Sorry if this topic has been discussed before and I missed it.
My son Will took clear 1st in the 8th grade and under section of a local scholastic tournament this weekend with 4.5/5. His rating dropped 8 pts.
My issue with the scholastic “other” rating system is where they start the rating for these kids. I’ve seen where ratings start based on grade, so 700 for 7th grade, 800 for 8th grade, etc. Why not start all players at 1000 and let their ratings adjust as they win/lose? This will keep them closer, as a group, to the non-scholastic tournament groups. It’s rather discouraging for kids to play in scholastic tournaments if they are 1000 pts higher than anyone in the scholastic group, when a single draw is costly as was for my son.
My solution is to adjust all current ratings below 1000 to 1000, and make it the Universal floor. This way we’ll strengthen competition in scholastic tournaments.
This topic has been discussed numerous times before.
Some of the issues are:
massive inflation from such a floor
inflation primarily affecting one segment of the rating pool (scholastic players)
incredible reality check once a scholastic player joins the adult rating pool and loses many of the rating points gained from a grossly overrated scholastic pool
4-10) damaging the capability of the rating system to properly reflect the percentage chances of the various results in games between players with a LOT of other ramifications including the USCF rating no longer being considered as a valid rating
A 1726 playing against a 1395, 1069, 1038, 930 and 729 should, on average, score better than 90%. Thus the 90% result is a performance rating in the 1400s and will lower the rating of somebody starting at over 1700.
I do empathize as I’ve had similar situations happen to me as an adult in some club events (with some under-rated adults whose only USCF-rated play is in those events), but that is just how the cookie crumbles.
I agree with the previous poster, and with no disrespect intended, I’m surprised a player of your strong ranking (similar to your son, it seems) is making this suggestion. Massive inflation/ inconsistency/ “rude awakening” results would be realized, no doubt.
Look at it this way (I was curious because of the previous posting, so I also looked at the actual tournament). If a high-B (1726) player competes with E and lower players (i.e., all of your son’s opponents except one), I think the former should win 95% of the time. There is a huge disparity between those levels. And further, playing a high D player (1395), the B player should win, say, 75% of the time. Would you agree? This arithmetic yields a 61% chance that this B player scores a 5.0 in the tournament.
And if he scores a 5.0, his rating would have seen only a modest boost - again, the result of the expectation. He’d have gone from 1726 to 1731. I think that the 8-point drop is (a) certainly quite inconsequential, and (b) appropriate for not meeting the reasonable expectation of a 5.0. Either change, admittedly, is very minor - as with any single-digit rating shift.
Now look further at what would have happened if the players had a “1000 floor” as suggested. Raise the ratings of the bottom 2 players to exactly 1000. The impact on your son’s rating would have been: 1718 at 4.5; 1731 at 5.0. NO CHANGE! So the rating system inherently takes into account the very strong expectation that a 1726 should beat a 930 or 729 with near certainty, and should do virtually the same against a 1000.
Then you might say that this ‘floor’ (hypothesized) would have also boosted the other players ratings - maybe to 1500, 1200, 1100 (respectively) - and with that, your son’s result would have been enhanced. But check that case out, too - and note I’m applying some significant ‘ups’ to those other ratings - and in each case (4.5 and 5.0), the outcome is only shifted by three rating points (1721, 1734). The impact of this fairly dramatic input data shift is nearly nil - again, reflective of the striking difference between a B player and the other competitors. But in any event, to suggest this type of floor and such shifts is an inappropriate rating inflation, which will spiral in the scholastic ranks - with all the downsides mentioned earlier.
I think, if anything, the runner-up’s experience in that particular tournament is more striking than your son’s. This student started out as a sub-1000, scored 4.0, and saw their rating drop, also (only 5 points in this case). This is more surprising (to me) than your son’s outcome, yet it also accurately reflects the arithmetic of the rating system - which MUST be applied universally to assure any kind of predictive stability.
I’m not sure what ‘other’ system you’re referring to.
USCF ratings for players under 26 are started based on the player’s age. For example, someone who is exactly 8 years old has a pre-event rating of 400 based on zero games.
That estimate has only a minimal effect on his or her first post-event rating, because it’s really based on the ratings of his or her opponents using the ‘special’, which is essentially the old ‘Win+400, Loss-400’ provisional/performance rating formula, but it is used in the initial steps of the ratings formula to come up with a (hopefully) more accurate estimate to use in the final step, when all the post-event ratings are computed.
Flooring all ratings at 1000 would effectively end the usefulness of the ratings system for scholastic players, because 87% of the recently active players under 16 would all have the same rating–1000.
