4ncl Congress Entry Conditions Change

Venues, fixtures, teams and related matters.
Brian Towers
Posts: 1058
Joined: Tue Nov 18, 2014 7:23 pm

Re: 4ncl Congress Entry Conditions Change

Post by Brian Towers » Fri Aug 17, 2018 7:48 pm

Nick Burrows wrote:
Fri Aug 17, 2018 1:17 pm
Roger de Coverly wrote:
Thu Aug 16, 2018 4:47 pm
Arguably though, players with ECF grades above 175 shouldn't be allowed in the second section of weekend tournaments unless it's expected that the top section will be exceptionally strong and well supported by titled players.
What do you base that on?
I would expect the answer to be something of the form - The top X% of players have grades over 175 and X% seems a reasonable cut off point for most competitions.

The question is what is X? I dare say Roger has all the data at his finger tips or else knows which section of which website to find exactly what he wants. For the rest of us it's a bit more tricky.

The ECF grading website will let me see who are the top 106 English players in the ECF list (7 players are tied on 214, hence the strange number) but it won't easily let me see how many players have a grading above 175. One solution is to download the latest ECF grading file, find someone (English) with a grading of 175, then back on the ECF grading website look at their profile to see their "Rank, excluding FIDE registered foreign player".

Doing that gives me 883 English players graded above 175.

Next step, find out how many English players have a grade using the same technique.

Doing that gives me 8960 players with a grade higher than the grade of the lowest graded players.

So, X looks like 10% almost spot on because a lot more than 13 (896-883) players have a grading of 175.

The top 10% of players should be in the Open? Sounds reasonable to me.

Coincidence? Or is RdC a lot cleverer than he lets on?
Ah, but I was so much older then. I'm younger than that now.

Nick Burrows
Posts: 766
Joined: Sat Aug 14, 2010 12:15 pm

Re: 4ncl Congress Entry Conditions Change

Post by Nick Burrows » Fri Aug 17, 2018 10:48 pm

Brian Towers wrote:
Fri Aug 17, 2018 7:48 pm
The top 10% of players should be in the Open? Sounds reasonable to me.
10% is still an arbitrary cut off. Why not 5%? or 33% in a 3 section event?

At the last 4ncl congress, using 10% or 175 ecf as the cut off, creates a rating variance in the open of 69 points between the 1st and last seed. 45 in the major section, and 50 in the minor. Therefore penalising the 175-190 ecf players who want "competitive" chess. If you choose to "play up" - great. But the majority choose not to. What are the benefits to forcing them? Especially if you risk them not playing at all.

Isn't the primary objective in holding a congress to attract as many players as possible to play satisfying games of chess?

Roger de Coverly
Posts: 16998
Joined: Tue Apr 15, 2008 2:51 pm

Re: 4ncl Congress Entry Conditions Change

Post by Roger de Coverly » Fri Aug 17, 2018 11:06 pm

Nick Burrows wrote:
Fri Aug 17, 2018 10:48 pm
Isn't the primary objective in holding a congress to attract as many players as possible to play satisfying games of chess?
Collecting the scalps of IMs, GMs even , is a fun sport. Too many of them and you get trampled underfoot.

It's only been the 4NCL Congresses and before that the e2e4 ones that had a second section defined by under 2100, 2050, 2000 or 1950. With the exceptions of Kidlington and Blackpool, the second section would have been under 170, 165, 160 or 155. Even Kidlington was under 170 until the revaluation.

Nick Burrows
Posts: 766
Joined: Sat Aug 14, 2010 12:15 pm

Re: 4ncl Congress Entry Conditions Change

Post by Nick Burrows » Fri Aug 17, 2018 11:45 pm

I'm sure it's a fun sport if you're good enough to do it.

Re: Kidlington, the top section was capped (at 210 ecf I think) which incidentally created more competitive sections for the majority at the expense of the elite.

A minor U140, Major U190 and Open creates 3 sections with approximate bands of 50 ecf points. And seems equitable across the playing spectrum.

