Yes, that assumption would be correct. Every time you eliminate someone from a tie, via one of the categories, you would technically begin again at step 1, but obviously it would pointless to recheck 1 and 2.

Cool - that's what I thought. H2H will be the only one that has to be rechecked each time since the other categories will always be tied if they move to the next level of tie-breakers together.

I would be very happy if we do this.

Brad will be writing it up in official form, including a level of description for each step to include in the floor rules. I'm sure he will post a final version for us to look over before making the change official. But it will go in effect before Gencon, after all regionals are completed.

do not like the one where it is sos vs same or better. Can you tell me what precisely is wrong with just plain old overall SOS?

Assuming 6 rd tourney

rd 1 lose to undefeated (or 5-1)
rd 2 lose to random with bad matchup who ends up going 3-3
rd 3 win (knock them to at least their 3rd loss)
rd 4 win (knock them to 3rd loss)
rd 5 win (knock them to 3rd loss)
rd 6 win (knock them to 3rd loss)

SOS vs same or better is 100% (or 86% if your loss was to a 5-1) and you are automatically in and you basically got there by beating nobodys.

Wouldn't that be 0%? You only faced 1 person at your level and you lost

nope it would be 100%

SOS of people that you played at yor point level or better. Only one you played went undefeated (100%)

It's not supposed to be the percentage. I'm not sure the best way to word it, but it should essentially be total victories by your opponent's at your point level or better I think.

so then it would penalize you if you haven't played any of them? And if you played 3 or 4 of them you would auto be in? That doesn't seem any more (or less) fair than any other system.

I still don't know what is wrong with plain old SOS?

Nobody can control who they are paired with so anything based on any part of who you played will have the same problem as people cannot control who they play.

In all reality we are talking abot 1 person who lost 2 games (or maybe 2 people) who if they had taken care of business wouldn't e worrying about tie-breakers. They really shouldn't make t8 anyway but we have to cut to a top 8 for convenience sake but it would be nice to cut to a (anybody with 1 loss) scenario.

Not any diffferent than a 2 point victory where commanders are left alive in the back and the answer is you should have played faster so you have time to go track them down.

Look at my initial suggestion. I had RECORD vs players at your level or above.

This was above SoS vs the same.

So if someone went 2-2 vs the top, they advance over guys who went 1-2 or 0-2.

In fact, now that I think of it, it should just be plain old # of WINS vs same level or higher opponents. That way if you beat more of the top level players, it ranks you higher, AS IT SHOULD BE.

It's kinda like a broader H2H of the top guys.

I don't know how many times we have to say it. You get penalized for WINNING against scrubs and players who drop.

You should never be penalized for WINNING. (At least not until much further down the list, as a final tie-breaker, AFTER several others. Even if it is against a scrub.

I DO think you should be rewarded for beating good players.

I DO think you should be penalized for losing to lesser players.

Having the bad luck of getting paired up with scrubs (or players who drop, or byes for that matter) should effect you LESS than the above statements, but if we go straight to SoS it gives equal weight. Unfairly.

This is why it needs to change

This. Just remember, we are trying to determine who had the better day. In the event of players being tied, I'll take the person who played the higher ranking players over those who beat people at the bottom.

which is exactly what SOS does.

Thank You Bill

No it doesn't. At least not in full. We all know a great deal of situations where full SoS does not equate to who we believe had the best day. For example, a person who plays in the first two rounds, a 0-7 and a 1-6 person, and from then on plays top 16 people will still have a mediocre SoS by days end. And the guy who never plays a single top 15 player can have a stronger one.

How many times in the past have we had someone make the top 8 who did not play a single person in the top 15? At least once a year.

At the same time, we also had about once a year, someone who missed the top 8 because of the early round bad luck. Why object to a partial SoS concept that at least attempts to compare fairly.

For example, in your actually thinking, not what the rules say, which day is better?
Player A played: 0-6, 1-5, 4-2, 6-0, 5-1, 5-1 = 21-15
Player B played 3-3, 4-2, 2-4, 3-3, 4-2, 5-1 = 21-15

Both are totally normal outcomes of Swiss, and look exactly the same from total SoS. Player A played at the top tables from round 3 on, and took 2 losses. Player B, lost his first two games and played from the bottom up, only reaching a top table game in the final round.

I'm totally fine with simply fixing H2H to be limited to only when one player has beaten all of the others in the tie. That's a given as a change. But why not while we are at it add additional layers that will continue to help differentiate players?

This conversation is good. it has made me realize things on both sides. Jason does have some good points, but I completely disagree with him on others. SoS alone is still piss poor.

But - he made me realize this:

If we go by SoS vs only your record or better players, that actually gives an unfair advantage to someone who LOST against a scrub. That shouldn't happen. You should be penalized for that.

But if we simply look at # of WINS vs same record or better players, and then # of LOSSES vs lower record players - it actually accounts for this. We might have been over-complicating it - it can be so much more simple. I still like top-half SOS, but only AFTER simple wins and losses are considered.

My NEW suggestion:


1. Score

2. Record

3. Head to Head - ONLY IF A PLAYER HAS BEATEN ALL OTHER(S) AT THIS LEVEL. (IE - if a total of 2 players are tied here and they played, the winner. If 3 players are tied here and 1 beat both the other 2, etc)

4. # of wins ONLY vs players with your same record or better.

5. # of losses ONLY vs players with a worse record

6. Strength of Schedule ONLY vs players you lost to

7. Strength of Schedule ONLY vs players with your same record or better.

8. Full strength of schedule

9. Roll off

Why not take a real life situation and put it in this machine.

Wisconsin Regionals.

