PGETC : Ratings

[Javaness2]

Everyone loves talking about ratings, so here is another excuse to do so.
In the PGETC there are rules to try to ensure that everyone's team is fielding its strongest players at board 1 and its weakest players at board 4.

4.7: The players of a team must be ordered (from 1 to 12) according to the rating list of the European Go Database available at the beginning of the season. The strongest player gets number one. The exact date of the list to be used is published by the project leaders
4.8: The order keeps unchanged for the whole season (see §1 (5)) regardless of any changes in the European rating lists

There are some teams for which this doesn't really work out very well. None of their players have up to date ratings, and no tournaments seem to happen in their countries. We could think about Bulgaria, Georgia, Greece, and until very recently Kazakhstan.
The question then arises, how could you make sure such players are actually following the rules, and not sacrificing board 1 to try to get points on boards 2 to 4.

[kvasir]

This tournament used to be class C. It helped motivate participation for us. It also turned out that some players with similar ranks diverged quickly in rating, giving reason to trust the players list or reason to change it.

It was possible to opt out of the rating, the OP apparently did, robbing me of a rating point or two

Some people objected, especially the German federation, if I remember then they objected that there was possibility for cheating and irregularities, that online games should never be rated, that every game or no game should be rated, and that online games should never be rated. It's my recollection. Since then it was separately decided that this tournament should not be rated and that national federations were free to hold online events as class D tournaments. It's not the most logical arrangement, the PGETC is a well organized, important tournament that brings in money from the sponsor and many people participate in. Class D tournaments on the other hand can be anything I don't know if should say it this way, I'm doing my best to organize an ongoing class D tournament myself, but PGETC is an important event.

I'd prefer that it was still class C. It is motivating to some players and signals that it is supposed to be serious. Cheating and irregularities are in my view better faced head on than in a retreat. Though I did suggest to someone at EGD that there could be a maximum number of rated online games per year per person, when that number of games is reached (i.e. 30, 40, 50,... whatever) they wouldn't count, which could be a way to avoiding some forms of irregularities (i.e. someone playing way too many online games). They could instead be weighted to count less an less, which might be complicating something simple.

[Javaness2]

The PGETC is a serious tournament I agree. As I older and grumpier, I found it harder to take it seriously. It's interesting to understand why. When I used to play chess in the evenings I could forget everything I did previously that day. When I played Go, I couldn't. I was still stuck in front of a computer, using a program that irritated me (anti-mouseslip & lag++), and hmm who knows. I am probably an exception rather than a rule though. The reason it was rated in the first place, as I remember it, was because it was sponsored and the sponsor or the EGF board thought that rating the event would make it seem more serious and important. Something like that. I'm surprised it hasn't stayed on as Class D at least. There is definitely something illogical going on there.

If the players have no EGF rating, should their online server ratings be used instead?

[kvasir]

If the players have no EGF rating, should their online server ratings be used instead?

The players list is required to have higher rated players above lower rated players. I think the spirit of the competition is to make adjustment to the order by rating when the ratings are outdated or don't exist. It is same as how you should declare an accurate rank when you enter a tournament and online rating is only one aspect to consider. Another aspect is that the need to make adjustments is greater with kyu players that can improve quickly and maybe didn't have an accurate rating to begin with. All the same, the rule of this tournament is that players have to be paired in the order of the players list, the players list needs to be realistic, and sandbagging is cheating.

[dsatkas]

There are some teams for which this doesn't really work out very well. None of their players have up to date ratings, and no tournaments seem to happen in their countries. We could think about Bulgaria, Georgia, Greece, and until very recently Kazakhstan.

Indeed, this is a valid observation. I don't know what other teams do, but for what is worth, the Greek team abides by the rules and doesn't sacrifice boards or put them out of order.

[quantumf]

Most players from South Africa do not have an EGD entry, and those that do will have it from a one-off WAGC event attended years ago. The PGETC guidance we have is to use our internal rating system, and have done so since we were added to the tournament.

[Javaness2]

Greece is certainly a country which plays fairly.

National ranks are certainly a good alternative to European, but why did South Africa never submit its tournament games to the EGD? Were you told that this was not allowed? It would essentially be a parallel rating world, apart from a tiny amount of cross-over at events like WAGC, EGC.