The problem with grade x 100 as an estimate is that you can have 3rd grade students who range from 7 to 10 years old, though the majority of them will be 8 at the start of the school year.
That also means that their initial rating would be the same if their first tournament was at the beginning of the school year as it would be at the end of the school year.
The problem is that your son won’t want to play in most scholastic tournaments any more. He has to be perfect or he loses points. Indeed he has little reason to be interested. He may do great and just lose a few points. But if we give players that “should” be 200, ratings of 1000, it will mess up the system as has been well described above.
Would this be yet another place to mention the idea of a separate scholastic rating system? What if this scholastic event were rated in a separate USCF scholastic rating system and not in the “regular” rating system? Then your son might still lose a few scholastic rating points with his performance, but he probably wouldn’t care about that rating any more, because he’s moved on to adult competition and has a “regular” rating. The result is that he would still be willing to play with the kids he goes to school with for the fun of it.
I keep hearing that USCF wants to keep its strong players active. In the scholastic world, I think it would help keep them active if once they had advanced beyond the usual scholastic level, they could play in those events and not worry so much about the rating problems of playing down 500 or more points.
I disagree with the whole premise that he should feel bad about losing points. Think of it this way: what is the percentage of (a) USCF members, (b) scholastic USCF members, (c) our general population… that can truly earn a 1700+ rating?? Exceedingly small, and obviously a>>b>>c. So he should be proud of his rating, without a doubt. But it carries with it the burden of maintenance - indeed, he SHOULD be expected to beat all 1400- players, regardless of age, a high percentage of the time. And he should lose to 2000+ players at a comparable frequency - except for the occasional win or draw, which he will cherish. It is similar for ALL players, once they are rated. We have a couple gentlemen in our club that are 2200+, and when they play a “mere” 1700 player, they have almost nothing to gain, rating-wise, and much more to lose. But they accept that mantle as part of being 2200+, as these very good players should accept as part of being 1700+. Indeed, among his scholastic peers, your son is a ‘master’ of sorts - and must work to defend/maintain that position.
A more common example is readily at hand. I am a ‘C’ player, near 1500. Say I play somebody who is rated 900. Obviously a less experienced player, and yet, someone who knows the rules, a little strategy, and might very well be highly motivated to give me a battle. Should I beat them, almost every time? Certainly… YES. And when I win, what do I get? On the flip side, should I lose, how much should I be docked? Let’s take a look, via the ‘estimator’. In that situation:
WIN (should happen) = 1502. I gain only 2 points for that game.
DRAW (could happen!) = 1482. I drop 18 points for my efforts.
LOSE (shudder) = 1464. I drop a full 36 points!!
Recently I was playing such a player and on the combination of his inspired play and my own gaffes and time struggles, this almost happened. I did pull it out, and he relished the near-miss… with my admiration. (I secretly also relished the near-miss of 30+ point free-fall.)
My point is this: the above scenario, and its seemingly disparate outcomes, is APPROPRIATE. I have to earn my spot in the 1500+ club (or any other level), and if I want to get to 1700+ any time soon (ans.: I DO), that level, too, has to come with a price - if it is to mean anything at all. I CANNOT get to 1700+ only by playing players ranked below or around me - it won’t happen. I have to BOTH (a) step up an beat some higher-ranked players, and (b) hold my own, very consistently defending my ranking, against people that are ranked significantly below me. Without both components, I won’t be able to succeed - and, I believe, I SHOULDN’T be able to succeed.
Giving your son’s opponents artificially-inflated ratings, therefore, would eventually undermine the value of his OWN rating, and his struggle toward (and ultimate value of) his reaching 1800+, 2000+… etc. I want to make those levels, too - and being older and out of chess for a while, it will be difficult - but if/when I get there, I know that it will truly MEAN something - because of our consistently-applied rating scale and process.
Regardless of what people (such as a 1700 rated kid considering whether to play in a weak scholastic event) maybe “should” feel, it’s better marketing to respect what they actually feel. And USCF has a mission to serve them, whether they feel one way or the other about it.
Frankly, I think he’s being very nice to play in such an event, even if he didn’t have to worry about rating. If he’s really 1700 strength, it’s not good for his chess to play a 1000 rated field. It will mainly give him bad habits, besides boring him to death. He is a promising player and should be trying to play A’s and Experts.
He can choose to do that, sure. He likely won’t find it in the scholastic events, but probably can do so with out much trouble within a drivable (for his parents) radius. Kudos if he gives it a shot!
If his 1700+ is accurate, he’ll take his lumps against As and Experts. If he’s underrated, he’ll surprise a couple of ‘A’ players. In any of these alternatives, however, the rating system stands to serve him and does so best if it is NOT perturbed artificially.
