tour of India and Sri Lanka where he took 7 wickets at 73 and I have no doubt that the experience was far more valuable than the cost to his bowling average. He is also a decent batsman that will slot in nicely at number 8 or 9 and have the potential to contribute. He will have some grown pains but is easily the most exciting spin prospect since Vettori.

Will Young (20) – right hand middle order

Former captain of the Under 19 side (Ish Sodhi was also in that side and hit 12 runs from the last three balls to win a quarter final against the West Indies) he hit his maiden first class century in his 8th first class game against a bowling attack with three current or former internationals (although one of those was Jeetan Patel). Will hopefully develop into the long term partner to Kane Williamson in our middle order.

Jacob Duffy (19) – right arm medium fast

Duffy debuted for Otago while still at school and has already played 8 first class, 5 one day and 12 T20 games. He is tall and quick and opens the bowling in all formats. His numbers aren’t all that flattering but I think they will improve this season. Never before have New Zealand been this blessed with depth and prospects in our fast bowling and Duffy should be a great addition to the Black Caps down the line.

Matt Henry (21) – right arm medium fast

Matt Henry is a player that I expect to make a debut for the Black Caps in an ODI at some point this season. He is good across all three formats (first class bowling average of 20.73, one day of 19.24 and T20 of 35.07) and is another reason why we should be excited about the potential of our fast bowling stocks.

Daryl Mitchell (22) – right hand middle order

Has the worst cricinfo profile photo of players on this list as he seems to be channelling Mitchell Johnson which is never a good idea. Mitchell doesn’t appear to be a strong performer in the limited overs formats but in first class cricket he has 1 hundred and 7 fifties at an average of 42.47. Struggled in the NZ A tour with first class returns of 0, 41, 0 and 10 but he will be better for the experience.

So that is me for. Now that cricket season is upon us I will be attempting to write a bit more regularly but we will see how it goes.

Any ideas for new and innovative statistics please let me know because I would love to hear about them.

Advertisements

From → Cricket, General

2 Comments
  1. the use or misuse of batting not outs in determining a batting average is another example of statistic stupidity

  2. I think there is certainly some room to improve on the not out. I can’t see any value in being not out when the overs expire in a limited overs game so why should the batsman be rewarded? There is some value in being the not out batsman when a score has successfully be chased down so maybe the reward is appropriate. Either way I agree that as it stands it is not only statistical stupidity it is a basic nonsense.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: