Head to head as tiebreaker

Head to head as tiebreaker

Post by Peter Fost » Fri, 08 Feb 2002 19:36:35


There has been a lot of recent discussion about the use of
head to head results as a tiebreaker when two teams are
equal on points.

What is wrong with this method? It certainly seems a
reasonable way to break a tie, but it can have a number of
unfortunate side effects. The most bizarre is that a match
can affect the relative placings of two teams who were not
actually playing.

For example, suppose there are 4 teams, A, B, C and D.
They are in the order ABCD and there is one match still to
play, between C and D. C wins this match and the final order
is BACD. I think that most people would agree that this is a
ridiculous outcome, but it is what can happen when using
head-to-head.

What to use as a tiebreak? Here are some thoughts:

1. Use bonus points in a modified form, as follows:
i) points (not bonus points)
ii) fewest bonus points conceded
iii) most bonus points gained
iv) NRR

2. S***bonus points entirely and just use the NRR.

3. NRR has a problem in that it can't be used for a rain
shortened match. Instead of NRR, simply use the total
winning margin in runs. Duckworth/Lewis can be used to
determine the winning margin when the side batting second
wins. For example, in the first final SAf won with the score
at  2/191after 45.1 overs. The D/L par is 156, so SAf won by
35 runs. The advantage of D/L is that it can also be used in
a rain shortened match (the disadvantage is that it is
totally inscrutable to most normal people). This also may
eliminate any need for bonus points, as big wins/losses will
be accurately reflected (more accurately than just using
NRR).

And no, I'm not whingeing about Australia missing the
finals. This was a particularly unusual tournament, as the
three teams finished with the same number of wins and
identical  win/loss ratios, so any tiebreak method will
produce an "unfair" result

Cheers,
Peter.
------------------
Peter Foster
Canberra, Australia

 
 
 

Head to head as tiebreaker

Post by djb » Fri, 08 Feb 2002 23:36:31

Why can't net run rate be applied in a rain shortened game ? It still
probably fairer than DL

The Duckworth/Lewis method is a method of adjusting run rate targets in lieu
of a smaller target due to inclement weather to make it easier for a side
batting second - I don't think that makes it a fairer  method than net run
rate. Especially as the DL method is a empiral method based on the 1992
world cup results.


Quote:
> There has been a lot of recent discussion about the use of
> head to head results as a tiebreaker when two teams are
> equal on points.

> What is wrong with this method? It certainly seems a
> reasonable way to break a tie, but it can have a number of
> unfortunate side effects. The most bizarre is that a match
> can affect the relative placings of two teams who were not
> actually playing.

> For example, suppose there are 4 teams, A, B, C and D.
> They are in the order ABCD and there is one match still to
> play, between C and D. C wins this match and the final order
> is BACD. I think that most people would agree that this is a
> ridiculous outcome, but it is what can happen when using
> head-to-head.

> What to use as a tiebreak? Here are some thoughts:

> 1. Use bonus points in a modified form, as follows:
> i) points (not bonus points)
> ii) fewest bonus points conceded
> iii) most bonus points gained
> iv) NRR

> 2. S***bonus points entirely and just use the NRR.

> 3. NRR has a problem in that it can't be used for a rain
> shortened match. Instead of NRR, simply use the total
> winning margin in runs. Duckworth/Lewis can be used to
> determine the winning margin when the side batting second
> wins. For example, in the first final SAf won with the score
> at  2/191after 45.1 overs. The D/L par is 156, so SAf won by
> 35 runs. The advantage of D/L is that it can also be used in
> a rain shortened match (the disadvantage is that it is
> totally inscrutable to most normal people). This also may
> eliminate any need for bonus points, as big wins/losses will
> be accurately reflected (more accurately than just using
> NRR).

> And no, I'm not whingeing about Australia missing the
> finals. This was a particularly unusual tournament, as the
> three teams finished with the same number of wins and
> identical  win/loss ratios, so any tiebreak method will
> produce an "unfair" result

> Cheers,
> Peter.
> ------------------
> Peter Foster
> Canberra, Australia


 
 
 

Head to head as tiebreaker

Post by Moby » Sat, 09 Feb 2002 12:35:01

Quote:

> There has been a lot of recent discussion about the use of
> head to head results as a tiebreaker when two teams are
> equal on points.

> What is wrong with this method?

There's nothing wrong.. as such.. it's just that it doesn't work very well
with some systems.

The ACB (and ICC) made a mistake in trying to make two changes at once.
Both, on their own (IMHO) would have improved triangular ODIous series,
but together they create conflict.

