Accelerated Pairing Systems

[Ian Thompson]

I've created a new topic for this as its got little to do with the British Championship thread it was in.

No, I don't think you are missing anything. I don't think accelerated pairings have any merit.

I'm sure someone will tell me if I'm wrong. I thought accelerated pairings were designed to give modestly rated players a better chance of getting a title norm if they got off to a good start than they would have with normal pairings. I don't think this is a good reason to mess around with the pairings. If someone's good enough to get a title they'll get it eventually anyway, and if the only way they can get one is with assistance from accelerated pairings then they don't deserve to have it.

I think its also questionable whether tournaments using accelerated pairings are valid for title norms. Title regulation 1.42g says "Tournaments that make changes to favour one or more players shall be excluded." Is that not exactly what accelerated pairings are doing?

Aren't they better suited to tournaments with rather fewer rounds and rather more competitors?

I don't think so.

If there are so many players in an event that two (or more) of them can finish with 100% scores, then no pairing system is going to prevent that.

If a normal pairing system is used and two players score 100%, you would expect them to be the two highest ranked players in the event. If an accelerated pairing system is used, then you would expect the two players to be the highest ranked player and some lower ranked player who had the good fortune to avoid playing the best players. Where's the benefit in that compared to a normal pairing system?

What's more likely to happen wih an accelerated pairing system is that no-one gets 100%, so you get a mass tie on a lower score. Again, what's the benefit of that?

[JustinHorton]

If there are so many players in an event that two (or more) of them can finish with 100% scores, then no pairing system is going to prevent that.

Indeed, but aren't we talking about "make substantially less likely" rather than "prevent"?

[Ian Thompson]

If there are so many players in an event that two (or more) of them can finish with 100% scores, then no pairing system is going to prevent that.

Indeed, but aren't we talking about "make substantially less likely" rather than "prevent"?

Yes, you do make it less likely that two players finish equal first with 100% scores, but you make it more likely that several players finish equal first with less 100%. Why would you want to that?

[JustinHorton]

Because equal firsts per se aren't considered to be so much of a problem as the situation where two players score 100% and (obviously) never meet each other?

[IM Jack Rudd]

I think its also questionable whether tournaments using accelerated pairings are valid for title norms. Title regulation 1.42g says "Tournaments that make changes to favour one or more players shall be excluded." Is that not exactly what accelerated pairings are doing?

I don't think so. The intention of the rule is that you aren't allowed to deviate from the correct pairings (under whatever system you're using) in order to make title norms more likely. It says nothing about what systems you are allowed to use in the first place.

[Ian Thompson]

Because equal firsts per se aren't considered to be so much of a problem as the situation where two players score 100% and (obviously) never meet each other?

Well perhaps they should be when the pairing system means that some players finishing equal first are likely to have played much stronger opposition than other players finishing equal first.

I think its well known that a normal swiss pairing system is good at finding the best players and worst players, but no good at ranking players who finish in the middle of a tournament because they will have played opposition of very different strengths, depending on whether they had a good start and poor finish, or vice versa. An accelerated pairing system exacerbates this effect so its not confined to players who finish in the middle of a tournament.

[Alex Holowczak]

the situation where two players score 100% and (obviously) never meet each other?

The only way that could ever happen is if log n / log 2 > x, where n is the number of players, and x is the number of rounds in the tournament. Since 75 players entered the Championship, you only need 7 rounds to avoid that possibility, and the event has 11. A more likely time that'll happen is if you have > 32 players in a section of a 5-round weekend congress.

[JustinHorton]

Indeed

[Sean Hewitt]

There are two main reasons for using accelerated pairings that I'm aware of.

The first is to avoid huge mismatches in the early rounds where almost every game goes to the higher rated player because the average rating differential is 300 or more. Such results renders the round practically meaningless.

The second reason is to make it less likely in shorter events that two players score 100%. This does not make it easier for the lower rated player to score 100% however as you always have at least the last two rounds paired normally - and such a player will have played opposition rated higher than him in at least the two rounds prior to the acceleration "unwinding".

