Is it possible to know an average variability of score on chess.com?

Question

If I play a number of 10 minute games on an "off day", on chess.com, my score can plummet by 100 points. Conversely, if I'm well-slept, just had some good news (no stress), fresh breakfast and coffee, I can win game after game, raising my score by 100+

I used to think a person's score was a fair indicator of skill level and therefore what I was "up against", but now I'm thinking not so much. I've played high score players making repeated blunders, and low score players whipping me.

Is there any data on the average variability or score bracket that players have?

For example, if I play a person with score 900, are they more likely to be bracketed 875-925, or maybe 850-950, or even 800-1000 ?

Answer 1

chess.com uses the Glicko rating system. The rating has an associated RD (rating deviation). If RD is low, the rating is more reliable or accurate.

Consider the rating info of Magnus at present.

{'chess_blitz': {'best': {'date': 1607707714,
                          'game': 'https://www.chess.com/game/live/5942875338',
                          'rating': 3192},
                 'last': {'date': 1607716342, 'rating': 3123, 'rd': 53},
                 'record': {'draw': 52, 'loss': 32, 'win': 137}},
 'chess_bullet': {'best': {'date': 1604349635,
                           'game': 'https://www.chess.com/game/live/5878212456',
                           'rating': 3350},
                  'last': {'date': 1607718902, 'rating': 3260, 'rd': 68},
                  'record': {'draw': 11, 'loss': 19, 'win': 59}},
 'chess_rapid': {'best': {'date': 1418582233,
                          'game': 'https://www.chess.com/game/live/7498870151',
                          'rating': 2862},
                 'last': {'date': 1613844867, 'rating': 2810, 'rd': 241},
                 'record': {'draw': 1, 'loss': 0, 'win': 5}},
 'fide': 2882,
 'lessons': {},
 'puzzle_rush': {},
 'tactics': {}}

He has a blitz rating info of:

'last': {'date': 1607716342, 'rating': 3123, 'rd': 53},

The rd is 53. The rating interval is:

min_rating = 3123 - 53*2 = 3017
max_rating = 3123 + 53*2 = 3229

So in a bad day he might play at lower end of the rating which is 3017 and in a good day he might play at a higher end of the rating which is 3229.

Different players may have different RD. If the number of games played by a player is low, the RD will be high and the rating range from max_rating - min_rating will be high too. If this range is high the accuracy of the rating is lower.

Answer 2

Rating does of course say something about your strength, but I once saw someone of about 1375 absolutely wiping out 1800’s. Rating says something, but other factors such as underestimation, luck, and your mood also play a role.