15 Apr 2007 10:31:42
Peter Foster
Net Run Rate alternative

A few weeks ago there was a thread (subject: NRR needs more
tuning...) that shows some problems with Net Run Rate.
Basically, if you need to boost your NRR it can make a huge
difference whether you are batting first or second, as you will
be batting for differing number of overs.

A better way is to use Duckworth/Lewis to extrapolate scores to
the corresponding 50 over score. Since all scores will then have
a common baseline we can use a straight run difference, which is
a lot easier to relate to.

For example, West Indies v New Zealand
Win made 177, NZ made 179/3 with 11 overs to spare. Using D/L, NZ
50 over score is 264, so NZ win by 87 runs.

A more complicated example is Ireland v South Africa
This was shortened to 35 overs. Ireland made 152/8, which Saf
reached with 3.3 overs to spare. The corresponding 50 over scores
for both teams are Ire 191 Saf 219 so Saf win by 28 runs
(incidentally, my D/L calculations are probably a bit off, but I
hope the idea is clear.)

Doing this for all super 8 matches up to NZ v Saf we get the
following table (use fixed pitch font):

points NRR For Against Difference
Aus 10 2.298 1522 1049 473
NZL 10 1.068 1573 1146 427
SrL 8 1.35 1324 973 351
SAf 6 -0.21 1465 1524 -59
Eng 4 0.079 1134 1136 -2
WIn 2 -1.212 1147 1428 -281
Ban 2 -1.431 878 1300 -422
Ire 0 -1.968 817 1304 -487

(I've included the existing NRR for comparison. )

Knowing the number of runs you are ahead/behind is much more
understandable that the fairly nebulous "Net Run Rate" (IMHO).

Interestingly, despite the huge NRR difference, Aus is only 46
runs ahead of NZ. And Sri Lanka is ahead of NZ on NRR but is 76
behind on run difference.

Comments/discussion welcome (but I will be away fo a week so I
won't be able to reply until then).

Cheers,
Peter
--
Peter Foster
Canberra, Australia


16 Apr 2007 08:51:12
Calvin
Re: Net Run Rate alternative

On Sun, 15 Apr 2007 20:31:42 +1000, Peter Foster
<pfosterNOS@PAMpcug.org.au > wrote:

> A better way is to use Duckworth/Lewis to extrapolate scores to
> the corresponding 50 over score. Since all scores will then have
> a common baseline we can use a straight run difference, which is
> a lot easier to relate to.

I like this a lot, the major drawback is that the DL method itself is so
flawed. A team can win a match despite scoring fewer runs, this method
then compounds the nonsenese of such a result.

But I think you are on the right track. Fix DL and it could be a real
winner.

--
cheers,
calvin


15 Apr 2007 17:03:17
Phil.
Re: Net Run Rate alternative

On Apr 15, 6:51 pm, Calvin <cal...@phlegm.com > wrote:
> On Sun, 15 Apr 2007 20:31:42 +1000, Peter Foster
>
> <pfoster...@PAMpcug.org.au> wrote:
> > A better way is to use Duckworth/Lewis to extrapolate scores to
> > the corresponding 50 over score. Since all scores will then have
> > a common baseline we can use a straight run difference, which is
> > a lot easier to relate to.
>
> I like this a lot, the major drawback is that the DL method itself is so
> flawed. A team can win a match despite scoring fewer runs, this method
> then compounds the nonsenese of such a result.
>
> But I think you are on the right track. Fix DL and it could be a real
> winner.
>

But you can't 'fix it' in the way you suggest, by definition a 'rain
rule' must allow the possibility of winning despite scoring fewer
runs.

Phil.




16 Apr 2007 10:16:41
Calvin
Re: Net Run Rate alternative

On Mon, 16 Apr 2007 10:03:17 +1000, Phil. <felton@princeton.edu > wrote:

> On Apr 15, 6:51 pm, Calvin <cal...@phlegm.com> wrote:

>
> But you can't 'fix it' in the way you suggest, by definition a 'rain
> rule' must allow the possibility of winning despite scoring fewer
> runs.

That's not its only failing.

--
cheers,
calvin


15 Apr 2007 18:14:53
R. Bharat Rao
Re: Net Run Rate alternative

On Apr 15, 6:31 am, Peter Foster <pfoster...@PAMpcug.org.au > wrote:
> A few weeks ago there was a thread (subject: NRR needs more
> tuning...) that shows some problems with Net Run Rate.
> Basically, if you need to boost your NRR it can make a huge
> difference whether you are batting first or second, as you will
> be batting for differing number of overs.
>
> A better way is to use Duckworth/Lewis to extrapolate scores to
> the corresponding 50 over score. Since all scores will then have
> a common baseline we can use a straight run difference, which is
> a lot easier to relate to.

