What if chess ratings for a tournament were based on relative performance ratings?

Random shower thought.

One of the biggest issues with ratings is them not keeping up with an improving player or not reflective of form. What if in a tournament, one's rating change was determined by first evaluating each player's performance rating, and then performing rating adjustments based upon the performance rating instead of the real rating?

I have no idea the mathematical effect this would have. The idea is to have a rating adjustment based on players 'short term rating' rather than their 'long term rating.' Perhaps this would do little other than slow rating changes since if you have a bad tournament then your declines would be less because your 'short term rating' would be relatively low, whereas if you had a great one your gains would also be less because your 'short term rating' would be relatively high.

Random shower thought. One of the biggest issues with ratings is them not keeping up with an improving player or not reflective of form. What if in a tournament, one's rating change was determined by first evaluating each player's performance rating, and then performing rating adjustments based upon the performance rating instead of the real rating? I have no idea the mathematical effect this would have. The idea is to have a rating adjustment based on players 'short term rating' rather than their 'long term rating.' Perhaps this would do little other than slow rating changes since if you have a bad tournament then your declines would be less because your 'short term rating' would be relatively low, whereas if you had a great one your gains would also be less because your 'short term rating' would be relatively high.

lizani

The TPR can be strongly influenced by a single game. For example, a much higher rated player tries to beat his opponent, overreaches and loses. For the rest of the tournament, both players may play according to their rating expectation but this one result will determine their overall TPR.

tpr

"One of the biggest issues with ratings is them not keeping up with an improving player"

For elo there is K = 40 for new or young players, K = 20 for most, and K = 10 above 2400.
Glicko-2 on lichess calculates an individual K for each player at each time.

"not reflective of form"

Form may change from day to day. Say you play well, then get sick, then get better again.

"determined by first evaluating each player's performance rating"

On how many games? 1? 2? 3?

"the mathematical effect"

It makes no sense mathematically. The elo or glicko-2 are simplified from the Kalman filter.

"One of the biggest issues with ratings is them not keeping up with an improving player" * For elo there is K = 40 for new or young players, K = 20 for most, and K = 10 above 2400. Glicko-2 on lichess calculates an individual K for each player at each time. "not reflective of form" * Form may change from day to day. Say you play well, then get sick, then get better again. "determined by first evaluating each player's performance rating" * On how many games? 1? 2? 3? "the mathematical effect" * It makes no sense mathematically. The elo or glicko-2 are simplified from the Kalman filter.