[gennan]

I'd expect that for a tournament result table to be entered in the EGD, it needs to be played in Europe (with some exceptions for international tournaments elsewhere where Europeans represent their country, such as the WAGC).

The definition of "played in Europe" might not be very strict (apparently it includes Israel and even Kyrgyzstan), but considering South Africa as "in Europe" seems a stretch to me.

[jlt]

Why not? After all, I learned from Eurovision 2023 that Australia is in Europe.

[Javaness2]

If South Africa was accepted as an observer member, why not let them contribute to the ratings system.

[kvasir]

I don't necessarily see a problem with South Africa entering tournaments in the EGD but it does raise some questions:

• What is the meaning of South Africa's observer status in EGF? Is it cooperation like this?
• Can the EGF or EGD have governance over anything that transgresses at South African tournaments? Do they in France, Germany or UK, for that matter?
• Is there a benefit to entering South Africa's tournaments? Something that potentially outweighs the trouble?
• Is it something anyone wants? Would the South African country organization request to be allowed to enter tournament results?
• What are the interaction between the communities in South Africa and Europe? It is very far away (for me at least).

This might be a funny hypothetical but if it were requested then I think it should be taken seriously.

I understand "European song contest" to mean something that is inclusive of Australia and Africa. Australian entry in the team championship would have time zone issues which would make it less attractive but there was no such issue with South Africa. Kyrgyzstan and Kazakhstan present time zone issues for some participants but not in the same way that Australia would.

As for EGD, to me it is a matter of if there is a wish for this and that the benefit is not outweighed by some real problems, something that I may not know about or related to the list above.

[quantumf]

We did discuss with the EGD administration about entering our tournaments onto the database. The main sticking point as I recall was that our rating system is much more active, as we accept club games into our rating system. This would have required frequent rank adjustments, which the EGD isn't a big fan of. If the benefits were more obvious perhaps we would have pushed harder. Since then (2015ish) the active go community has declined quite a bit, and Covid hasn't helped.

[gennan]

We did discuss with the EGD administration about entering our tournaments onto the database. The main sticking point as I recall was that our rating system is much more active, as we accept club games into our rating system. This would have required frequent rank adjustments, which the EGD isn't a big fan of.

Club games are allowed in the EGD if they are played in an organised and somewhat offical tournament-like setting. I know that some larger clubs send their periodic club competition results to the EGD.

Why would frequent rank adjustments be required for that? Besides, rank adjustments are not a problem in the EGD. People declare different ranks all the time. This doesn't affect the ratings (unless it is a double promotion, in which case the rating is reset to the new rank).
I think it's only a problem when you frequently want to manually adjust EGD ratings, because that is not how the EGD is supposed to be used.
Does the South African go association have some sort of a rating commission? Would someone be responsible for validating rating resetting promotions in a certain range?

Another thing is that uploading results to the EGD is usually done by designated persons who are expected to be vetted by the national go association.
In the past there have been some cases of fraud and then an EGD admin would need to manually purge it from the database. EGD admins are volunteers and they prefer to have as little manual interventions in the EGD as possible.

If there are only few encounters between players from Europe and players from South Africa, there is little benefit in having these mostly separate populations in one database. The South African go association might as well implement their own national rating system, using the EGD algorithm (possibly modified). France and Russia do that and IIRC this is also done in South America.

[quantumf]

>Club games are allowed in the EGD if they are played in an organised and somewhat offical tournament-like setting. I know that some larger clubs send their periodic club competition results to the EGD.

We capture all games, even club night games. We don't play enough tournaments to keep our ranks relevant to our ability, so we believe we need these club games captured. That the EGD tolerates ranks being far behind reality seems like a weakness in the EGD system, but presumably the manual adjustments at tournament time (if necessary) is deemed to be a good enough workaround.

>Why would frequent rank adjustments be required for that? Besides, rank adjustments are not a problem in the EGD. People declare different ranks all the time. This doesn't affect the ratings (unless it is a double promotion, in which case the rating is reset to the new rank).

