Tournamant Pairings in Round 1

[hammy]

The problem with AP at Melbourne was the rating system used to determine the tiers had a flaw in it and put quite a few people in the wrong tier. This was fixed in time for the next IWF.

True but if you look at the tournament used to 'test' the AP system the player that won played one player from the top quarter of the rankings while the one who placed 2nd played IIRC five from the top 10 This was with supposedly reliable ratings as the comp only included Australian players.

In Athens there were several bottom half vs bottom half games at the top of the comp in rounds 3 and 4 as most if not all the 'top guns' only managed one win and one draw or winning draw in the first two games.

I understand the theory of AP but it is both difficult to understand and produces IMO far less good final results than a random or even seeded draw.

I have won large tournaments from both seeded and non seeded (playing a seed in round 1) starts. I have also done less than brilliantly in both situations. For me I like to see the games later on in a tournament being the hard ones as it means the tension and pressure builds as the event goes on.

[nikgaukroger]

The problem with AP at Melbourne was the rating system used to determine the tiers had a flaw in it and put quite a few people in the wrong tier. This was fixed in time for the next IWF.

That'd be the one won by Graham Evans who said the games got easier as the comp went on Peoples gut instinct is that games should get harder as the comp goes on (especially at the top end) which means that if this doesn't happen they think the system is wrong.

The main issue for the DBMM "world ratings" is that they were not relaibale, even after the huge fudge applied, which really rather discredited them. In terms of something to be used at a comp I don't think they're worth the effort to be honest.

[hammy]

The problem with AP at Melbourne was the rating system used to determine the tiers had a flaw in it and put quite a few people in the wrong tier. This was fixed in time for the next IWF.

That'd be the one won by Graham Evans who said the games got easier as the comp went on Peoples gut instinct is that games should get harder as the comp goes on (especially at the top end) which means that if this doesn't happen they think the system is wrong.

The main issue for the DBMM "world ratings" is that they were not relaibale, even after the huge fudge applied, which really rather discredited them. In terms of something to be used at a comp I don't think they're worth the effort to be honest.

Graham won in Athens, Dave Handley won in Melbourne

Also I think you may have added an extra M somewhere.

FWIW I think the final DBM Glicko was plausible but there were several players ranked very highly from countries I had no experience of playing people from so while most of the players at the top were in line with my gut feelings (which I trust a lot when it comes to player ability) there were some I just had to take on trust.

[grahambriggs]

The problem with AP at Melbourne was the rating system used to determine the tiers had a flaw in it and put quite a few people in the wrong tier. This was fixed in time for the next IWF.

That'd be the one won by Graham Evans who said the games got easier as the comp went on Peoples gut instinct is that games should get harder as the comp goes on (especially at the top end) which means that if this doesn't happen they think the system is wrong.

The main issue for the DBMM "world ratings" is that they were not relaibale, even after the huge fudge applied, which really rather discredited them. In terms of something to be used at a comp I don't think they're worth the effort to be honest.

Dave Handley won it, to a degree because locals chose to fight his unbreakable Saba hordes head on. He did fight Terry Shipman-Shaw in the final round though.

[lawrenceg]

The problem with AP at Melbourne was the rating system used to determine the tiers had a flaw in it and put quite a few people in the wrong tier. This was fixed in time for the next IWF.

True but if you look at the tournament used to 'test' the AP system the player that won played one player from the top quarter of the rankings while the one who placed 2nd played IIRC five from the top 10 This was with supposedly reliable ratings as the comp only included Australian players.

Not according to www.iworg.com/files/AcceleratedPair(v2).doc

What’s effect did Accelerated Pairing have? For Cancon, the eventual winner was put up against the best players the tournament had to offer. The winner Bram Jakins went through 3 out of the 5 top ranked players in the event. Similar results were achieved at the state championships.

In Athens there were several bottom half vs bottom half games at the top of the comp in rounds 3 and 4 as most if not all the 'top guns' only managed one win and one draw or winning draw in the first two games.

The winner and 3rd place at Athens both won their first two games. 2nd place won and then drew, but do you feel his ultimate finishing position was influenced by easy pairings overall?

I understand the theory of AP but it is both difficult to understand and produces IMO far less good final results than a random or even seeded draw.

It is intended to give the best results for the winner, so possibly it gives better results for the top places at a cost of worse results lower down.

For me I like to see the games later on in a tournament being the hard ones as it means the tension and pressure builds as the event goes on.