For mine, the head to head system is the best way to sort out results
where there are no bonus points.  However, where bonus points come into
it, NRR would be a better option.

The first lessons the likelyhood of a repeat of a WI/Aus, WC99-type match,
and the other decreases the chance of matches like the NZRSA and the
RSA/Aus matches this year.

Moby

 
 
 

Head to head as tiebreaker

Post by Paul Robs » Sun, 10 Feb 2002 01:12:14


Quote:
(Peter Foster) writes:
>And no, I'm not whingeing about Australia missing the
>finals.

Of course not.

How about Alphabetical order ?
Failing that, order them in distance from the South Pole, smallest
first ?

 
 
 

Head to head as tiebreaker

Post by Ro » Sun, 10 Feb 2002 01:27:47



Quote:


>> There has been a lot of recent discussion about the use of
>> head to head results as a tiebreaker when two teams are
>> equal on points.

>> What is wrong with this method?

>There's nothing wrong.. as such.. it's just that it doesn't work very well
>with some systems.

>The ACB (and ICC) made a mistake in trying to make two changes at once.
>Both, on their own (IMHO) would have improved triangular ODIous series,
>but together they create conflict.

>For mine, the head to head system is the best way to sort out results
>where there are no bonus points.  However, where bonus points come into
>it, NRR would be a better option.

Can't really agree, especially in a 3 team competition. If 2 teams are tied for
2nd place, and one has a 3-1 win record against the other, then by default that
side must have a worse record against the top side.
Essentially the side that will get through will be the side that has the worst
record against the top side. Hardly logical.

The same principle applies to tournaments with more teams competing. Should 2
sides be even, then the side that will progress is the one that has the worst
record against the other sides.

Given that we should be trying to see the best performed side progress, it
doesn't really make sense that the head to head record is used when it is
clearly not as good an indicator of performance in the tournament as net run
rate.

 
 
 

Head to head as tiebreaker

Post by Moby » Sun, 10 Feb 2002 06:30:50


<snip>

Regardless of the whinge factor, it's still a good point.  NRR is the best
way to break a deadlock in a triangular series with bonus points.
Head-to-head is a far better way of breaking the deadlock without bonus
points.

The problem would have been avoided by implementing one or the other, no
both.

Moby.

 
 
 

Head to head as tiebreaker

Post by Moby » Sun, 10 Feb 2002 06:47:50


Quote:



> >> There has been a lot of recent discussion about the use of
> >> head to head results as a tiebreaker when two teams are
> >> equal on points.

> >> What is wrong with this method?

> >There's nothing wrong.. as such.. it's just that it doesn't work very well
> >with some systems.

> >The ACB (and ICC) made a mistake in trying to make two changes at once.
> >Both, on their own (IMHO) would have improved triangular ODIous series,
> >but together they create conflict.

> >For mine, the head to head system is the best way to sort out results
> >where there are no bonus points.  However, where bonus points come into
> >it, NRR would be a better option.

> Can't really agree, especially in a 3 team competition. If 2 teams are
> tied for
> 2nd place, and one has a 3-1 win record against the other, then by
> default that
> side must have a worse record against the top side.
> Essentially the side that will get through will be the side that has
> the worst
> record against the top side. Hardly logical.

Well, I agree if you're saying that what you've said is hardly logical.

While they have, by default, a 3-1 record against the other, then they
probably have a pretty poor record against the team they are competing for
a spot against.

SO... if there was a game played to determine who would go through (and
surely this is the best system, because it's the one they use for finals),
then the one who is likely to go through is the one with the better record
-between the two teams-.

See?

Moby.

 
 
 

Head to head as tiebreaker

Post by Peter Fost » Sun, 10 Feb 2002 09:40:42

On 08 Feb 2002 16:12:14 GMT, in rec.sport.cricket,

Quote:


> (Peter Foster) writes:

> >And no, I'm not whingeing about Australia missing the
> >finals.

> Of course not.

Paul, did you actually *read* my post?
Are people posting from .au domains not permitted to comment
on the system?
Do you honestly think that it is a good system?

Head-to-head can lead to problems, because it results in
favouring particular teams:
- In WC99 in resulted in a team finding it advantageous to
deliberately underperform (and I objected to it at that
time).
- In the VB series, it resulted in teams finding it
advantageous to deliberately underperform.

Do you really think that deliberate underperformance is a
good thing? If not, then you must agree that the rules are
flawed.

