[Hanoi5]

What are the rules to solve a tie situation? Let's say two teams in a swiss tournament are tied and:

a. They had a match against each other
b. They didn't have a match against each other

I would think that in case (a) the winner of the match should be the winner of the tie. If they're tied there or if (b) happens then the team with the most imps overall should win. What is the official order?

[mjj29]

Hanoi5, on 2011-August-24, 08:25, said:

What are the rules to solve a tie situation? Let's say two teams in a swiss tournament are tied and:

a. They had a match against each other
b. They didn't have a match against each other

I would think that in case (a) the winner of the match should be the winner of the tie. If they're tied there or if (b) happens then the team with the most imps overall should win. What is the official order?

Depends on your jurisdiction. The EBU's White Book defines the split-tie procedure for their events and when to apply it and in which order the various tie-splitters occur. In general you only split if you have to (money and green points can just be divided between the teams, only a trophy or qualification for another event must be split) and if there is a direct match then that will be one of the higher-precedence tie splitters, but it depends on the exact form of competition.

If you want to know the full gory details I suggest you read up in the White Book (available from the EBU website), but if it's not an EBU tournament then you'll have to check the Conditions of Contest.

[RMB1]

mjj29, on 2011-August-24, 08:30, said:

..., but if it's not an EBU tournament then you'll have to check the Conditions of Contest.

But expect to be disappointed

[bluejak]

There are a number of jurisdictions which rely on the EBU White book for matters for which they have no regulation themselves. This seems a sensible approach.

[jeffford76]

Hanoi5, on 2011-August-24, 08:25, said:

What are the rules to solve a tie situation? Let's say two teams in a swiss tournament are tied and:

a. They had a match against each other
b. They didn't have a match against each other

I would think that in case (a) the winner of the match should be the winner of the tie. If they're tied there or if (b) happens then the team with the most imps overall should win. What is the official order?

This varies by jurisdiction. Last time I was involved in a Swiss tie in the ACBL (top 16 advancing to a KO for GNTs) the tiebreak was total VP's of the the teams you played in the second half - presumably intended to discourage the "Swiss gambit", but I never understood why this was more fair than total VP's of all the teams you played.

Total VP's of all the teams you played was the tiebreak in the current Juniors world championship.

In general I think something involving opponent strength is more fair than just using the head-to-head result.

[Hanoi5]

Ok, I see now. I thought it was a new thing. I still don't see the logic of breaking a tie between two teams which played a match against each other by looking at something different than that match. But the rules are the rules are the rules.

[Echognome]

Hanoi5, on 2011-August-24, 12:59, said:

Ok, I see now. I thought it was a new thing. I still don't see the logic of breaking a tie between two teams which played a match against each other by looking at something different than that match. But the rules are the rules are the rules.

I can see head-to-head result being one tiebreaker, but of course you could have a 3-way tie between A, B, and C where A beat B, B beat C, and C beat A.

[barmar]

Hanoi5, on 2011-August-24, 12:59, said:

Ok, I see now. I thought it was a new thing. I still don't see the logic of breaking a tie between two teams which played a match against each other by looking at something different than that match. But the rules are the rules are the rules.

I think the rationale may be that more data is better. Breaking the tie based on the results of just one match is not as good as using the results of many matches.

[aguahombre]

I always thought that the norm was to start with the heads-up result; then if still tied (or they didn't play each other) go to performance of the teams each team played. Obviously, the rules are whatever the conditions state. But somehow I thought that is what they usually state.

This could be unfair if a team we met very early never won another match afterward and it was an accidental result of first or second round swiss matching. Only counting teams met in the second half would reduce that effect.

[mrdct]

It is just so important to have conditions of contest that cover tie splitting properly.

A couple of year ago my team tied for first place in the Northern Victoria GNOT qualifier with the winner getting an airfare subsidy and entry fees for the GNOT Finals at Tweed Heads. Under the Victorian Bridge Association Tournament Regulations, my team would've won the tie split but under the Australian Bridge Federation Tournament Regulations the other team would've won. The GNOT Regulations said that it was up to individual convenors of regional qualifiers to establish their own conditions of contest, but the convenor in this case (who was also a member of other tied team) had not done so. In the handful of regions where conditions of contest were put in place, some referred to the State Body's regulations and some referred to the National Body's regulations. Ultimately it was referred to the ABF for a decison which took almost 2 months and was in favour of the other team so I missed out on the trip.

[Ant590]

I understand that the head-to-head is the standard way to split ties, but I have always felt that this is actually unfair.

If team A beat team B, and they finished on equal VPs, then team B will have won more points against other pairs. It seems that the match has almost counted twice, once to give team A enough VPs to tie, and then again to win the split-tie.