I might not be remembering the details. In my defense, it was discussed 8 or 9 years ago. We could probably get used to a different system. Nonetheless, asking people to declare their ranks seems a weird thing to build into the system. Why should they have to guess their rank? Why can't they just know (as our system tries)?

>I think it's only a problem when you frequently want to manually adjust EGD ratings, because that is not how the EGD is supposed to be used.

Our system does always reset the rating at every rank adjustment, but I guess there is no particular reason for us to insist on alignment in that regard.

>Does the South African go association have some sort of a rating commission? Would someone be responsible for validating rating resetting promotions in a certain range?

Yes, we do. Mostly they get called upon when players are improving so fast that even our club game capturing can't keep up. Less common these days with fewer new players.

>If there are only few encounters between players from Europe and players from South Africa, there is little benefit in having these mostly separate populations in one database.

I think that's the main thing. When we first joined PGETC I think closer collaboration had been hoped for, but declining go populations and Covid have put paid to that.

>The South African go association might as well implement their own national rating system, using the EGD algorithm (possibly modified). France and Russia do that and IIRC this is also done in South America.

Yeah, we have a system. It's not at all ELO based, but works reasonably well. The main concern, and I'm not sure it really is a concern, is that we are somewhat over-ranked compared to the EGD ranks. This is partly due to some inherent rank inflation in the system (to encourage activity you get a few points just for participation), and partly due to limited interaction with European players. A few players have been demoted locally after playing in the PGETC, so there has been a moderate level of correction. Probably not enough tho, we still lose more than 50% of our games against equally ranked European players. That said, EGD ranks are often way out of date (see my comments above), especially in the lower divisions, and even more so after it was decided to stop capturing PGETC games.

[gennan]

I think there is a gap between what some people expect from the EGD versus what the EGD can reasonably be expected to do.

The EGD was originally conceived to track and compare the levels of stronger European tournament regulars (mostly stronger SDK and dan players). I'd say it works fairly well for that.

But for weaker (DDK) players who progress quickly while playing few tournament games, the EGD can't do much more than just follow the player's progression in their rank declaration history, because the system doesn't have enough data to calculate an accurate rating from game results. The player should have some external source (outside the EGD) to estimate their level to the best of their knowledge and they would just declare that level when participating in tournaments. In the past, this external source would usually be casual club games. Nowadays it can also be online games.

Even within Europe there is variation between different regions. This happens because some regions are more strict in rank promotions than other regions. So it can and does happen that France has tougher ranks than the UK for example. And at lower levels of play there is even more variation across Europe, because weaker players don't participate much in tournaments, so there is little contact between them across Europe.

What we generally do in the clubs that I have attended: players have a club rank that is determined by the relative handicap between them (starting from the top players ranks). The club ranks may drift over time (mostly depending on the influx and improvement of new players). Periodically we then recalibrate (shift) all the club ranks so that on average the club's top players' ranks coincide again with their "official" stable national ranks. The top players' national ranks would also be reflected in the EGD, because these players tend to be on a plateau for years and they tend to play at least a couple of tournaments per year, which is enough to keep their EGD ratings up-to-date.

I think the above is how EGD ratings should be used and this should work reasonably well (after all, this is how it was done before the EGD even existed).
But some countries/clubs won't allow such a procedure. At one extreme, they may force maintaining EGD (or similar) ratings upon their rapidly improving weaker players and won't allow them to declare a higher rank at tournaments than the rank matching their current EGD rating, let alone resetting their EGD rating without permission. Even though they may be quite aware when these players are (much) stronger than their "official" EGD rating, they demand their weaker players to sandbag and "earn" all their EGD rating points in tournament games. I think this gate keeping policy is detrimental, causing them to have a lot of underranked/underrated players at the lower levels and tournament games that aren't fun for either player.
I think this over-reliance on EGD ratings at lower levels has been one of the downsides of the existance of the EGD, but it seems to be gaining ground in the last decade.

So from what you told about ranking in South Africa, I'd say what you're doing is just fine. Joining the EGD will probably result in a better match with EGF ranks at higher levels (assuming sufficient tournament contact between South African dan players and European dan players), but that benefit may not outwheigh the risk of getting caught into this trap of over-reliance on EGD ratings at lower levels.