Cheers,
Peter
------------------
Peter Foster
Canberra, Australia

 
 
 

Head to head as tiebreaker

Post by Peter Fost » Sun, 10 Feb 2002 09:50:52

On Fri, 8 Feb 2002 13:35:01 +1000, in rec.sport.cricket,

Quote:


> > There has been a lot of recent discussion about the use of
> > head to head results as a tiebreaker when two teams are
> > equal on points.

> > What is wrong with this method?
> There's nothing wrong.. as such.. it's just that it doesn't work very well
> with some systems.

Moby, you snipped the rest of my post where I answered my
own question, showing that head to head is actually a poor
tiebreaker. It certainly seems a logical method, but as I
showed, it can lead to ridiculous outcomes. The example I
gave showed that a game can affect the relative rankings of
two teams who are not actually playing. For example: team A
is ahead of team B. If C can beat D then B will move ahead
of A.  I think you will agree that this is an absurd
outcome.

Cheers,
Peter
------------------
Peter Foster
Canberra, Australia

 
 
 

Head to head as tiebreaker

Post by Takla Saa » Sun, 10 Feb 2002 12:02:54

Quote:




> >> There has been a lot of recent discussion about the use of
> >> head to head results as a tiebreaker when two teams are
> >> equal on points.

> >> What is wrong with this method?

> >There's nothing wrong.. as such.. it's just that it doesn't work very well
> >with some systems.

> >The ACB (and ICC) made a mistake in trying to make two changes at once.
> >Both, on their own (IMHO) would have improved triangular ODIous series,
> >but together they create conflict.

> >For mine, the head to head system is the best way to sort out results
> >where there are no bonus points.  However, where bonus points come into
> >it, NRR would be a better option.

> Can't really agree, especially in a 3 team competition. If 2 teams are tied for
> 2nd place, and one has a 3-1 win record against the other, then by default that
> side must have a worse record against the top side.
> Essentially the side that will get through will be the side that has the worst
> record against the top side. Hardly logical.

Neither is it logical that one team should have a 3-1 record against another
and yet watch the latter go through.
The head-to-head tie-breaker works worst in a triangular tournament, and even
there, it prevents a very strong team deciding its final opponent by
tanking against one team and playing full strength against another.
For instance, if Team A could win all 4 matches against B, while B had a
3-1 record against C, A could give C a leg-up by going 2-2 against them,
and by suitably rigging net run rates.
A head-to-head tie-breaker would negate this strategy.
Not that I think it matters much in the case of such a lopsided field.

Quote:

> The same principle applies to tournaments with more teams competing. Should 2
> sides be even, then the side that will progress is the one that has the worst
> record against the other sides.

And what's wrong with that per se ?

Quote:

> Given that we should be trying to see the best performed side progress, it
> doesn't really make sense that the head to head record is used when it is
> clearly not as good an indicator of performance in the tournament as net run
> rate.

If the NRR were used, a team chasing 280 could easily choose to chase 250
instead, and preserve their NRR rather than run the risk of being shot out
for 200 trying to get it all. Besides, it would reward a team
getting 202/7 in 43 overs more than a team getting 202/2 in 47 overs,
though both were chasing 200 in 50. There's no good reason for this, IMO.
On the other hand, if you had the net run rate normalized to 50 overs
per innings using the D/L method, I can see some fairness being restored.
 
 
 

Head to head as tiebreaker

Post by Ro » Sun, 10 Feb 2002 15:03:40



Quote:





>> >> There has been a lot of recent discussion about the use of
>> >> head to head results as a tiebreaker when two teams are
>> >> equal on points.

>> >> What is wrong with this method?

>> >There's nothing wrong.. as such.. it's just that it doesn't work very well
>> >with some systems.

>> >The ACB (and ICC) made a mistake in trying to make two changes at once.
>> >Both, on their own (IMHO) would have improved triangular ODIous series,
>> >but together they create conflict.

>> >For mine, the head to head system is the best way to sort out results
>> >where there are no bonus points.  However, where bonus points come into
>> >it, NRR would be a better option.

>> Can't really agree, especially in a 3 team competition. If 2 teams are
>> tied for
>> 2nd place, and one has a 3-1 win record against the other, then by
>> default that
>> side must have a worse record against the top side.
>> Essentially the side that will get through will be the side that has
>> the worst
>> record against the top side. Hardly logical.

>Well, I agree if you're saying that what you've said is hardly logical.

>While they have, by default, a 3-1 record against the other, then they
>probably have a pretty poor record against the team they are competing for
>a spot against.