Based on the previous 4ncl congress and using the same percentage of players that chose to "play up" at the previous congress, the numbers in each section would have been:

Open: 26 players
Major: 37 players
Minor: 39 players

Versus the actual 28/37/37

Almost no difference in entries, just a more equitable rating distribution, that doesn't penalise a small section (8%) of players

LawrenceCooper
Posts: 4750
Joined: Tue Dec 20, 2011 8:13 am

Re: 4ncl Congress Entry Conditions Change

Post by LawrenceCooper » Sat Aug 18, 2018 6:56 am

Nick Burrows wrote:
Fri Aug 17, 2018 11:45 pm
I'm sure it's a fun sport if you're good enough to do it.

Re: Kidlington, the top section was capped (at 210 ecf I think) which incidentally created more competitive sections for the majority at the expense of the elite.
Kidlington was Under 225 when I used to play, Cherniaev being one entry that I recall, albeit he was swatted aside by a young Ameet Ghasi.

User avatar
Adam Raoof
Posts: 2255
Joined: Sat Oct 04, 2008 4:16 pm
Location: NW4 4UY
Contact:

Re: 4ncl Congress Entry Conditions Change

Post by Adam Raoof » Sat Aug 18, 2018 9:07 am

Hampstead is Under 2200 (any non FIDE rated player with an ECF over 200 could play, and they have)
Under 1900 and Under 160 (=if you have an ECF standard grade then you must be both)
ECF Under 135 (don't care what the FIDE is if they have an ECF standard grade)

this is because several players who were 190 were playing in the Under 1900 with low FIDE ratings, and this seemed a little unfair. I also noticed that in some of the tournaments mentioned you had the same issue, so I took action.

Richard Bates
Posts: 2693
Joined: Fri Nov 14, 2008 8:27 pm

Re: 4ncl Congress Entry Conditions Change

Post by Richard Bates » Sat Aug 18, 2018 12:10 pm

Nick Burrows wrote:
Fri Aug 17, 2018 10:48 pm
Brian Towers wrote:
Fri Aug 17, 2018 7:48 pm
The top 10% of players should be in the Open? Sounds reasonable to me.
10% is still an arbitrary cut off. Why not 5%? or 33% in a 3 section event?

At the last 4ncl congress, using 10% or 175 ecf as the cut off, creates a rating variance in the open of 69 points between the 1st and last seed. 45 in the major section, and 50 in the minor. Therefore penalising the 175-190 ecf players who want "competitive" chess. If you choose to "play up" - great. But the majority choose not to. What are the benefits to forcing them? Especially if you risk them not playing at all.

Isn't the primary objective in holding a congress to attract as many players as possible to play satisfying games of chess?
I would argue that you may be slightly conflating two different things - maximising the competitive/“satisfying” number of games for the players, and having sections tightly enough banded that most players have realistic chances of winning prizes. 4NCL type tournaments tend to be a bit atypical in that they offer incentives for GMs to play (as well as free entry for others) and are, generally stronger at the top end than most regular Opens. Most Opens will always suffer from being at the extreme end of the normal distribution. And whilst the overall grading spread can be quite wide, the median grade will often be less than halfway from top to bottom. And when you exclude the few “outliers” near the top, most games will often be fairly reasonably matched.

Nick Burrows
Posts: 766
Joined: Sat Aug 14, 2010 12:15 pm

Re: 4ncl Congress Entry Conditions Change

Post by Nick Burrows » Sat Aug 18, 2018 12:31 pm

Adam Raoof wrote:
Sat Aug 18, 2018 9:07 am
Hampstead is Under 2200 (any non FIDE rated player with an ECF over 200 could play, and they have)
Yes, thanks Adam. This definitely fills a much needed "gap in the market". It seems to be a relatively popular bracket in large European swisses

