Head to head record as first tie breaker?

[Joemanji]

Something suggested by Jimany at the weekend. How would people feel about head to head record being the primary tie breaker in case of players having equal points? Seems a fairer way than bonus points or TD difference that can be slightly arbitary. This would also keep tournaments closer... it would be harder for one person to have won the tournament by game 5.

[Grumbledook]

I like it, takes away the advantage of someone having an easier run up till that point, either by coaches or races.

[DoubleSkulls]

Best bet is strength of schedule - i.e. the total points all your opponent's get. So if you have an easy ride to the upper ranks you place worse than someone who played all the top coaches.

[mubo]

I'd prefer it. Not sure how many results it would affect though.
Could get a bit messy if you use it throughout the tourney. You mean only at the end?

Best bet is strength of schedule - i.e. the total points all your opponent's get. So if you have an easy ride to the upper ranks you place worse than someone who played all the top coaches.

True, but you can only beat who you're drawn against. Also harder to calculate. Especially if you use it to determine order for Swiss in earlier rounds.

[Purplegoo]

I always prefer the situation where on pitch results count more than TDs scored or CAS inflicted; be it making sure that the best record wins, or as tie-breaker, so I'm a fan. Could be a bit sticky mind, I agree with Nick that this perhaps could only come in late on / at the end when the field has spread out. Perhaps it might be ok for smaller tourneys throughout?

As for strength of schedule; the ideal thing in principle. In practice over only 5 or 6 games, the draw plays too heavy a role. I suggested this at the GT last year, but the thing is too big for it to be fair over 6 games.

[mubo]

If you have three coaches on 4-1-1, two of whom have played one another... what happens then?
Mind you, I guess 3 coaches having the same record is unlikely at the top end of a tourney. You could just use it to determine roster places at the end.

I think Ian's idea could only really work in big tourneys, as in smaller ones Swiss should take care of that. However in a big tourney, the records of your opponents would depend on their draw as well.

[besters]

If you have three coaches on 4-1-1, two of whom have played one another... what happens then?
Mind you, I guess 3 coaches having the same record is unlikely at the top end of a tourney. You could just use it to determine roster places at the end.

If I had won my last game at Poo Bowl at the weekend 3 coaches would have been on 5 - 0 - 1, so perhaps not so unlikely? Each coach would have beaten one of the other two and lost to the other one, I don't see how an individual match up can work in this case?

[Purplegoo]

I'm unsure on strength of schedule, and how fair it is. I mean, if you (Nick) whoop Rycros and Geoff first two games and they each go on to finish typically well, and I get two new guys with Goblins who finish a lot worse (and then our games played are much of a muchness after that), sure, if we both finish on the same record, you've had a harder weekend than me and your record is more impressive than mine. But it's not like I can choose who I play, the names have just been chucked out of score at random. When you make the final and I don’t on tie breakers (in this fictional GT-esq tournament we seem to be doing rather well in! ), I’m sure that on the day you deserve it more on games played, but I never had a crack to see if I did too!

I was on the side of strength of schedule tie breakers at the GT last year when I had the idea, I'm not sure I am now.

Perhaps it's as good a tie breaker as any. Afterall, should one of us go 10-0 against Goblins in game one and the other have a normal game, that's just as much a wrinkle of the draw as sos.

What we really need is a week long tourney where everyone plays everyone. Let’s do it properly!

Spot on on the 4-1-1 front. Perhaps you'd have to revert to traditional methods if more than 2 guys need splitting.

[mubo]

I'm unsure on strength of schedule, and how fair it is. I mean, if you (Nick) whoop Rycros and Geoff first two games and they each go on to finish typically well, and I get two new guys with Goblins who finish a lot worse (and then our games played are much of a muchness after that), sure, if we both finish on the same record, you've had a harder weekend than me and your record is more impressive than mine. But it's not like I can choose who I play, the names have just been chucked out of score at random. When you make the final and I don’t on tie breakers (in this fictional GT-esq tournament we seem to be doing rather well in! ), I’m sure that on the day you deserve it more on games played, but I never had a crack to see if I did too!

Agree completely. I'm definately not advocating using strength of schedule. You can only beat who you get drawn against.
Point I was trying to make above is that it isn't something that you can measure well/objectively. In a tourney big enough to need it, the records of the guys you played mean less.

Solution may be a tier seeding system as discussed on Leip's/Geoff's thread.

I was on the side of strength of schedule tie breakers at the GT last year when I had the idea, I'm not sure I am now.

Biggest problem with the GT is that it's effectively a 5 game swiss, which is daft for 160 coaches. 6 games isn't enough, I reckon it needs 7/8.

[Joemanji]

If you have three coaches on 4-1-1, two of whom have played one another... what happens then?

Just use the next tie breaker, for example TD diff.