Nahhh -- a far better scenario is to use a lot of different tie
breakers
before coming to NRR. Off the top of my head, I'd use

Points
head-to-head records
# of wins
# of wins against teams with same or better records
# of wins against teams qualified for the playoffs (or tied on points)

and maybe a couple more, and *then* only go to NRR.
and first have NRR in head-to-head (applies if 3 or more teams tied
obv)

And if you have n >2 teams, go through the list until you eliminate one
or
more teams, and then restart with the remaining teams from the top
criteria, until only one team left.

Bharat



16 Apr 2007 21:55:46
dp
Re: Net Run Rate alternative

On Apr 16, 6:14 am, "R. Bharat Rao" <rao2_nor...@yahoo.com > wrote:
> On Apr 15, 6:31 am, Peter Foster <pfoster...@PAMpcug.org.au> wrote:
>
> > A few weeks ago there was a thread (subject: NRR needs more
> > tuning...) that shows some problems with Net Run Rate.
> > Basically, if you need to boost your NRR it can make a huge
> > difference whether you are batting first or second, as you will
> > be batting for differing number of overs.
>
> > A better way is to use Duckworth/Lewis to extrapolate scores to
> > the corresponding 50 over score. Since all scores will then have
> > a common baseline we can use a straight run difference, which is
> > a lot easier to relate to.

I had made the same suggestion over here:
http://groups.google.com/group/rec.sport.cricket/browse_frm/thread/6ba57c446d683c1e/f108390519c54fff?lnk=st&q=&rnum=1#f108390519c54fff

It makes sense. So I don't expect ICC to ever implement it in their
lifetime (of which thankfully there isn't much left).

>
> Nahhh -- a far better scenario is to use a lot of different tie
> breakers
> before coming to NRR. Off the top of my head, I'd use
>
> Points
> head-to-head records
> # of wins
> # of wins against teams with same or better records
> # of wins against teams qualified for the playoffs (or tied on points)

None of these are useful as tie-breakers in the most common scenario:
4 teams in a group, 3 of which end up with 2 wins each (they all beat
up on the minnow and then win one of the other two matches).

Net run rate is a reasonable method of breaking ties as it indicates
the margin of victories. However, the way it is calculated at present
it is heavily loaded against teams who chase and win (as explained
with examples in the thread linked above). All that is needed is to
use D/L table to convert all wins by wickets into wins by runs.

dp



17 Apr 2007 08:23:18
R. Bharat Rao
Re: Net Run Rate alternative

On Apr 17, 12:55 am, dp <dpuse...@yahoo.com > wrote:
> On Apr 16, 6:14 am, "R.BharatRao" <rao2_nor...@yahoo.com> wrote:

> > > A better way is to use Duckworth/Lewis to extrapolate scores to
> > > the corresponding 50 over score. Since all scores will then have
> > > a common baseline we can use a straight run difference, which is
> > > a lot easier to relate to.
>
> I had made the same suggestion over here:http://groups.google.com/group/rec.sport.cricket/browse_frm/thread/6b...
>
> It makes sense. So I don't expect ICC to ever implement it in their
> lifetime (of which thankfully there isn't much left).

I'm sorry -- it makes no sense to me. D/L has enough flaws as it is,
changing
wildly when a wicket falls. Game shortened to 35 overs, and instead
of teams
slogging wildly in overs 34-35, they are playing to preserve wickets
to
artificially elevate their D/L score at the end of 50 overs -- no
thank you.

> > Nahhh -- a far better scenario is to use a lot of different tie
> > breakers
> > before coming to NRR. Off the top of my head, I'd use
>
> > Points
> > head-to-head records
> > # of wins
> > # of wins against teams with same or better records
> > # of wins against teams qualified for the playoffs (or tied on points)
>
> None of these are useful as tie-breakers in the most common scenario:
> 4 teams in a group, 3 of which end up with 2 wins each (they all beat
> up on the minnow and then win one of the other two matches).

Really -- tell that to South Africa -- which will surely figure in
some of
the tie-breaking scenarios. I'm amazed at the "not useful" statement,
when in the last 2 world cups, so-called minnows have beaten many of
the top teams. Kenya last time, and Bangladesh this year. So, I'm
guessing the "most common" scenario is not that common, is it?


> Net run rate is a reasonable method of breaking ties as it indicates
> the margin of victories. However, the way it is calculated at present
> it is heavily loaded against teams who chase and win (as explained
> with examples in the thread linked above). All that is needed is to
> use D/L table to convert all wins by wickets into wins by runs.

