Yes, ratings only reflect your strength within a pool of players. If the pool is two people your rating will only reflect how you do vs that one player.
This is how a convict manipulated the system to get to 2600 USCF. He would get other inmates to register for the USCF lose to them intentionally during their provisional period so they had inflated ratings and then beat them over and over until they gave no more points.
Originally posted by plopzillaIf two 2600 players played each other constantly with a 1300 rating and one of them won 100% of 100 games, then the winner would eventually hit a peak rating which he couldn't rise above. If you look at the graph of one of the top rated players on this site and look for a win against a 1300 player, you'll notice that their rating doesn't increase by a single point even though they won. This is built into the elo system of grading. A players rating eventually reaches a point where the win expectancy is 100%. At that point a player doesn't receive a rating increase. Different rating systems set this expectancy at different points, but i believe on this site a 600 point advantage means you won't receive a rating increase for a win. Going by this system our 2600 player would probably struggle to reach 1800, as his opponents grade would be dropping significantly. In fact, i'd say the maximum he could probably achieve would be 1600 (+300) while his opponent would be graded 1100 (-300). As there would be a 600 point difference between them, no further increase/decrease would be possible without a change in result...
so are we saying if you mostly only play people roughly around your strength, unless you pretty much beat them constantly your rating will never really progress that much..
Originally posted by MarinkatombFor the K constant of 24 (about 2100 to 2400 here), you're close - Any rating differential of 597 or greater won't net you any rating increase. But at the lower ratings where the K constant is 32, any rating differential of 720 or more won't get you a rating increase. (Verified with my handy-dandy RHP rating spreadsheet. 🙂 )
If two 2600 players played each other constantly with a 1300 rating and one of them won 100% of 100 games, then the winner would eventually hit a peak rating which he couldn't rise above. If you look at the graph of one of the top rated players on this site and look for a win against a 1300 player, you'll notice that their rating doesn't increase by a si between them, no further increase/decrease would be possible without a change in result...
The formula for the Delta R at which you cease getting rating points is:
Delta R = 400 * log ((K/(K - 0.5)) -1)
The 0.5 constant in the equation is the rounding down of 1/2 rating point. (Any rating increase less than 0.5 is rounded down to zero, and you get bupkis.)