Nick Burrows
Posts: 766
Joined: Sat Aug 14, 2010 12:15 pm

Re: 4ncl Congress Entry Conditions Change

Post by Nick Burrows » Sat Aug 18, 2018 12:58 pm

Richard Bates wrote:
Sat Aug 18, 2018 12:10 pm
4NCL type tournaments tend to be a bit atypical in that they offer incentives for GMs to play (as well as free entry for others) and are, generally stronger at the top end than most regular Opens. Most Opens will always suffer from being at the extreme end of the normal distribution. And whilst the overall grading spread can be quite wide, the median grade will often be less than halfway from top to bottom. And when you exclude the few “outliers” near the top, most games will often be fairly reasonably matched.
Yes, I think you are right. Even Re: 4ncl congresses, my initial (and small) calulation of the average grading discrepancy per game in the open vs major was significantly distorted by using FIDE grades. Using ECF grades it doesn't seem to hold up at all. In the opens you have a small number of outliers at both top and bottom ends (low rated players playing up 2 sections!) but then a more solid middle section.

I guess the pure chess experience will be equally interesting for me in an open vs a major. Which just leaves the part about the only tournament that offered a decent prize to 175-190 ecf rated players just disappeared. As i mentioned to Mike, after he kindly emailed me - winning a decent prize is a strong motivator for most players, especially when you are low waged and a typical tournament leaves you a couple of hundred quid out of pocket.

Part of the impetus behind the decision was an attempt to address sandbagging by a minority of players.
Part of the problem in being able to accurately address this issue is the disconnect between the two rating systems, that leaves a gap for a narrow band of players to fall down.

Alex Holowczak
Posts: 8391
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: 4ncl Congress Entry Conditions Change

Post by Alex Holowczak » Sat Aug 18, 2018 9:19 pm

Nick Burrows wrote:
Sat Aug 18, 2018 12:58 pm
Part of the impetus behind the decision was an attempt to address sandbagging by a minority of players.
Part of the problem in being able to accurately address this issue is the disconnect between the two rating systems, that leaves a gap for a narrow band of players to fall down.
The issue with the disconnect isn't an entirely one-way street though. There are the opposite players - people who have a much higher FIDE-rating than their ECF grade converts to. However, people don't tend to complain about them because:
(a) They tend not to enter the FIDE-rated congresses I'm aware of in the first place, because rather than "sharking" the section, they will have the opposite.
(b) People's psychology is such that they tend to view the FIDE system as "wrong" when there is a disparity.

You do get people like this person: https://ratings.fide.com/card.phtml?event=11603925

He entered the Basingstoke Under 1825 section earlier this year, and following that, his rating went above 1825 for the first time, to his lifetime high rating of 1835. He has now dropped back below it again. 1825 converts to 150. If you look up his grade, you see that on all but two grading lists, his ECF grade has been above 150.

Inevitably, there was a complaint that there was a 166-graded player was in an Under 1825 section, even though until that point, his FIDE-rating had never been above 1825. So there couldn't have been a problem with him playing in an Under 1825. He finished in 3rd place in Basingstoke.

We also have to consider that the two systems include different sets of games; evening league games are shorter than FIDE-rated games, except for Adam's events which are of similar length. It might be genuinely true that someone is better at 3-hour games than they are at 4-hour games, and so their grade is higher. This may, or may not, explain Bovtramovics - although his Rapid grade and rating perhaps suggest the opposite might be true!

So I think so long as long as there are two rating systems, you will get these problems. In addition, people always seem to think the FIDE rating is "wrong", and so you always get complaints about those that you never get in the reverse situation.

Nick Burrows
Posts: 766
Joined: Sat Aug 14, 2010 12:15 pm

Re: 4ncl Congress Entry Conditions Change

Post by Nick Burrows » Sat Aug 18, 2018 11:09 pm

Erika Orsagova who played in Basingstoke had a fide rating of 2040 and an ecf of about 136 :?

They do though, seem few and far between. Often players with an ancient Fide rating.