As a player who played in a lot of scholastic tournaments growing up, I just knew that it was very possible to lose rating points and it’s better to just not allow it to get to you. Kids are usually improving rapidly, alert tactically, hard fighters, and brave players. Upsets are going to happen, and that’s part of what lures players to play in those tournaments. The same kind of thing happens regularly at the US Open when up and coming players knock out GMs. Rating is a nice indicator of success, but it’s not the only indicator and shouldn’t be something that holds you back from wanting to compete.
Many fine points have been mentioned, but since the discussion is interesting I’d like to press on.
Over time, ratings would go up and down and certainly 86% would not stay stuck at 1000 as they play each other, their ratings would still fluctuate relative to each other.
If by “massive inflation” you mean that ratings would be higher than they should be - so what? The ratings may be higher, but the disparity should be less. I’m not ignoring the math expressed here that my idea will not avoid the same 8 pt loss. I see that. But by “two rating systems” I mean this:
I caution my son when he plays scholastic tournaments because a 900 player who ONLY plays scholastic tournaments may actually be a 1400 player if he/she were to play in adult tournaments. The ratings of exclusively scholastic players fluctuate within a different range. When a scholastic player moves out of that group and plays adults, they often catch an adult hundreds of points higher - but skill-wise, not that much better. And their ratings get adjusted on both sides. It makes one wonder, what is a true 1700?? What is a true 900??
I see a lot of arguments against a 1000 floor and that is fine. But perhaps many don’t see how it is possible that a pool of 900 rated players may actually be C class players, but because they stay in their own pool their ratings do not reflect their strength?
C’mon, will someone admit they see this happening as well? I can’t be the only one!
Why not just add 900 to everyone’s current rating and increase all the floors by 900 as well? That way nobody would be rated under 1000. And I’d make 2300.
Tell your son that happened because he played one game to a draw. Tell him he needs better competition because that’s where the payoff is. There is probably not a significant statistical difference between his rating before and rating after the event. Rating points are not the be-all and end-all. They are a means to measure and assure fair competition.
Another point is that the scholastic ratings are underrated as a group. If I wanted to start with a rating above 1000, my first tournament wouldn’t be a scholastic tournament, because the current rating system computes provisional ratings relative to the strength of the group one is playing against. Hence, I may lose every game in both a scholastic tournament (avg rating 500) and an adult tournament (avg rating 1500), but in the former I start at 100, whereas in the latter I start at 1100. Check it out with the rating estimator and see! How does that make sense? By starting everyone at the same rating you eliminate this stupid dichotomy. Poor Johnny thinks in the former he’s a 100 and is afraid to play adults, but in the latter he’s (ta da!) 1100, and is too good to play other kids!
Because of the lack of a base starting rating you have created this ridiculous situation and it’s time to fix it.
These threads usually end up with proposals that fix the problems that are statistical outliers in ways that cause new problems for the overwhelming majority of the game results.
I like it! I have been vying for Grandmaster status for some time now, but I find those last 900 or so points to be a bugaboo. This will solve my problem. Thanks!
In general, I haven’t seen that in our area. It is certainly true that scholastic players are more likely to improve from their current strength so that it can take a while for their ratings to catch up to that new strength - but that is different than claiming the scholastic ratings are low in general. If anything, with the artificial floor of 100, it could certainly be argued that they are artificially high at the very low end compared to adults in the 900-1400 range (where there is no floor).
That isn’t how the system works. As long as you start out losing all your games, you’ll actually be about 400 points below the lowest person you played, not below the average.
Of course, there is also the opposite side which I tend to see more often than your worry - a scholastic player going undefeated in his first tournament and getting an unrealistically high rating. Much of our difference in perspective may have to do with the types of scholastic tournaments held in our respective areas. Rather than trying to change the rating system without a valid statistical reason, you might be better off trying to change the way local scholastic are organized in your area. Instead of basing sections on grades, it might be better to use rating classes - say K-12 Open, K-12 U1200, K-12 U1000. …
Could you explain why you think that your initial suggestion (“My solution is to adjust all current ratings below 1000 to 1000, and make it the Universal floor.”) wouldn’t just add 900 points to everyone’s rating (since the current universal floor is 100) and not really change anything relatively?
This doesn’t follow from experience. Back when we had an absolute floor at 100 (as opposed to the current “floating floor” from 100-150), we had a massive pile-up of players at the 100 point floor. I believe that the argument to change to a “floating floor” was to keep those players from being discouraged. I don’t have any idea if that has been the effect.