Jonny - 3-2, 9pts. Lost to Bill (3-2), and Tim (5-0). Beat Matt (4-1)
Bill - 3-2, 9pts. Lost to Tim (5-0) and Jason (3-2). Beat Jonny (3-2)
Jason K - 3-2, 9pts. Lost to Tim (5-0) and Jake (4-1). Beat Bill (3-2)
Tristan - 3-2, 9pts. Lost to ? and ? Beat nobody at or above same record.

So all are tied for record and points.

First H2H check (step 3) is skipped since there is no player that has beat all the others

4. # of wins vs players at same record or above:
Jonny - 1
Bill - 1
Jason - 1
Tristan - 0

Tristan is eliminated. Re-check H2H, and it is once again skipped since there is no player that has beat all the others.

5. # of losses vs players at same record or below
Jonny - 0
Bill - 0
Jason - 0

6. SoS ONLY vs players you lost to.
Jonny (5-0 + 3-2 = 8-2, 80%)
Bill (5-0 + 3-2 = 8-2, 80%)
Jason (5-0 + 4-1 = 90%)

Jason gets 4th seed. Re-check H2H, and since Jonny lost to Bill, Bill gets 5th seed and Jonny gets 6th.

This IS what we came up with, but through totally different means. Actually if all we changed was the way we rule H2H, then Jonny would have been 4th.

So is this unfair to say that Jason had a better day because he only lost to the 5-0 and 4-1, while Jonny and Bill both lost to a 3-2? I think it's way more fair than taking the complete SoS.

With this system is accounts for WHAT YOU DID much more heavily that luck of the draw.

Reward for beating top players

Punish for losing to lesser players


AND - to top it off - it is super simple to figure out steps 4 and 5 (not that the others are that hard)

H2H
# of wins vs top players
# of losses vs bottom players
SoS of players you lost to
SoS vs top players

in WI

Jonny's best win was a gainst a 4-1

Bill and my best were against a 3-2

Not sure from who had a better day standpoint that Bill or I should have even been considered in this conversation. Jonny was at the top tables all day (playing against all 3 of the top 3 players)

Many Swiss systems use SOS as a tiebreaker (not just the hated WOTC) . Just perusing wikipedia in the chess area

The idea behind the methods based on the games already played is that the player that played the harder competition to achieve the same number of points should be ranked higher. from http://en.wikipedia.org/wiki/Tie-breaki ... ournaments


I guess all I am saying that SOS overall takes into account your overall day not just 1 or 2 matchups. 6 (or7) games is already a small sample size.

Plus SOS is realtively easy to calculate where our 12 step formula is much more complex for essentially the same result.

Also someone dropping DOES NOT hurt your SOS. If anything it helps it. 0-2 added to your SOS is much better than 1-5 (or 2-4)

Answer Flip a coin should be number 3,4,5,6 lol

Someone dropping DOES hurt you, if they would have won games after that.

They would not have hurt you if they were going to keep losing.


Real life situation: 2007, GenCon. I play Ben G from MI round one. I win an incredibly tough match, then he gets paired with Deri in round 2. Ben loses then he drops.

I beat the guy with the same record who ends up with the seventh seed, even though he didn't beat anyone in the top 16.

I lost to the 1st seed and the 3rd seed.

I just miss making the top 8.

Ben hurt me by dropping, badly. Had he just played 2 more matches and won them, (can't see how he wouldn't have), I would have made the top 8.

Now this is valid. In my suggested formula we simply flip #6 and #7, so that we are weighing SoS of the best heavier than SoS of who you lost to (essentially the difference between you and Jonny that day).

BUT I standby that it is necessary to have steps 4 and 5 first. Tristan SHOULD have been eliminated by not playing the top opponents. Full SoS might have come up with that, or it might not have. Goofy situations arise, why not account for them?


6-7 games IS a small sample size, which is why full SoS breaks down.

The sample size is not big enough to give an accurate measure by SoS, that's why we have to eliminate unfair factors of SoS - and focus on the valid ones.

I really think simply looking at WINS vs top opponents and LOSSES vs lesser opponents should be weighed, then move onto SoS variants.

Whether SoS vs your losses or SoS vs same record or better players comes first I don't care, sincerely. We just have to pick one and move on.




The word "essentially" is all the difference. If we can catch one situation out of 100 to make the better player advance, we've succeeded. For something like GenCon, do we really want to be lazy and not take 5 more minutes (literally, how long did it take us to do the math in our heads for this general hypothesis at Wisconsin? 3 minutes?)

My bottom line is this:

SoS is out of your control.

Beating a good opponent is what YOU DID that day

Losing to a bad opponent is what YOU DID that day

except that because as you stated "who you play is out of your contrl" it means exactly that.

I can't control hopw many top 8 (or 16) or whatever cutoff we decide people I play against. Why should someone who only plays 1 game against same record or better be penalized? They don't control who they play. Maybe they could have beaten any number of people at their pts level but for whatever reason weren't matched up against them.

I would rather have a sample size for SOS of 6-7 than 1 or 2.

The question really is which will be right more often as any system will have errors and possible ways to create scenarios where the "wrong" person makes it to the playoffs. I think that the best we can do for this year's gencon IMO is make head to head apply only if you have played and beaten all tied and go with SOS after that. Then at Gencon this year we can actually analyze the tournaments more and see how our various ideas would affect standings/etc and make any future changes from there. I think that overall SOS will be "right" more often than any of the other scenarios.

It is also quite possible that a 4-2 with 12 pts will have played other 4-2 people that for some reason only have 11 or 10 pts and in your scenario they wouldn't be counted in the tb formula because they don't have the same pts or better (ala how we didn't count the 8 pter at WI).