International round-up 15/7/12

[Jonathan Rogers]

@ Jonathan - much easier with manual pairings ... if you manipulate the draw to maximise norm chances, then there's every chance of invalidating all norms achieved in the event sadly.

Paul Dargan

oh, has that happened often then?

[David Sedgwick]

@ Jonathan - much easier with manual pairings ... if you manipulate the draw to maximise norm chances, then there's every chance of invalidating all norms achieved in the event sadly.

Paul Dargan

oh, has that happened often then?

Not often, but it has happened.

I think a British organiser would endeavour to ensure that there were the right number and mix of foreigners to make such problems unlikely.

However, as has been stated, Ryan Rhys Griffiths was extremely unlucky.

[Jonathan Rogers]

@ Jonathan - much easier with manual pairings ... if you manipulate the draw to maximise norm chances, then there's every chance of invalidating all norms achieved in the event sadly.

Paul Dargan

oh, has that happened often then?

Not often, but it has happened.

I think a British organiser would endeavour to ensure that there were the right number and mix of foreigners to make such problems unlikely.

However, as has been stated, Ryan Rhys Griffiths was extremely unlucky.

I'd like some statistical assessment of this "very unlucky" claim. I wonder whether the probability of meeting 7/9 Frrench players in this event was not so outlandish, in which case some of the problem is to be attributed to the employment of computer pairings with no possibility of human intervention.

[Sean Hewitt]

I'd like some statistical assessment of this "very unlucky" claim. I wonder whether the probability of meeting 7/9 Frrench players in this event was not so outlandish, in which case some of the problem is to be attributed to the employment of computer pairings with no possibility of human intervention.

Cue Alex If he hasn't done it by the time I get home, I'll do it!

[Jonathan Rogers]

This must be what people have in mind when they talk about the testosterone fuelled environment of the forum

[LawrenceCooper]

I'd like some statistical assessment of this "very unlucky" claim. I wonder whether the probability of meeting 7/9 Frrench players in this event was not so outlandish, in which case some of the problem is to be attributed to the employment of computer pairings with no possibility of human intervention.

Meeting seven in the first seven rounds when there's only 38 out of 63 does seem somewhat unfortunate I'll leave it to others to bombard you with statistics and graphs

[Alex Holowczak]

I'd like some statistical assessment of this "very unlucky" claim. I wonder whether the probability of meeting 7/9 Frrench players in this event was not so outlandish, in which case some of the problem is to be attributed to the employment of computer pairings with no possibility of human intervention.

Cue Alex If he hasn't done it by the time I get home, I'll do it!

Assuming everyone was of equal standard, the probability of meeting a Frenchman was 38/63 in a round.

63C9 / 38C7 / 25C2 = 23667689815 / 12620256 / 300 = 6.25 to 1.

In other words, the probability of meeting 7 Frenchmen and 2 non-Frenchmen is 0.160 (3dp).

This isn't strictly accurate, though. For example, if the 25 foreigners were the bottom 25 seeds, the probability of meeting 2 of them would decrease if you fulfilled every other norm criterion, because you're likely to be on different scoregroups from the foreign players. I haven't the inclination to take the ratings of all of the players into consideration when performing the calculation.

[Christopher Kreuzer]

Can you adjust that probability assessment based on the ratings of the players and the likely score and opposition needed for a norm? If the non-French players were more in the upper or lower half of the ratings, the probability changes, I think, but no idea how to calculate it. (It's really a different question, what probability of meeting 7 French players over 9 rounds if you play to the standard for an IM norm and everyone else plays to their rating level? i.e. Take aim at Alex's assumption 'Assuming everyone was of equal standard' and blow it out of the water).

Oh. It seems Alex had answered that as I typed this...

[Jonathan Rogers]

I ask because we have a lot of norm seekers in our Club and although most are too sensible to consult me, I still like to have some pearls of wisdom for those who do.

It sounds as though one answer to the question "is it worth playing a nine round swiss abroad to get a norm?" might be

"try to find out whether they do computer pairings and roughly how many foreigners played there as a proportion of the overall entry last year. Even in a metropolitan city there can easily be a sixteen or so percent chance that you will in any event fall foul of the foreigner rule, if the event does computerised pairings which don't take special account of the needs of norm seekers. And that is on top of the possibility that by the same token you may fail to get an appropriate pairing from the elo point of view when it matters, or may have played too many unrated or very lowly rated players early on. If a norm is your main objective you'd be much better off if you could find an otherwise feasible ten round event".

[Alex Holowczak]

I ask because we have a lot of norm seekers in our Club and although most are too sensible to consult me, I still like to have some pearls of wisdom for those who do.

It sounds as though one answer to the question "is it worth playing a nine round swiss abroad to get a norm?" might be

"try to find out whether they do computer pairings and roughly how many foreigners played there as a proportion of the overall entry last year. Even in a metropolitan city there can easily be a sixteen or so percent chance that you will in any event fall foul of the foreigner rule, if the event does computerised pairings which don't take special account of the needs of norm seekers. And that is on top of the possibility that by the same token you may fail to get an appropriate pairing from the elo point of view when it matters, or may have played too many unrated or very lowly rated players early on. If a norm is your main objective you'd be much better off if you could find an otherwise feasible ten round event".

It depends on the Swiss. I mentioned 1.43e above. If the event is big enough to meet the foreigner requirement of its own accord, so you don't have to, then the Swiss is fine.

An APA has the insurance of knowing who you're opponents are in advance, and being certain that a norm is on when you sit down in round 1.