[mgoetze]

Ant590, on 2011-August-26, 03:41, said:

I understand that the head-to-head is the standard way to split ties, but I have always felt that this is actually unfair.

If team A beat team B, and they finished on equal VPs, then team B will have won more points against other pairs. It seems that the match has almost counted twice, once to give team A enough VPs to tie, and then again to win the split-tie.

Indeed in Go, the standard tie breaker is sum of opponents' scores (SOS), and head-to-head is only considered in formats such as round robin. The idea being that two players with the same score may not have had to work equally hard for that score.

[paulg]

Ant590, on 2011-August-26, 03:41, said:

I understand that the head-to-head is the standard way to split ties, but I have always felt that this is actually unfair.

If team A beat team B, and they finished on equal VPs, then team B will have won more points against other pairs. It seems that the match has almost counted twice, once to give team A enough VPs to tie, and then again to win the split-tie.

On the other hand, following the win team A will have played theoretically stronger opponents than team B.

Personally I prefer the winners of the match between two teams to be used to split a tie. At least that is a contest between the two teams. Failing that, their scores against common opponents.

What I dislike is sum of opponent's scores (SOS). This can lead to the splitting of a tie being based on a match between two teams not involved in the tie, perhaps even at the bottom of the field.

But it probably doesn't make a whole of difference as there will be a large element of luck in any method.

[Jeremy69A]

Although a direct match is a common way to split, in tournaants such as the Brighton Teams where you may need to split a tie amongst more than two teams for a place in either of the two finals they use Swiss Points i.e. the results of your opponents. Ever since I won a 4 way split tie I've considered this method to have some merit.
I've also played in a Swiss in La Baule where after round 2, in addition to your vp score, you automatically had this amount added and shown on the ranking list round by round which seems to me a good idea.

[campboy]

paulg, on 2011-August-26, 05:40, said:

On the other hand, following the win team A will have played theoretically stronger opponents than team B.

Does not follow. Team A will be on a higher score than B when assignments for the round immediately after their match are made, but in subsequent rounds B could easily be on a higher score, and facing supposedly stronger opponents, than A. Of course, SOS is a better guide to the strength of each team's opponents than the head-to-head result.

Quote

What I dislike is sum of opponent's scores (SOS). This can lead to the splitting of a tie being based on a match between two teams not involved in the tie, perhaps even at the bottom of the field.

The result of a head-to-head tiebreak can also be changed by the result of a match between two other teams if that match introduces other teams into the tie. (A beats B in a tiebreak between A and B does not imply that A beats B in a tiebreak between A,B and C.)

[paulg]

campboy, on 2011-August-26, 06:21, said:

paulg, on 2011-August-26, 05:40, said:

What I dislike is sum of opponent's scores (SOS). This can lead to the splitting of a tie being based on a match between two teams not involved in the tie, perhaps even at the bottom of the field.

The result of a head-to-head tiebreak can also be changed by the result of a match between two other teams if that match introduces other teams into the tie. (A beats B in a tiebreak between A and B does not imply that A beats B in a tiebreak between A,B and C.)

Using a tie break that involves just the matches played by the teams to be split, for example A, B and C, is preferable to one that involves a match between teams not in the tie break, for example, D and E.

In my opinion.

Which is what I said originally.

[campboy]

I don't think your proposed method achieves this. In order to know whether A, B or C wins the tie-break you need to know that D is not involved in the tie -- which depends on the result of the match between D and E.

A tie-break such as net IMPs would have no dependence on results of matches between other teams.

[aguahombre]

Not so profound expression of the obvious:

In the 3-way tie, if they each played each other, shouldn't it just be considered a round-robin and broken by quotient, like we do every day in the earlier stages of a K.O.? This might have been addressed within some other post, but if so I missed it.

[bluejak]

I have no idea what is the fairest method.

But I have played in Swiss Teams tournaments where the results were shown in both VPs and total imps. This leads to no ties [with incredibly rare exceptions] and a simple method everyone understands and can see. It seems best to me.

[Mbodell]

In some chess tournaments that are also played swiss we used to use sum of a players score. That is your score after round 1 + score after rounds 2 + ... + score after round n. This way a team that took "the swiss strategy" is penalized for the expected weaker field, and it doesn't matter as much how your early round opponents end up doing (who, when considering tie breakers of the top teams, will tend to vary between the worst teams and the mediocre teams, and the best teams will be expected to beat them). But the chess tournaments have Win, Lose, Draw scoring so don't have 20, 25, or 30 VP per match to award.

There was a complicated method that involved estimating the results against all the teams as if it were a full round robin (using IMP in other matches and SoS to simulate the full round robin). That seems like it might be fairest, especially for events where you will have played a reasonable percentage of the field.