You could enforce this by giving the seedings a high influence on pairing in all rounds. One of the problems with Swiss Chess as we use it is top guns are likely to win their first game big and then play another top gun in the second round. Maybe we should be using a "retarded pairings" system rather than accelerated, to stop them meeting until later.

Round 1: top half vs bottom half
Round 2: top quarter vs 2nd quarter, 3rd vs 4th.
Round 3: top eighth vs 2nd eighth etc
Round 4: top 16th vs 2nd 16th etc. or purely on points.

I think this might give the best unseeded player a relatively high chance of winning the tournament through an easy draw, but you can get the general idea.

Whatever scheme one comes up with, it can always be messed up by players winning games they were not expected to win.

[dave_r]

The main problem with Accelerated Pairing is the number of lower tier players who suddenly get unleashed in round three.

Since you can only play in your own "half" in the first two rounds then potentially 25% of the bottom half are on two big wins, meaning that the one or two players from the top half get an easy remaining draw.

Jason Broomer mentioned that at Dublin the first two games were difficult, but after that it was easy.

[lawrenceg]

You can read my previous post if you quote it. For some reason it's not showing in the normal view.

[nikgaukroger]

Clearly discussions on accelerated pairings are so dull that the forum software blanks it

[lawrenceg]

The main problem with Accelerated Pairing is the number of lower tier players who suddenly get unleashed in round three.

Since you can only play in your own "half" in the first two rounds then potentially 25% of the bottom half are on two big wins, meaning that the one or two players from the top half get an easy remaining draw.

Jason Broomer mentioned that at Dublin the first two games were difficult, but after that it was easy.

Not quite, the second round is not restricted to own half.

In theory, the bottom quartile players all lose in the first round to third quartile.

Then all bottom half players that won lose in the second round to top half players.

With a random first round, potentially 50% of the bottom half are on two big wins after two rounds. With a seeded first round the chances of this are greater.

Of course, there are some inconsiderate low ranked players who insist on beating their betters (The impudence of these upstarts!), which messes up any pairing system based on scores.

[hammy]

I fixed the problem and have replied. The problem was the () in the url Lawrence posted.

True but if you look at the tournament used to 'test' the AP system the player that won played one player from the top quarter of the rankings while the one who placed 2nd played IIRC five from the top 10 This was with supposedly reliable ratings as the comp only included Australian players.

Not according to www.iworg.com/files/AcceleratedPair(v2).doc

What’s effect did Accelerated Pairing have? For Cancon, the eventual winner was put up against the best players the tournament had to offer. The winner Bram Jakins went through 3 out of the 5 top ranked players in the event. Similar results were achieved at the state championships.

The event I was talking about was Cancon 05 which was I believe the first outing of AP as a draw system. The player who won played players ranked 74,29,36,59,45 and 51 while the player who came second played 37,17,16,5,4 & 1

I understand the theory of AP but it is both difficult to understand and produces IMO far less good final results than a random or even seeded draw.

It is intended to give the best results for the winner, so possibly it gives better results for the top places at a cost of worse results lower down.

I am still anything but convinced that it is better even in terms of making the top finishers draw harder.

You could enforce this by giving the seedings a high influence on pairing in all rounds. One of the problems with Swiss Chess as we use it is top guns are likely to win their first game big and then play another top gun in the second round. Maybe we should be using a "retarded pairings" system rather than accelerated, to stop them meeting until later.

Round 1: top half vs bottom half
Round 2: top quarter vs 2nd quarter, 3rd vs 4th.
Round 3: top eighth vs 2nd eighth etc
Round 4: top 16th vs 2nd 16th etc. or purely on points.

I think this might give the best unseeded player a relatively high chance of winning the tournament through an easy draw, but you can get the general idea.

I am fairly sure that is exactly what it would do. There are a number of things that need to be considered in wargaming draw systems, one of them is that when good players play there is a higher proportion of draws than when bad players play. Bad players are more likely to make big mistakes and big mistakes lose games. Good players tend not to do so and as a result there are fewer big results between them.

Whatever scheme one comes up with, it can always be messed up by players winning games they were not expected to win.

True but the AP system used for the WICs in the past almost automatically messed itself up by ensuring that going in to the third game there would be a higher than usual number of players from the bottom half of the pool with two wins and that very few players from the top half would be in the same boat.

[hammy]

In theory, the bottom quartile players all lose in the first round to third quartile.

Then all bottom half players that won lose in the second round to top half players.

With a random first round, potentially 50% of the bottom half are on two big wins after two rounds. With a seeded first round the chances of this are greater.

Of course, there are some inconsiderate low ranked players who insist on beating their betters (The impudence of these upstarts!), which messes up any pairing system based on scores.

