Wednesday, 1 June 2016

Playing loose and fast with stats - a rejoinder to Cricketing View


A wag once remarked that when you have a hammer in your hand, everything looks like a nail.

Nowhere is this truism more evident than in the shoddily written clickbait article by Karthikeya Date on ESPN Cricinfo following the IPL final.

Let us examine the claims made in the article.

Date points out, rightly it must be said, that Kohli scored 54 off 35 balls and therefore scored his runs at less than the asking rate at the start of the innings.

To buttress his point further, Date picks on a stat, which was that Kohli scored only 15 runs off his first 18 balls, something Date contends a “true-blue T20 hitter” would not have done when his team was chasing 10.45 per over to win.

The stat itself is indisputable – but this is where the hammer-nail analogy comes in. 

A piece of reasonably well-known cricketing wisdom is that when batting alongside a partner who is batting well, hitting the ball and looking in total control, the worst thing a batsman can do is to deny him the strike. It upsets the rhythm of the set batsman and generally hurts the team. 

Added to that is another significant stat that Date missed (hopefully, and did not just gloss over) – 4 overs into the RCB chase,  Gayle had faced 19 balls and scored 32, while Kohli had played a mere 5 balls of which one yielded a boundary. 

When you consider these two facts together, it made sense for Kohli to defer to his more aggressive and already settled partner which was what he did. 

Date then offers this gem “When the asking rate is ten runs per over over 20 overs, there simply aren't many options. This is not a problem of the imagination, it is a compulsion of arithmetic.”

At this stage Date declares AB de Villiers the best RCB player at IPL 2016. 

It is not entirely clear what that claim is based on but consider the stats – which Date did not bother to provide (source: ESPNCricinfo)

Virat Kohli
AB de Villiers
Runs
973
687
Strike Rate
152.03
168.79
Average
81.08
52.84
Fours
83
57
Sixes
38
37

Green highlight denotes the better performance – and considering that the only place where de Villiers outscores Kohli is on strike rate, it is difficult to accept, solely on the basis of that stat, the contention that de Villiers, in this IPL at least, was the better batsman.

In other words, and to respond to another of Date's unbacked-by-data claims, a team of 11 Kohli's would beat a team of 11 de Villiers' by about 320 runs (assuming each batsman on either team maintained the averages above).

On that count – strike rate – by the way, Date offers us the fact that Kohli played 233 balls more than De Villiers with the suggestion that this goes against the former. 

It would be interesting to know why this is considered a negative. 

The fact that Kohli played more deliveries could - and does - indicate a number of things:
  • He came in to bat earlier than de Villiers each time, which was the role he was presumably meant to play,
  • On average he batted 14.5 balls longer than AB in each innings,
  • He got out cheaply less often than de Villiers did.
  • In other words, he was generally more consistent. 
None of these stats makes those 233 extra balls that Kohli played look wastefully spent, as evidenced by the 286 additional runs that he managed off them. 

Date also tells us that “He (Kohli) used up more than 25% of the deliveries available to his team and scored at less than the overall asking rate.”

As mathematical facts go this one too is indisputable. 

But even here Date does a great job of presenting a half-truth.

When Kohli got out, RCB needed 69 off 43 balls i.e. an asking rate of 9.62 runs per over. 

This means two things:

One, Kohli’s opening partnership with Gayle (who scored 76 from 38, at a strike rate of 200) had reduced a 10.45 starting rate to an asking rate of 9.62. This is where Kohli’s initial preference for giving the strike to a rampaging Gayle paid dividends. 

(On a side note, when Gayle got out RCB needed 95 off 57 i.e. an RPO of 10, which by the time Kohli got out was brought further down to 9.62)

Of course, there is also the small matter that by the time Kohli was out RCB had scored more than two-thirds of the runs RCB were chasing, which meant that the remaining batsmen had to score less than a third of the original target. 

Two, it means that the remaining 8 batsmen that followed Kohli and Gayle mustered a total of 61 runs between – that was just 7 runs more (in 13 balls more) than Kohli managed. 

Essentially that meant that 
(i) the remaining batsmen were no longer looking at the higher rate that Date speciously suggests Kohli’s seemingly slow batting left them with, but also that 
(ii) they even failed to score the already reduced RPO that was required of them with 69 needed off 43.

But, by far the most damning indictment of the lazily written hatchet job on Kohli is what the piece yet again, forgets to mention.

Even making allowance for the fact that the Bangalore wicket was flat and full of runs, RCB’s problems were not created merely when they batted.

Their bowling line up consisted of three relatively inexperienced domestic bowlers in the form of Sreenath Arvind, Iqbal Abdulla and Yuzvendra Chahal. 

But in keeping with their strategy, they also chose to have their two most experienced international bowlers – Shane Watson and Chris Jordan – bowl their last three overs. 

Apart from the fact that both bowlers had a pretty bad day and leaked 52 runs in the last three overs, their overall figures which combined to read 8-0-106-3 were not just significant in the context of the match but, in the final analysis, decisive.

Looked at holistically this way then, the statistics tell the story of a team that lost its way despite being in control more than once in the match - that seems a more reasonable conclusion than questioning the credentials of one man!