True, but that doesn't necessarily mean that they had a better tournament (and
therefore deserve to be the team that progresses through).

Quote:
>SO... if there was a game played to determine who would go through (and
>surely this is the best system, because it's the one they use for finals),
>then the one who is likely to go through is the one with the better record
>-between the two teams-.

No, I would have thought that the team likely to win is the one that has
performed best in the series. And IMO, the best indicator of performance (2nd
to games won of course) is net run rate. Head to head only looks at the result
of games between the two sides, ignoring everything else that team has done
through the series.

But in saying that, I would only expect someone to agree with this if they
believe that it is the best performed side *throughout the tournament* that
should progress.

 
 
 

Head to head as tiebreaker

Post by Colin Lor » Sun, 10 Feb 2002 23:46:28


Quote:
> There has been a lot of recent discussion about the use of
> head to head results as a tiebreaker when two teams are
> equal on points.

> What is wrong with this method? It certainly seems a
> reasonable way to break a tie, but it can have a number of
> unfortunate side effects. The most bizarre is that a match
> can affect the relative placings of two teams who were not
> actually playing.

> For example, suppose there are 4 teams, A, B, C and D.
> They are in the order ABCD and there is one match still to
> play, between C and D. C wins this match and the final order
> is BACD. I think that most people would agree that this is a
> ridiculous outcome, but it is what can happen when using
> head-to-head.

> What to use as a tiebreak? Here are some thoughts:

> 1. Use bonus points in a modified form, as follows:
> i) points (not bonus points)
> ii) fewest bonus points conceded
> iii) most bonus points gained
> iv) NRR
> 2. S***bonus points entirely and just use the NRR.

The bonus point is actually duplicated in NRR. If you win by the such a
large amount that you earn the BP then you must have given your NRR a
significant boost anyway. The NRR is much smoother, and rewards consistent
performance over the odd large victory. Assuming 50 runs is the cut off for
the BP (ie Team scores 250), if team A won 1 game by 51 runs (BP) and 3
games by 1 runs (no BP) it would in fact be higher on the table than team B
that won all 4 games by 49 runs (no BP). Yet, you would have to be mad to
argue that Team A was more *** than team B, despite Team A sitting on
top of the table? NRR is a far better guide, because as pointed out earlier
the BP victory margin is already worked into the equation.

Quote:
> And no, I'm not whingeing about Australia missing the
> finals. This was a particularly unusual tournament, as the
> three teams finished with the same number of wins and
> identical  win/loss ratios, so any tiebreak method will
> produce an "unfair" result.

I don't think Aus was knocked out unfairly as such. The rules were there
before the tournament started, poor rules or not all the teams had to play
by them.
 
 
 

Head to head as tiebreaker

Post by Peter Fost » Wed, 13 Feb 2002 20:14:38

On Sat, 09 Feb 2002 14:46:28 GMT, in rec.sport.cricket,

Quote:

>Peter Foster wrote
> > And no, I'm not whingeing about Australia missing the
> > finals. This was a particularly unusual tournament, as the
> > three teams finished with the same number of wins and
> > identical  win/loss ratios, so any tiebreak method will
> > produce an "unfair" result.

> I don't think Aus was knocked out unfairly as such. The rules were there
> before the tournament started, poor rules or not all the teams had to play
> by them.

Oh, I don't think Australia was unfairly eliminated (that's
why I put the word in quotes). I was just saying that
whichever team was eliminated could point out that they
actually beat one of the other teams that finished above
them.

And yes, the rules were known to all. The question is
whether it is felt that a line has been crossed. The
underarm delivery crossed a line, and the lawmakers reacted
by fixing the rules. If the lawmakers, after what happened
in WC99 and the VB series, decide to retain head to head,
then they have given tacit approval for teams to indulge in
this trickiness.

Cheers,
Peter

------------------
Peter Foster
Canberra, Australia

 
 
 

Head to head as tiebreaker

Post by Colin Lor » Sat, 16 Feb 2002 20:41:35


Quote:
> On Sat, 09 Feb 2002 14:46:28 GMT, in rec.sport.cricket,
> And yes, the rules were known to all. The question is
> whether it is felt that a line has been crossed. The
> underarm delivery crossed a line, and the lawmakers reacted
> by fixing the rules. If the lawmakers, after what happened
> in WC99 and the VB series, decide to retain head to head,
> then they have given tacit approval for teams to indulge in
> this trickiness.

Too true. They need to find a better way before the next WC. Either that or
validate the tactics as you suggest (which I hope does NOT happen).