I disagree -- my problem with NRR is that it can be easily manipulated
(see the "go slow" by Waugh's Aussies in the 99 WCup) and is also
amenable to being "fixed". That manipulation applies triple when you
look at D/L extrapolations (wickets lost, etc.), which I hope will
*NEVER*
be taken into account by the authorities.

I'm not saying that D/L should not be used as a tie-breaker. Just
that
as many tie-breakers as possible should be used before it comes into
play -- with those tie-breakers being based upon wins and records.

Bharat



17 Apr 2007 08:34:51
RishiX
Re: Net Run Rate alternative

> I'm sorry -- it makes no sense to me. D/L has enough flaws as it is,
> changing
> wildly when a wicket falls. Game shortened to 35 overs, and instead
> of teams
> slogging wildly in overs 34-35, they are playing to preserve wickets
> to
> artificially elevate their D/L score at the end of 50 overs -- no
> thank you.

If a team *has to slog wildly* to win the game, there is no way they
would be ahead on D/L just by preservin wickets. Just take any random
situation and check D/L.

> I disagree -- my problem with NRR is that it can be easily manipulated
> (see the "go slow" by Waugh's Aussies in the 99 WCup) and is also
> amenable to being "fixed". That manipulation applies triple when you
> look at D/L extrapolations (wickets lost, etc.), which I hope will
> *NEVER*
> be taken into account by the authorities.

In terms on deciding that particular match, I think D/L is the best we
have had so far. It does not take into account things like a pinch-
hitter sent ahead to boost the run-rate etc because it cannot. But
other than that it is pretty reasonable.

> I'm not saying that D/L should not be used as a tie-breaker. Just
> that
> as many tie-breakers as possible should be used before it comes into
> play -- with those tie-breakers being based upon wins and records.
>
> Bharat

That I agree with. NRR is not the best way to break a tie.

-- Rishi.




17 Apr 2007 08:53:28
dp
Re: Net Run Rate alternative

On Apr 17, 8:23 pm, "R. Bharat Rao" <rao2_nor...@yahoo.com > wrote:
> > I had made the same suggestion over here:http://groups.google.com/group/rec.sport.cricket/browse_frm/thread/6b...
>
> > It makes sense. So I don't expect ICC to ever implement it in their
> > lifetime (of which thankfully there isn't much left).
>
> I'm sorry -- it makes no sense to me. D/L has enough flaws as it is,
> changing
> wildly when a wicket falls. Game shortened to 35 overs, and instead
> of teams
> slogging wildly in overs 34-35, they are playing to preserve wickets
> to
> artificially elevate their D/L score at the end of 50 overs -- no
> thank you.

I suggest that you go back to that thread and read the proposal
carefully. In my proposed change D/L comes into picture only when the
chasing team wins, so I don't see how a team that needs to be slogging
wildly in 35th over to win can artificially elevate their D/L score.
Most of this NRR is significant only when a team wins by huge margin.
In close matches, it doesn't make much of a difference how you
calculate it.

>
> > None of these are useful as tie-breakers in the most common scenario:
> > 4 teams in a group, 3 of which end up with 2 wins each (they all beat
> > up on the minnow and then win one of the other two matches).
>
> Really -- tell that to South Africa -- which will surely figure in
> some of
> the tie-breaking scenarios. I'm amazed at the "not useful" statement,
> when in the last 2 world cups, so-called minnows have beaten many of
> the top teams. Kenya last time, and Bangladesh this year. So, I'm
> guessing the "most common" scenario is not that common, is it?

ok, make it expected to be common scenario. Cricket being a game of
"glorious uncertainties", anything can happen ofcourse. (aside, good
to see BD revert back to form against Ireland).

>
> > Net run rate is a reasonable method of breaking ties as it indicates
> > the margin of victories. However, the way it is calculated at present
> > it is heavily loaded against teams who chase and win (as explained
> > with examples in the thread linked above). All that is needed is to
> > use D/L table to convert all wins by wickets into wins by runs.
>
> I disagree -- my problem with NRR is that it can be easily manipulated
> (see the "go slow" by Waugh's Aussies in the 99 WCup) and is also
> amenable to being "fixed". That manipulation applies triple when you
> look at D/L extrapolations (wickets lost, etc.), which I hope will
> *NEVER*
> be taken into account by the authorities.

Incorporating D/L doesn't provide any more room for manipulation than
what is there already.

> I'm not saying that D/L should not be used as a tie-breaker. Just
> that
> as many tie-breakers as possible should be used before it comes into
> play -- with those tie-breakers being based upon wins and records.

That I agree. But there will still be cases when you don't have any
other option. It doesn't hurt to fix the NRR calculation to make it as
fair as possible when it needs to be used in those cases.

dp
>
> Bharat