Junior Grades

[Neil Graham]

I have just noted a young 7-year old player with a rapidplay grade of "0". Examination of the past six month period shows he played 14 games, winning six and losing eight.

How can anyone possibly end up with a grade of zero unless they lost every game they'd played? I assume that a player cannot be scored with a minus grade for losing a game. That would indicate to me that they didn't even know the moves and if this youngster won six games surely he must be given credit for these. In addition he should also get an age addition shouldn't he?

Imagine coming home after your rapidplay event and saying "I did quite well, I nearly won half my games - I'm really looking forward to getting my first published grade". That is until you see it comes out as "Nil"!


Addition to above - another junior rapidplay grade for an eight year old - Jan 2012 P10 W5 D3 L2 Grade "0" . Opponents (all juniors) had eight "Nil" grades, two with a grade of "4".

[Alex Holowczak]

How can anyone possibly end up with a grade of zero unless they lost every game they'd played?

I think you answered your own question:

Addition to above - another junior rapidplay grade for an eight year old - Jan 2012 P10 W5 D3 L2 Grade "0" . Opponents (all juniors) had eight "Nil" grades, two with a grade of "4".

Not so much a flaw in the grading system, but a floor in the grading system.

Why were these tournaments even graded?

[Matthew Turner]

Since all juniors start the grading process with no grade, then if you have 'pockets' of juniors who just play other juniors then they will all end up with grades of 0. It will not matter if they lose all their games or win all their games.

[Christopher Kreuzer]

Presumably my first-ever grade was obtained by beating people with a grade? Is the lesson here that juniors should only really be graded once they start beating (or drawing with) people with grades?

[Roger de Coverly]

Not so much a flaw in the grading system, but a floor in the grading system.

Logically those who lose to players graded below 50 get a negative performance for those games. If that isn't offset by positive performances, the grade comes out negative. The grading team took the decision in principle many years ago not to publish negative grades, but to reset them to zero. The issue of minimum grades was one that should have been looked at as part of the regrading exercise, as allowing grades under 50 presumably has a delaying effect on the increase in grades for improving players.

In its original form, with the 1a to 9b grading codes, the Clarke system had a floor of around 100. Even after it went purely numeric, it was believed that the default for new players was e100. Events containing very weak players wouldn't have been graded in the days when it was all done with pen and paper.

[Neil Graham]

Surely you cannot get a minus grade for losing a game. We assume the minimum grade on the system to be "Nil" (as it clearly is otherwise a large number of these juniors would have a minus grade). If we had minus grades they would be "anti-chess" which would indicate that the players couldn't actually play the game at all. Once we establish a baseline of nil and supposing our player who played 14 games winning six and losing eight played 14 players each graded nil he would score 8 x nil for the losses; 6 x 50 for the wins making his grade 14 x 300 = 21. if each of the 14 players started with a nil grade as presumably they all did; there must be a way to determine that the player who scored +5; =3; -2 is better than the player who scored +0; =0; -10 who would be the only one with an actual grade of "nil" when the gradings were published.

[Roger de Coverly]

Surely you cannot get a minus grade for losing a game.

Of course you can. The grading scale is arbitrary and contains no information about the ability to play chess. Zero is where it is, because Sir Richard Clarke chose 248 as the top number in his grading scheme. He could equally have chosen 298 or 348.

Zero is where players score 10% against those with a grade of 40, or 25% against those with a grade of 25.

We assume the minimum grade on the system to be "Nil"

I would make no such assumption. I don't think it's any more sophisticated than resetting all grades that come out negative to zero. At the risk of inflating the lower end of the grading scale, it might be better to treat 51 as a minimum grade for calculation purposes. This would ensure that all published grades, which would only be used for pairings, eligibility and board orders came out as at least 1.

You could publish negative grades, the grading team and perhaps the wider BCF didn't want to. The same effect would be achieved by arbitrarily adding 100 or 500 to all existing grades.

[E Michael White]

There are several issues with grades this year apart from the continued use of junior increments in the muddled way the ECF have made their own. I know several adults who started/returned to chess and showed increases of 10/15 points per year. Why aren’t their grades subject to predictive correction and the increment related to the number of games/years since starting/returning to chess ? Alternatively a better general prediction method could be employed.

The file from the grading download section shows :-

• 6 juniors with RP grades of 0 who played 0 games in the last 12 months
• 5 of those were shown as RP E grades and 1 a D
• 58 shown as D/E grades and 0 games shown

Are these correct ?

If the iterative process is used for juniors, just as at the top end where a few very high grades emerge some very low grades will emerge as a result of losing to other juniors who lose to others etc. The difference at the lower end is that there are many more in the same situation, which means that many players will find it difficult to move up from those grades due to the inertia effect discussed on this forum previously of playing against equally biased grades.

