T20 Franchises With Much to Learn

18th May, 2017.


Over the last year or so, I've been building a cricket database and model with data far in excess of which is in the public domain, and with the T20 database completed mid-way during the IPL, I want to focus on the quite incredible, and bizarre, decisions that IPL franchises make.

Given the millions thrown around at IPL auction, it is truly incredible that franchises can possibly start the tournament with such unbalanced teams, yet this is what Kings XI Punjab, and Gujarat Lions in particular, managed.  I wrote about Gujarat Lions' issues here - and after I tagged coach Brad Hodge in the promotion of the piece, its not impossible to think he was actually trolling me with his team selection subsequently.  

Eventually Nathu Singh did get a chance in two matches but picked up an injury, not before recording superb figures of 15-1 from his four overs, and he is definitely one to keep an eye on in the future.  

Several days on Twitter, I tweeted the screenshot below, illustrating the performance of overseas batsmen after the group stages:-


We can see here that Sunrisers were out on their own as having the best performance from overseas batsmen, with David Warner, Moises Henriques and Kane Williamson major contributors to this.  Kings XI were actually second, with strong displays from Hashim Amla and Glenn Maxwell in particular - bringing into question coach Virender Sehwag's comments here - how can you criticise a guy who averages 31.00 with a strike rate in excess of 170, with good bowling numbers too?  


However, Sunrisers' and Kings' bowling attack was not so impressive, demonstrated above.  Finalists Pune were the only bowling attack to have overseas bowlers outperform the IPL mean, so it is quite reasonable to assume that signing quality overseas bowlers is very important to a T20 franchise.  

Kings XI's overseas bowlers bowled fewer than 35 overs, with Gujarat's just over 60 - ever other franchise had numbers of around 100 or over, and I'd venture to suggest that this team imbalance went a long way to their failure to qualify.


The final screenshot, above, shows the mean deviation combined from batting and bowling for overseas players.  Three of the top four franchises qualified, with only Mumbai qualifying from those with lower data.  This is purely because Mumbai have such good Indian players that they could be a major threat without great overseas players - imagine if their overseas players performed as well...

Player recruitment is just one area where franchises are lacking.  Despite most employing analysts, they still make incredible mistakes regarding their batting orders.  There is no doubt that T20 is one of the most solvable sports from a game theory basis, and it is vital that batting orders are flexible, on that basis.

Step forward, Sunrisers Hyderabad.  In yesterday's eliminator (scorecard here) they were in trouble, batting first at 99-4 after 15.3 overs, after Yuvraj Singh was dismissed.  10 an over from there would have got them to 144, and had to be a realistic goal, but they needed players with the skill-set to be able to achieve this.  

Vijay Shankar was the other batsman at the crease with Yuvraj, and actually the numbers don't mind him - particularly from a batting average perspective - but the new batsman, Naman Ojha's data over the last couple of years show he wasn't the man for the job of getting SRH to a total that at least bordered on competitive.  Before yesterday's match, in this IPL he hit just FOUR boundaries in 49 balls faced.  Considering that he usually bats at a time where scoring fast is required, these numbers are simply not good enough - 16% is my requirement for a player to be considered a threat, and Ojha's 8.16% isn't that.  

What about if we increase sample size?  Well, last year's IPL data for the wicket-keeper was barely any better - he hit 12 boundaries in 138 balls (8.70%) so it clearly isn't a variance issue.  He's just not a quick enough scorer.

When Shankar was out at 118-5, with 11 balls remaining, who did Sunrisers pick to come in?  Chris Jordan.

However, Jordan's boundary hitting isn't a strength either - since the start of the 2015, in T20 internationals and all major franchise leagues combined, he's hit just 31 boundaries from 266 balls (11.65%) - not much better than Ojha, with an even worse strike rate, as I will demonstrate below.

The irony here is that SRH had two players - Bipul Sharma, Rashid Khan - who were much better suited for the job, and arguably three if you include Bhuvneshwar Kumar, but only Bipul batted, and he faced a mere three balls in the final over.

In this superb article, Tim Wigmore demonstrated the value of a bowler being able to improve their batting to average 18 off 10 balls, and both Bipul and Rashid have the potential to play these types of innings.

Below is a small part of my data prepared in advance for the match:-

SRH

Overall Exp Ave

Overall Exp SR

Overall Exp SR

Ave Dev

SR Dev

SR Dev


Overall

Pace

Spin

Overall

Pace

Spin








Warner

40.34

142.32

149.58

1.45

1.01

1.06

Dhawan

29.67

128.96

102.45

1.07

0.92

0.73

Williamson

37.49

137.17

119.05

1.35

0.98

0.85

Yuvraj

21.52

141.42

101.02

0.77

1.01

0.72

Shankar

29.09

126.89

126.89

1.05

0.90

0.90

Jordan

14.58

108.81

108.81

0.52

0.77

0.77

Ojha

16.11

121.37

89.94

0.58

0.86

0.64

Kumar

6.98

149.05

149.05

0.25

1.06

1.06

Rashid Khan

11.03

151.08

151.08

0.40

1.07

1.07

Bipul Sharma

16.23

157.07

157.07

0.58

1.12

1.12

Kaul

3.88

67.70

67.70

0.14

0.48

0.48


Here we can see that Kumar, Rashid and Bipul, according to my model, all have high expected strike rates (versus the 'average' T20 bowler) around the 150 mark, while Jordan's is below 110 and Ojha's at 121 for pace and a mere 90 against spin.

While Jordan and Ojha have more solid averages than Bhuvneshwar and Rashid, batting stability isn't the required job at this stage of an innings, quick scoring is, and a wicket is not much different to a dot ball.  There are limited resources left and maximising them is key.  Quite simply, selecting two batsmen who are far below average boundary hitters with below-average strike rates, with three batsmen with much better strike rates are left sitting on the bench, is not maximising resources.  

As for Bipul Sharma, my numbers suggest that there's a pretty decent T20 player in there, and he's currently being under-used, both from a selection basis and an in-match basis.  If SRH decide not to retain him in the future, a franchise could make many worse signings.

Yesterday's batting order is just one example of poor strategy from T20 franchises - there are many - but with the importance of this match considered as well, it is as bad a mistake as the much-derided decision by Delhi Daredevils batting Angelo Mathews above Chris Morris here, with 87 required from just 41 balls.  Mathews, another player with below-average boundary hitting data, was just not going to score quickly enough, while in Morris, Delhi could not have wished for many better players suited to that particular scenario.

I'll leave you with one thought - imagine if SRH had got 10 an over to get to 144?  How tough would that six-over chase have been for Kolkata Knight Riders?






Comments