Monday, 14 January 2013

Form is temporary?

Faf du Plessis
“Form is temporary, class is permanent.” It’s a cliché that is fairly frequently repeated. But can we measure form.  But can we measure form, and are some players more prone to big swings in form than others?  In the press conference after day 2, Faf du Plessis said

“It’s nice to average 110. But I understand it’s a good time for me and I need to score as many runs as I can because there’s going to come a time when you don’t score as many runs.  I’m in a little bit of good form and just have to make the most of it.”

It made me wonder if du Plessis was actually a player who had been prone to streaks of form.

To do this I had to find a method that would differentiate between a player being streaky and a player being inconsistent. 
An inconsistent player might have scored: 108, 41, 8, 0, 29*, 37, 6, 114*, 5, 4, 102.   
Whereas for a streaky player that might have read:  6, 0, 37, 108, 41, 102, 29*, 114*, 5, 4, 8.  
One of them has had a hot patch, while the other one has just occasionally scored runs.

In order to test for this I looked at the 15 innings rolling average, and compared it to the cumulative career average.  (both of these terms are explained in the glossary) I found the difference between these two averages at every point in a career where they both existed and then averaged this out.

What this gave me was how far away (either above or below) from the career average a players recent average was. This then gave me a number to describe how streaky a player was.  The difficulty then was knowing how to interpret this number.  Players who tended to get a larger variation in scores would get a higher average difference than a player who was equally streaky, but much more consistent.

As a result I then found what that number was for 1000 random arrangements of the batsman’s innings, and how the average difference compared to those.  This allowed me to accurately look at a player’s streakiness.

I decided to look at first class innings, as du Plessis hasn’t played sufficient tests to be able to tell anything yet.  Also there is a purity to first class results that comes from the similarity of situations that a batsman find themselves in.  In limited overs cricket, there are often times where the requirements of a batsman are significantly different, while in first class cricket it’s rarely a bad idea to just score as many runs as possible, at whatever pace suits the player.

I wanted to have some players to compare him to, so I asked some of the South African journalists in the box who they felt was a streaky player.  The first suggestion was Morne van Wyk. A suggestion of another inconsistent player was Neil McKenzie.  I also wanted to look at a particularly consistent player, so I chose Stephen Fleming. The first thing that I did was to look at their 15 innings averages.  I also looked at 7 innings averages, but they seemed too random to be useful. Here are some graphs of each players 15 innings averages and cumulative averages:


We can see that Faf has had a remarkable run recently, and also why I was suggested to look at Morne Van Wyk. He went on some remarkable streaks of both good form and terrible form.

While McKenzie was obviously inconsistent, there weren't many extended patches of good or poor form.  Instead it looked like he was just an inconsistent batsman.
Fleming was remarkably consistent at the start of his career.  He was described as someone who averaged 40 by regularly scoring 40.  Later on in his career he started getting more big scores, but generally he remained a batsman who was fairly consistent.

We can also look at their consistency by looking at their 15-innings averages in a box and whisker graph.


We can see here that du Plessis has the largest range, by quite a margin, but that Morne van Wyk has a significantly larger interquartile range.  

After applying the randomisation technique, there are a couple of interesting results. The randomised results all came out with a very close approximation to a normal curve, so I’ve used mean and standard deviation to find the probability of a batsman being prone to high swings of form. A player who is always on a streak will have a value of 1, while a player who is completely consistent will have a value of 0.  Here are the 4 batsmen that we have the graphs of.

Stephen Fleming0.135
Neil McKenzie0.136
Morne van Wyk0.984
Faf du Plessis0.783

However, if we remove the start of du Plessis’ good run of form and re do the analysis, it comes out completely differently: 0.212  

In other words du Plessis was a quite consistent player until he started on his incredible run where he has averaged over 99 for the past 17 innings. It might be that he is finally having a hot spell, or it might be that he has actually just got a lot better.  This might actually be part of a new, consistently good career.

It will be interesting to find out. I hope for his sake that any time where he doesn’t score as many runs will be shorter than the times he spends at the top.

No comments:

Post a comment