[Jimjimjimany]

The off hand comment I made was that if two coaches were to finish on the same points, one having beat the other already, it seems right that they finish higher. Joe is right that it would just be used as the first tiebreaker, so if the two had drawn move on to the next criterion. In the hypothetical (damn you rycros!) situation Besters described it may also not work, however even though those 3 coaches would have had the same records, they may not have had the same points as there were bonus points on offer. So if two had the same points then the head to head between them would have been used as the tiebreaker.

[DoubleSkulls]

I'm unsure on strength of schedule, and how fair it is. I mean, if you (Nick) whoop Rycros and Geoff first two games and they each go on to finish typically well, and I get two new guys with Goblins who finish a lot worse (and then our games played are much of a muchness after that), sure, if we both finish on the same record, you've had a harder weekend than me and your record is more impressive than mine. But it's not like I can choose who I play, the names have just been chucked out of score at random. When you make the final and I don’t on tie breakers (in this fictional GT-esq tournament we seem to be doing rather well in! ), I’m sure that on the day you deserve it more on games played, but I never had a crack to see if I did too!

Now I really do not understand this. If anything it sounds counter-intuitive to me.

Lets say we have identical records and play exactly the same opponents apart from one. You get to play the guy who finished last with a joke team who was at his first tourney and looking for a laugh. You won 6-0. I played Geoff's 'zons and squeeze out a 1-0 win, and Geoff finishes 3rd. So using TD as the tiebreak you'd wallop me - but quite clearly I've had a much harder opponent to get the same overall record - and you are finishing ahead of me. That just does not feel fair - our ranking does not really reflect performance over the weekend.

Now your argument is that you didn't get a chance to play better opponents - that is fair enough - but what ranking is trying to do is say who played best over the weekend. So did the coach playing the easier set of opponents - and ending up with an otherwise identical record - play better than the coach who faces a harder set of opponents? You faced an easier set of opponents so your identical record was easier to achieve - so no. You should have got a better record!

The only reason I don't normally use strength of schedule is that I like including TD/Cas in the primary scoring for the tournament (e.g. 50 pts for a win, +1 per TD/Cas) as that seems to encourage people to keep playing even when they have lost. For Euc Bowl this year I'm doing something different (to encourage the tier 2/3 teams to play each other) which means I can't do that anyway so strength of schedule will be the first tie-break.

As for "how difficult is it" well my own tournament software supports it. So it doesn't actually take me any time to work out myself, I just need to configure the application. That's freely available for download and I'm happy to configure tournaments for people.

[Purplegoo]

The simplest way to say it (Ian) is that any of the traditional tie breakers aren't really that satisfactory (TDs / Diff / CAS / Diff), largely because they rely too heavily on wrinkles in the draw. Whilst I understand what you are saying there, it's another example of the same issue. There's nothing to say Geoff hasn't had an off game, and I haven't coached like a demon for my six TDs. Or that I wouldn't have done Geoff by two or more if the roles reversed. We don't know, it was just a feature of the draw, so whilst on paper it looks an OK breaker, it’s a bit unfair on the poor guy who could only beat what‘s in front of him. Head-to-head between the two of us would have been the fairest way (of course - not always possible).

To be honest, swiss has all these minor issues (especially as tournaments get bigger as Nick suggests), but what are we going to do about it? Insist people take a week off of work and play leagues? It's the best system for what we have (and works perfectly fine enough of the time), so in that regard, whatever tie breaker you choose is fine, since there needs to be one! There is a possibility here we're just taking it too seriously, afterall! I don't see sos as any better or worse than TDs, but I do prefer head to head, if it's just two guys on the same score, or 3+ and there is a clear leader.

[Grumbledook]

maybe rolling a die is fairer then ;]

[Pipey]

Looking at tiebreakers in general, and the reasons for them…

The win should be rewarded above all else, that’s why more and more I think 3/1/0 or 2/1/0 W/D/L ratios are preferable. It’s undesirable for someone with a 5/0/1 record to win a tourney ahead of someone with a 5/1/0 record because of bonus points accrued by racking up loads of TDs/CAS.

If you make it 3/1/0 or 2/1/o like in soccer, then you have the problem of what to make the tiebreak. If you make it net TDs then it favours AG teams; if you make it net CAS then it favours bash teams. If you make it net TD + net CAS then it’s fairer but still seems a bit of a lottery.

So I think it’s better to keep the 3/1/0 or 2/1/0 ratio, keep some kind of bonus, but weight the W/D/L points more heavily than the bonus points. This is like what they do at Water Bowl i.e. 6/3/0 (same ratio as 2/1/0) but with a maximum +1 bonus per round. This rewards the win, but makes it unlikely that lots of players will end up on the same score. So I nicked it and did something similar for Monkey Bowl V!

I don’t ever remember someone winning a tourney when being tied on points with an opponent they had previously lost to. So I don’t see a problem being fixed here. It might make sense though as a first tiebreaker, provided Score can work it out easily and assuming a sensible second tiebreaker is in place.