[Alex Holowczak]

The file from the grading download section shows :-

• 6 juniors with RP grades of 0 who played 0 games in the last 12 months
• 5 of those were shown as RP E grades and 1 a D
• 58 shown as D/E grades and 0 games shown

Are these correct ?

Yes, I'm sure they are.

Why not e-mail Richard Haddrell or Howard Grist though? After all, they can explain it if you e-mail them, whereas they might miss the question here. And it'd save 20 posts conjecturing about whether the ECF have got the grades right or not.

[Adam Raoof]

I think it is essential that as many junior events are graded as possible. This is for a variety of reasons, not least of which is that I believe it would add value to the events, and might encourage more juniors to break out into other tournaments. As they get older they might be more likely to keep playing, rather than completely give up.

Technically it would be possible to have a separate grading list for junior-only events, which could be updated monthly as results came in via the usual channel. It could appear on the ECF grading database as a separate field and when juniors entered open competitions without a regular ECF grade it could be treated like any other source of information.

I don't understand the ins and outs of the arguments to do with grading algorithm (as you will easily see), but it must be possible to grade events with no ECF graded players at all retroactively based on the final results? How was the very first ECF graded tournament graded otherwise?

[Paul Sanders]

Very quick idea which is not at all thought through - what about a Junior Elo that is conditionally convertible to an ECF grade, for instance if there is enough data or the rating is high enough? Then you can get data for pairings, and can make strategic decisions about how and when to move players into the main grading system.

[Adam Raoof]

Very quick idea which is not at all thought through - what about a Junior Elo that is conditionally convertible to an ECF grade, for instance if there is enough data or the rating is high enough? Then you can get data for pairings, and can make strategic decisions about how and when to move players into the main grading system.

I resisted mentioning Elo It would be possible, and it would make rating junior events easy. Would it be so easy to do the grading work? Using ECF grading it would be very simple as a lot of junior events use Tournament Director.

[Paul Sanders]

Using ECF grading it would be very simple as a lot of junior events use Tournament Director.

Do you make new roads for old cars, or for cars not yet launched? If any ECF policy is predicated on Tournament Director then it will take a very long time indeed to change, if indeed any change is desired!

[Alex Holowczak]

I don't understand the ins and outs of the arguments to do with grading algorithm (as you will easily see), but it must be possible to grade events with no ECF graded players at all retroactively based on the final results?

Someone did this in Warwickshire for an Under 11 event. We managed to have kids with grades of 100+ who weren't even candidates for selection to Warwickshire junior teams, but players who were regular players in such teams who had grades of about 100. The regulars, who hadn't played in that event, were of a significantly higher quality. This is why we shouldn't grade things where everyone(ish) is ungraded. Richard Haddrell doesn't grade all National Schools matches for this reason; he just chooses matches where it's a good idea to do so.

How was the very first ECF graded tournament graded otherwise?

Probably with a series of special rules that were implemented as a kick-start.

For example, I did an Elo test cricket software thing, where I applied the Elo algorithm to all results in the history of Test cricket. (I got bored when I got to 1997...) To do this, I started England and Australia at 2500 in 1877. They were the only teams who played for about 15 years. I then set a criterion about how new countries joined the system. I said that a new country's initial rating would be using the initial rating system that FIDE uses applied to their last 10 rated games, so long as one of those 10 games was a win. (As opposed to two draws.) This was completely arbitrary, but seemed to be a sensible way on which to base an initial rating, because it meant that a team like New Zealand - who didn't win a game for 25 years or so - wouldn't keep going lower and lower and lower before they won, meaning that the whole system got dragged down. If New Zealand also started at 2500, then the teams who played them first would get maximum benefit from beating them, whereas the other teams would get a minimal benefit when their rating plummeted to 2200 by the end of a series or two.

However, the point is, the two cases are distinct:
(1) How you kick off your rating system
(2) How new players join your rating system

It seems sensible to me that with (1), you start with a small group of people who have an arbitrary initial rating, and that (2) are treated as ungraded, rather than as you'd treat them in (1).

[Roger de Coverly]

I don't understand the ins and outs of the arguments to do with grading algorithm (as you will easily see), but it must be possible to grade events with no ECF graded players at all retroactively based on the final results? How was the very first ECF graded tournament graded otherwise?

You get a ranking order where the differences between the grades are derived from results but you don't know the absolute values. So in the very first season or seasons of BCF grading, they would have had to have decided what the grade of the top player (Penrose?) was to be and then everything else, including the relative strength of players with a zero grade, fell into place. That's what they did with the regrading exercise, cooking the results so that the top players stayed more or less the same.