[LawrenceCooper]

I ask because we have a lot of norm seekers in our Club and although most are too sensible to consult me, I still like to have some pearls of wisdom for those who do.

It sounds as though one answer to the question "is it worth playing a nine round swiss abroad to get a norm?" might be

"try to find out whether they do computer pairings and roughly how many foreigners played there as a proportion of the overall entry last year. Even in a metropolitan city there can easily be a sixteen or so percent chance that you will in any event fall foul of the foreigner rule, if the event does computerised pairings which don't take special account of the needs of norm seekers. And that is on top of the possibility that by the same token you may fail to get an appropriate pairing from the elo point of view when it matters, or may have played too many unrated or very lowly rated players early on. If a norm is your main objective you'd be much better off if you could find an otherwise feasible ten round event".

It depends on the Swiss. I mentioned 1.43e above. If the event is big enough to meet the foreigner requirement of its own accord, so you don't have to, then the Swiss is fine.

An APA has the insurance of knowing who you're opponents are in advance, and being certain that a norm is on when you sit down in round 1.

That's probably why FIDE charge us so much to run them

[Jonathan Rogers]

That, and probably FIDE assumes that many organisers are corrupt and use APAs to help home favourites to get norms, and reason "well, we can get an extra slice here!"

[Jonathan Rogers]

I ask because we have a lot of norm seekers in our Club and although most are too sensible to consult me, I still like to have some pearls of wisdom for those who do.

It sounds as though one answer to the question "is it worth playing a nine round swiss abroad to get a norm?" might be

"try to find out whether they do computer pairings and roughly how many foreigners played there as a proportion of the overall entry last year. Even in a metropolitan city there can easily be a sixteen or so percent chance that you will in any event fall foul of the foreigner rule, if the event does computerised pairings which don't take special account of the needs of norm seekers. And that is on top of the possibility that by the same token you may fail to get an appropriate pairing from the elo point of view when it matters, or may have played too many unrated or very lowly rated players early on. If a norm is your main objective you'd be much better off if you could find an otherwise feasible ten round event".

It depends on the Swiss. I mentioned 1.43e above. If the event is big enough to meet the foreigner requirement of its own accord, so you don't have to, then the Swiss is fine.

An APA has the insurance of knowing who you're opponents are in advance, and being certain that a norm is on when you sit down in round 1.

Oh yes, APAs are obviously better from the point of view of opponent compatibility anxiety. But not everyone finds them attractive, and one imagines that it is harder to make the required performances, because you are probably competing directly with other under-rated ambitious norm seekers. Put it this way, how often do you see two norms in a APA? My main point was meant to be that those who want norms in open swisses should probably be advised strongly towards ten round events rather than nine (is that obvious? what do seasoned swiss competitors think?)

[Maxim Devereaux]

In other words, the probability of meeting 7 Frenchmen and 2 non-Frenchmen is 0.160 (3dp).

Even in a metropolitan city there can easily be a sixteen or so percent chance that you will in any event fall foul of the foreigner rule...

A point to note is that Alex's probability is that of meeting *exactly* 7 Frenchmen and 2 non. We also need to consider 8 Frenchmen and 1 non, and just nine Frenchmen.

8-1: 63C9 / 38C8 / 25C1 = 23667689815 / 48903492 / 25 = 1 in 19.4 = 0.052
9-0: 63C9 / 38C9 = 23667689815 / 163011640 = 1 in 145 = 0.007

So the chance of failing the foreigner requirement in this tournament is theoretically as high as 22%.

In general, most of the lowest quarter of an international swiss will probably be home players, so the probability is likely nowhere near as bad as this in reality. Another point that hasn't been mentioned is seeding number: if all the foreign players had the same colour in round 1, then the chances of playing only home players would be substantially higher than if they were evenly spread. Even though *some* players change colour sequence in the course of a tournament, there is less pressure on players to do so in a big open, and although I cannot claim to have done any analysis on it, I would be very surprised if there were not a strong bias (of the order of up to 7/9) towards playing opponents with the opposite oddness of seeding number to oneself.

However, one of the biggest factors when choosing an international tournament can be *where* it is. Almost all French tournaments now use a centralised computer system run by the FFE for pairings as far as I know, and they accelerate, so rating average is probably not an issue in a tournament, but the foreigner requirement might be. Elsewhere (e.g. Germany) you might get a huge open with 500 people and no acceleration, and however nice the organisers are, if you play two or more relatively low rated players in the first rounds, it can be very hard to pull the average back up afterwards (especially since you can now only use the title rating floor for one of them), however accommodating the organisers try to be, and number of titled players can also be a problem if you drop an early point or more. I played such a tournament in Germany in 2008: scored 7/9 for 2nd-14th place, with a 2500+ performance, but only played three titled players, despite there being maybe 40-50 (FM or higher) in the event.

So, my advice would be to check the previous year's info about the event, if available, to see the composition, what pairing system they used and so forth. For a long-established event, things are likely to be much the same from year to year, and you can get a fair idea of how good your chances of a norm might be from that.

[Alex Holowczak]

In other words, the probability of meeting 7 Frenchmen and 2 non-Frenchmen is 0.160 (3dp).

Even in a metropolitan city there can easily be a sixteen or so percent chance that you will in any event fall foul of the foreigner rule...

A point to note is that Alex's probability is that of meeting *exactly* 7 Frenchmen and 2 non. We also need to consider 8 Frenchmen and 1 non, and just nine Frenchmen.

You're right. I couldn't be bothered to work that out, since it wasn't part of the question, and the ham, cheese and mozzarella panini I had ordered from the coffee shop I was in had just arrived.