Site Ideas
13 Jan 09
20 Jan 09
Originally posted by Ice ColdYou're correct that no system, including rating floors, is foolproof.
Good idea, and rec'd on my part.
The only problem I see is a guy could throw his 30th game, and retain a lower floor.
You know the deal, rule breakers tend to use and abuse the rules and laws to their advantage etc etc.
I mean, nothing even forces them to attain a high rating in the first place. Just throw enough games in the provisional stage, and keep your rating down in the 1400s. Then you can play at your full strength for games that 'count', and 'blunder' away the ones that don't.
23 Jan 09
Ice and SG, you are both correct, nothing could stop all the douches who are guilty of this garbage from keeping themselves on a lower rating floor. However, these are the types of people who love nothing more than to have stats on a web page stroke their egos. I'm in a 1350-1375 banded tourny right now with some fruit loop named jayo who used to be rated mid 1700s, its aggravating. The thing about these types of peoples is that I highly doubt they could live with a low rating, or one that doesn't truly reflect them. There will be a few who don't care, but the vast majority of these pricks wouldn't hear of being rated 1400 when they're supposed to be 1800.
- Joined
- 07 Jun 05
- Moves
- 5301
23 Jan 09
Originally posted by SwissGambitTo refer back to the old thread, see here: Thread 93130
For the benefit of the viewing audience, my response to the three points:
- Inflation from floors is negligible; and besides, the main point of ratings is to compare relative playing strength.
- Admin burden is negligible; there is no need to even consider floor-removal requests unless the player has been stuck on their floor for some time [an ...[text shortened]... Floors allow them to immediately join tournaments against players close to their own strength.
So my reasons for raising the three main issues again:
- Rating inflation will be an effect of adding floors. This is no longer in question. I contend that the extent of rating inflation is unknown at present, and therefore hardly negligible. If inflation is stoked by adding a floor, relative strengths from one year to the next become harder to judge.
-admin burden : Even deleting a request takes some time, and the number of requests will not be seen until the system is implemented. Add in the time to check whether rating has been low long enough, and admin burden becomes significant.
-TER works just fine. It prevents people resigning a load of games just to get their rating low enough to enter a low tournament band. Sure, TER does fall, but the tournament in question will have closed after 100 days, and most of us are here to play, not wait for ratings to fall. Adding a permanent TER floor of 200 below the max rating would be an addition, but could solve the "problem" of TER falling over time. I don't think it would be worth it though.
Response to your edit: Would punishment be that bad? However, your comment is plain wrong - tournament entry rating indicates the band in which you can enter a tourney - it does not exclude you until your rating is equal to your TER.
To go into a little more depth about why I don't want to see rating floors:
As User 380105 mentioned above, ability depends on handling of time. If you drop a bunch of games for whatever reason, then your rating ought to fall, until your rating reflects your performance. Going away and coming back to the site do not change your ability, but someone who leaves and then comes back is likely to do so again, so their rating should be lower than that of someone of the same ability who plays without leaving. Rating reflects performance, not ability.
V-graphs give an indication to those who care to look that the player sometimes takes breaks. It is a lot harder to spot those if you prevent rating falling.
Rating falling acts, in effect, as a punishment - I think that it is harder to get games at your ability again until your rating gets back in line. This acts as a disincentive to mass resignations, as you have to put time into games in order to get your rating back where you think it should be.
One final one, which I had not thought of until recently. Everyone starts at 1200. How do you cope with a player who has not yet found their floor after 30 games? More work for the admins? But why would a player bumping along a floor of say 900 want his rating lower - there is little for him to gain - Tournament bands generally stop at 1100, and a lower rating looks worse when looking for games.
Simply put, in my opinion, adding a rating floor causes a bunch of problems, and is not worth the benefits.
23 Jan 09
1 edit
Originally posted by gezza1) Admin burden
To refer back to the old thread, see here: Thread 93130
So my reasons for raising the three main issues again:
- Rating inflation will be an effect of adding floors. This is no longer in question. I contend that the extent of rating inflation is unknown at present, and therefore hardly negligible. If inflation is stoked by adding a floor my opinion, adding a rating floor causes a bunch of problems, and is not worth the benefits.
The time to check how long a player has been on their floor should be insignificant. One look at the rating's graph would instantly tell them. A dashed line could even be added to the graph to show where the floor is. As for dumb questions, there are always going to be people who do not know the rules, or read FAQs. The rest of us should not be made to miss out on a good feature due to the misguided questions of a few ignorant people.
2) TER
TER does not even cover games played outside tournaments. Within tourneys, it does not work well enough - there are loopholes, as already pointed out. So, not only does it not 'work fine', but it doesn't even impact tons of rated games played here [clans, open invites, sieges, etc.]
3) response to edit
Correction: TER punishes everyone else when a 1200 player with a high TER gets to play in a 1800+ section.
4) "why I don't want to see rating floors"
You would fully rate all games 'lost' by timeout while on work-related travel with no access to a computer. This is clearly absurd. Some of us prefer our chess ratings to actually represent chess skill/performance, not life prioritization skill/performance.
27 Jan 09
Originally posted by SwissGambitAgreed, and very well said!
It's high time RHP adopted rating floors similar to what the US Chess Federation uses.
The idea is this: After X games at a certain rating class, a player's rating cannot fall 200 points below that class.
Example: Player A's rating remains over 1900 for 30 games. This establishes a rating floor of 1700. His rating can no longer fall below 17 ...[text shortened]... that player has been stuck on their rating floor for a long time [and not due to absence!].
Rec'd!
10 Mar 09
1 edit
Originally posted by PhlabibitThe floor should be set based on a certain number of finished games. If we set it based on days, players could get floors just by remaining idle at a certain rating.
Could you spare a bit more support?
I support this idea A LOT!
The trick is a good range and formula. What's the best one offered so far? 200 points based on 100 days???
P-
Originally posted by SwissGambit
Here's the logic I would use if programming rating floors:
After each rated game is finished, and rated:
1) Grab the player's last 30 ratings (the most recent 30 values from their rating graph), excluding provisionally rated games. If they have not yet played 30 non-provisional games, do nothing [skip remaining steps].
2) Find the lowest rating in the group; round it down to the nearest hundred.
3) Subtract 200 from that value.
4) If result is lower than the current rating floor, do nothing [skip remaining step].
5) If result is higher than the current rating floor, set it as the new rating floor.
The "30" is open to adjustment. Experience should show if needs to go higher, or lower.
10 Mar 09
1 edit
Indeed the floor should be permanent.
To have the highest rating obtained over 10 games minus 200 seems a good formula to me.
And it should be easy to program. Existing players floor could be obtained by a batchprogramme going through all users and finding the appropriate value.
In the running system a small macro should do the trick of tuning the rating floor.
Edit: Didn't saw Swiss Gambit's post. Is much easier the way he described it.