I suspect that if you look at the actual results from Melbourne and Athens you will find that a significant number of players beat higher ranked opponents.

You will also find a lot of low ranked players won two games on the bounce at the start of the comp

After 2 rounds in Athens the quartiles of the top 20 players were (in order):
2,3,3,1,1,1,1,2,4,2,3,2,3,4,1,1,1,1,3,2

This meant that the four quartile 1 players in 4th thru 7th place essentially had a lottery for an easy or hard game. The top Q1 player gets a Q3 opponent, the second and third Q1 players play each other and the fourth gets a Q2 opponent.

You are more than welcome to have copies of my analysis of the various tournament run using AP but my reading of the information I have gathered is that it is just a load of mumbo jumbo.

[lawrenceg]

True but the AP system used for the WICs in the past almost automatically messed itself up by ensuring that going in to the third game there would be a higher than usual number of players from the bottom half of the pool with two wins and that very few players from the top half would be in the same boat.

To have two big wins, a bottom half player needs a big win against a top half player.

It is difficult enough for a bottom half player to beat a top half one. By your own argument it is even harder to get a big win against a top half player. So the AP system certainly does not ensure it.

If a good player wins his first two games then he may well have an easier third game. But he has already had two difficult games, so overall his draw is not that easy.

Anyway, I thought you weren't going to get involved in this discussion.

[hammy]

True but the AP system used for the WICs in the past almost automatically messed itself up by ensuring that going in to the third game there would be a higher than usual number of players from the bottom half of the pool with two wins and that very few players from the top half would be in the same boat.

To have two big wins, a bottom half player needs a big win against a top half player.

It is difficult enough for a bottom half player to beat a top half one. By your own argument it is even harder to get a big win against a top half player. So the AP system certainly does not ensure it.

If a good player wins his first two games then he may well have an easier third game. But he has already had two difficult games, so overall his draw is not that easy.

Anyway, I thought you weren't going to get involved in this discussion.

I know we said we weren't but

In Athens there were 12 players with two 'big' wins after round 1 (counting 23+ as a big win). Of these there were 4 Q1, 4 Q2, 3 Q3 and 1 Q4

The problem is not the good player having two hard games at the start. It is the round three lottery when the pools are mixed together.

[dave_r]

To have two big wins, a bottom half player needs a big win against a top half player.

Not at Athens and Dublin - the Top half and bottome half were completely separate until round three.

[lawrenceg]

In Athens there were 12 players with two 'big' wins after round 1 (counting 23+ as a big win). Of these there were 4 Q1, 4 Q2, 3 Q3 and 1 Q4

Well, that's not 25% of the bottom half with two big wins.

The problem is not the good player having two hard games at the start. It is the round three lottery when the pools are mixed together.

The problem with a random first round is:

The round one lottery (by definition) and
The round two lottery when top guns with a big win can meet either a top gun with a big win or a poor player with a big win over a poor player.

[dave_r]

The problem with a random first round is:

The round one lottery (by definition)

That is why we do seeding!

The round two lottery when top guns with a big win can meet either a top gun with a big win or a poor player with a big win over a poor player.

Which happens in AP as well, just at a later time so less opportunity to "sort" the chimps from the gibbons.

[madcam2us]

Just as long as the IWF pits me versus 6 players I don't normally face I'll be happy...

In N.O. I got to play 3 americans, 1 Dane & 1 Brit.... Beat the americans, lost to the Dane and the Brit!

Looking to redeem myself vs the f'ners.



Madcam.

[lawrenceg]

To have two big wins, a bottom half player needs a big win against a top half player.

Not at Athens and Dublin - the Top half and bottome half were completely separate until round three.

If that's the case then they were doing it wrong.

[nikgaukroger]

Just as long as the IWF pits me versus 6 players I don't normally face I'll be happy...

A view I suspect that is shared by most players

In fact I think that for the vast majority of players the "problems" that AP was supposed to solved just didn't exist.

In N.O. I got to play 3 americans, 1 Dane & 1 Brit.... Beat the americans, lost to the Dane and the Brit!

Looking to redeem myself vs the f'ners.



Madcam.

I suspect that this year is the year where an American has the best chance of winning the IWF.

[hammy]

To have two big wins, a bottom half player needs a big win against a top half player.

Not at Athens and Dublin - the Top half and bottome half were completely separate until round three.

If that's the case then they were doing it wrong.

You are assuming that Dave is right Lawrence. As usual, he isn't

The second round was bottom half winners against top half losers but even so four bottom half winners managed that feat.