[Ian Thompson]

There are two main reasons for using accelerated pairings that I'm aware of.

The first is to avoid huge mismatches in the early rounds where almost every game goes to the higher rated player because the average rating differential is 300 or more. Such results renders the round practically meaningless.

So all you succeed in doing is to delay the mismatches until the first round where accelerated pairing are not used. In round 4 of the British Championship 13 of the games had rating differences exceeding 200 points and 2 had differences of more than 300 points.

This does not make it easier for the lower rated player to score 100%

If that's true (and I doubt that it is) then it implies either:

1. The lower rated player has played exactly the same strength of opposition as he would have done with normal pairings, in which case all the accelerated pairings have done is to change the order in which he plays his opponents in the tournament. Overall it has had no effect, so it was pointless.

2. The lower rated player has played stronger opposition than he would have done with normal pairings, in which case a higher rated player must have played weaker opposition than he would have done with normal pairings. Overall the accelerated pairings have made it more difficult for a lower rated player to win the event and easier for a higher rated player, which I don't think many people would consider desirable.

[Alex Holowczak]

1. The lower rated player has played exactly the same strength of opposition as he would have done with normal pairings, in which case all the accelerated pairings have done is to change the order in which he plays his opponents in the tournament. Overall it has had no effect, so it was pointless.

2. The lower rated player has played stronger opposition than he would have done with normal pairings, in which case a higher rated player must have played weaker opposition than he would have done with normal pairings. Overall the accelerated pairings have made it more difficult for a lower rated player to win the event and easier for a higher rated player, which I don't think many people would consider desirable.

Well, that could be important for tie-breakers. I've played in tournaments where the sum of progressive scores is a tie-breaker. So the order you play opponents could be very important. If a tie-breaker is scores of opposition, or grading performance of the opposition, an alternative path through the tournament to get through the tournament will make a difference to those tie-breakers.

[JustinHorton]

Tournaments round my way often use sum of progressive scores. This tends to penalise me (as I tend to drop a lot of draws and indeed whole points to lower-rated opposition and come through the pack rather than lead from the front) and rightly so, I think.

[Alex Holowczak]

Tournaments round my way often use sum of progressive scores. This tends to penalise me (as I tend to drop a lot of draws and indeed whole points to lower-rated opposition and come through the pack rather than lead from the front) and rightly so, I think.

Agreed. However, with the accelerated pairings, you're playing someone stronger earlier, so the progressive scores will favour the third quarter of players, rather than the top half.

[Ian Thompson]

Well, that could be important for tie-breakers. I've played in tournaments where the sum of progressive scores is a tie-breaker. So the order you play opponents could be very important. If a tie-breaker is scores of opposition, or grading performance of the opposition, an alternative path through the tournament to get through the tournament will make a difference to those tie-breakers.

Yes, and it can be worse than that if the prize money is distributed using the Hort system and sum of progressive scores is used to allocate the prizes. I've played in a couple of tournaments in France where I've been affected by that.

The first time I and a friend both scored 5/7, both having beaten all the lower rated players we played and lost to the 2 higher rated ones. Because I was rated a little higher than him, I went one round longer than him before I lost, so my sum of progressive scores was better than his. I got about 5 times as much prize money as he did.

The second time, I had an upset loss to a lower rated player in an early round. I got considerably less for my final score of 5/7 than anyone else on the same final score.

[Matthew Turner]

I have always thought that using Accelerated pairings for the British was wrong. There is no reason why you couldn't get a norm with normal swiss pairings (you could discard two wins against weaker opposition if necessary to increase your average rating).
However, I think Accelerated pairings worsens the Championships. Take a look at David Howell. He played Mark Hebden in Round 2, so it is perfectly possible that he could be due for ten 'top' players in a row. There is a serious chance that he could have played all the other contenders by round 9 and a couple of other will have KO'ed themselves, so he might end up with relatively weak opposition in the last couple of rounds, which could lead to an anti-climactic finish. Why not have a couple of rounds of mismatches to start off with and build to a crescendo.