Matt Fletcher
Posts: 209
Joined: Wed Mar 09, 2011 9:42 pm

Re: 4ncl Congress Entry Conditions Change

Post by Matt Fletcher » Sun Aug 19, 2018 12:17 am

I had a couple of goes previously at working out a best (linear) fit converting actual ECF grades to FIDE ratings.

In 2014, FIDE = 7.5xECF + 675 was (close to) the best fit
In 2017, it was FIDE = 7.5xECF + 640

As you can see from the graph I put together in 2014, you do get oddities both ways, across the rating spectrum.

Image

Links to relevant bits of the previous thread:
viewtopic.php?f=4&t=3994&start=60#p131552
viewtopic.php?f=4&t=3994&start=90#p195169

Brian Valentine
Posts: 379
Joined: Fri Apr 03, 2009 1:30 pm

Re: 4ncl Congress Entry Conditions Change

Post by Brian Valentine » Sun Aug 19, 2018 8:28 am

Matt picks out a lot of the issues around this formula. I have reviewed this from time to time and my preferred data sets are 1. All players not identified as junior(in either list) on both lists and 2. Those listed as ENG not junior on both lists. A rule of thumb is that ECF grades are on average about 8 points away from underlying strength. This implies that this conversion difference will be around 11-12 grading points, say 80 Elo.

Graders have always been clear this formula can only be a guide for sectioning events. There are a number of knotty problems.
1. Any conversion has to take into account non FIDE alternatives, e.g Welsh or Scottish Elo.
2. It is not clear that all organisers want the safety margin on the side of FIDE as opposed to on the side of ECF
3. If you fit the conversion from FIDE to ECF, rather than what I think Matt has done, one doesn't get the reciprocal parameters.
4. What to do about juniors?

The grades and ratings are moving around on plates.
1. In an English context we find that running FIDE over a long period it deflates at about 1 Elo point a month on average. There are two contributers that have been identified, there may be others. The first is that where juniors play juniors their average grade stagnates, whereas all the evidence is that they improve. The second is that the initial rating set is not even sided; outperforming against rated opponents is not as highly rewarded as a mirror underperformance. Makes sense at the individual level, side effects on the community, especially with a c15% turnover.
2. The ECF grades in 2009 were set too close together. They are still dispersing to a more stable position.

This discussion is likely to rear up from time to time!
Brian Valentine
Manager ECF grading

Richard Bates
Posts: 2693
Joined: Fri Nov 14, 2008 8:27 pm

Re: 4ncl Congress Entry Conditions Change

Post by Richard Bates » Sun Aug 19, 2018 8:32 am

Matt Fletcher wrote:
Sun Aug 19, 2018 12:17 am
I had a couple of goes previously at working out a best (linear) fit converting actual ECF grades to FIDE ratings.

In 2014, FIDE = 7.5xECF + 675 was (close to) the best fit
In 2017, it was FIDE = 7.5xECF + 640

As you can see from the graph I put together in 2014, you do get oddities both ways, across the rating spectrum.

Image

Links to relevant bits of the previous thread:
viewtopic.php?f=4&t=3994&start=60#p131552
viewtopic.php?f=4&t=3994&start=90#p195169
Might be worth building in a greater level of analysis to factor in activity levels across the two systems, for anyone who wants to make the effort. The ECF lists of course has ready made categories for this (A, B, C ...etc), a bit of thought would have to be given to how to sensibly split FIDE ratings. Logically it also makes sense to compare like with like ie only use January and/or July lists.

Richard Bates
Posts: 2693
Joined: Fri Nov 14, 2008 8:27 pm

Re: 4ncl Congress Entry Conditions Change

Post by Richard Bates » Sun Aug 19, 2018 8:42 am

I guess one other issue is the question of whether increased frequency of lists work towards making ELO system more or less accurate. I think there are offsetting arguments around accuracy vs volatility/‘noise’ which don’t really apply to the ECF system (at least once it reverted to the rolling 1 yr list basis).

Post Reply