Wednesday, July 28, 2010

Shooting Yourself in the Foot

A few months ago I wrote a post on a parameter to gauge how efficient a partnership is. Yesterday was a classic case of a team getting it wrong and losing a chance to gain an advantage.

Virender Sehwag set out like a house on fire like he always does, scoring 64 off 63 before close of play on Day 2. India were chasing a mammoth score of 642 on a flat pitch and needing to win this test if they wanted to win the series. The only way it could happen was if Sehwag batted through the day, and even then the odds for an Indian win would have been tall.

Well they certainly didn't help themselves in trying to shorten them. The pair did decently well on Day 2, working at an efficiency of 54 percent. However, for some inexplicable reason Vijay ended up taking a lot more strike on the third morning. He faced 62 of 90 balls from the start of the 21st over to the end of the 35th. His scoring rate (32 off 62) during this period was decent by test standards Sehwag at the other end was scoring much faster (21 off 28). The partnership efficiency for this passage of play was just 31 percent.

What it also did was slow down Sehwag's charge as he was getting hardly any strike. Even before he got out on 99 he had been dropped at backward point when he was on 89 and had offered a half chance to backward short leg. The man likes to see runs being scored and the opposition always looks to get him out by cutting off his boundaries, although he's smart enough to keep his tally going by taking the singles on offer. However, he can't do anything about the score if he doesn't get the strike. It sounds very simplistic and obvious and yet if you see most of his innings you'll see he invariably ends up facing less deliveries than his partner(s).

I'm not privy to what was discussed in team meetings or on the pitch but if I were Gary Kirsten I'd instruct my batsmen to look for singles when they're batting with Sehwag. Put away the odd bad delivery but the first instinct should be find a gap and get Sehwag back on strike. Not only will it increase the scoring rate, it will also get the other batsmen going and they won't be stuck when Sehwag is finally dismissed.


  1. was discussing exactly the same thing with homer... in fact i told him that this is where gg is best coz he knows and understands viru's itch that is why you wont see gg blocking one ends up...

    simplistic but true to T... :)

  2. And we all know sehwag tends to get out when he's not getting the strike. Safe to conclude murali vijay got his first test wicket.

  3. I don't like your definition of partnership efficiency. If one partner hasn't scored a run but has faced at least one ball, then you get an "efficiency" of zero, regardless of how many balls each faced.

    I think a method more in line with your discussion in the linked post would be this:

    Take the strike rate of the faster batsman, and then work out how many runs would have been scored if he'd faced all the bowling and scored at that rate. Then divide the actual number of runs scored by this "ideal" figure.

    In the Sehwag-Vijay partnership, Sehwag was the quicker batsman and had a strike rate of 75. So if he'd faced all 90 balls, he would have scored 67.5 runs. The actual partnership yielded 53, so the efficiency was 53/67.5 = 78.5%.

  4. Welcome to the blog, David.

    You're right, the efficiency definition I use can fail if one of the batsmen hasn't scored. Maybe I should've added that this parameter should only be looked at once there's been a substantial partnership, but then the very definition of substantial partnership can be vague.

    The reason I came up with my definition was I wanted to figure out what should be the ratio of balls faced by each batsman. I quite like how you've defined it. The only thing to be considered in that scenario is it's almost impossible to have 100 percent efficiency since you never get a situation in which one batsman plays all the deliveries.

  5. Mahek, not true. If both batsmen have the same strike rate the efficiency will be 100%. The difficulty is that David's suggest measure seems biased towards batsmen who score at the same rate.

    The way around that is to measure efficiency as the percentage run-rate of the partnership between the slowest batsman and the fastest.
    ( 53/90 - 32/62 ) / ( 21/28 - 32/62 ) =
    ( 0.588 - 0.516 ) / ( 0.75 - 0.516 ) = 30.7%

  6. You're right Russ, it can be 100% even if both batsmen have the same strike rate. Like I said before, I came up with this parameter to figure out the least number of balls the batsman scoring faster should face in a partnership. That is the point of 100% efficiency. It's a bit of a misnomer because this figure can also be greater than 100% and efficiency by definition can't be more than 100%. I'm afraid I'm getting pedantic here so I'll stop.

    BTW, is it a coincidence that the efficiency figures we arrived at are quite close?

  7. Russ, could you talk a bit about what your measure is trying to explain? If the batsmen have equal strike in a partnership, then your measure gives an efficiency of 50% every time.

    I agree that my definition is biased towards batsmen who score at the same rate, but that is kind of the point - if they're scoring at the same rate, then Mahek can't complain about the faster batsman not getting more of the strike.

  8. David, are you sure? If by "equal strike" you mean they faced the same number of balls, I don't agree, by my calculation. If by equal strike you mean they have the same strike rate then my measure is undefined (divide by 0).

    I'll try and diagram it:

    --- 0.516 slowest batsman strike rate = 0% efficiency
    |-- 0.588 partnership strike rate = 31% efficiency
    --- 0.75 fastest batsman strike rate = 100% efficiency

    That is, my efficiency measure calculates how close a partnership gets to giving the fastest batsman all of the strike. Which, in reply to Mahek, is the point where a partnership is scoring at its maximum potential.

  9. Let r1 = runs scored by the faster batsman, b1 = balls faced by faster batsman.

    r2, b2 the same for the slower batsman.

    As I understand it, your formula is:

    [(r1 + r2)/(b1 + b2) - r2/b2] / (r1/b1 - r2/b2)

    Now suppose that b1=b2=b.


    [(r1 + r2)/(2b) - r2/b] / (r1/b - r2/b)

    = [(r1 + r2 - 2*r2)/(2b)] / [(r1 - r2)/b]

    = [(r1 - r2)/(2b)] / [(r1 - r2)/b]

    = 1/2

  10. I understand your diagram now. I think I am still right - equal strike gives an efficiency of 50% each time. You basically calculate the efficiency on a scale that runs from the speed of the slower batsman to that of the fastest, with equal strike giving a run rate in the middle.

  11. David, yes, I get you now, you are completely correct. A partnership should be 50% efficient, on average, anyway so I don't think the quirk you point out is a weakness in the approach.

  12. I think that to describe this sort of phenomenon, you need both definitions, or something like them - one to say how inefficient the distribution of strike was (your method), and one to say how much damage this caused (mine).

    Your formula can be simplified by the way - it's just the percentage of balls faced by the faster batsman.

  13. I think I know where Russ is going with this and it's a pretty good measure. However, I don't think a partnership should be 50% efficient because you hardly get both batsmen scoring at the same rate. This is especially the case with batsmen like Virender Sehwag or Tamim Iqbal who almost always score faster than their partner.

    As a captain I'd like such batsmen to get more strike than their partner, and efficiency can be a pointer to how much more strike they should get.

    Everyone knows the opposition captain is going to try and frustrate an attacking batsman by spreading the field, the best way to counter it is to keep getting him back on strike. That way he gets to play more deliveries and the team scores at a fair clip without taking the risks associated with hitting boundaries.