Unofficial OGS rank histogram (and graphs) 2022

A year has passed since the last time. I don’t have time to test these, let’s hope everything is ok.

Traditional six-month player distribution by rank:

Percentiles at the bottom are based on green part as all weaker players + half of given rank.

Accurate percentiles (change since the last time)
  0: 59.0k (+9.4)
  1: 29.0k (+0.6)
  2: 25.8k (+1.1)
  3: 24.0k (+1.1)
  4: 22.5k (+1.5)
  5: 21.4k (+1.5)
  6: 20.5k (+1.6)
  7: 19.8k (+1.5)
  8: 19.2k (+1.5)
  9: 18.6k (+1.4)
 10: 18.0k (+1.4)
 11: 17.5k (+1.3)
 12: 17.0k (+1.3)
 13: 16.6k (+1.2)
 14: 16.1k (+1.2)
 15: 15.8k (+1.1)
 16: 15.4k (+1.0)
 17: 15.0k (+1.0)
 18: 14.7k (+0.9)
 19: 14.4k (+0.8)
 20: 14.1k (+0.8)
 21: 13.8k (+0.7)
 22: 13.5k (+0.7)
 23: 13.2k (+0.7)
 24: 13.0k (+0.7)
 25: 12.8k (+0.6)
 26: 12.5k (+0.6)
 27: 12.2k (+0.7)
 28: 12.0k (+0.6)
 29: 11.8k (+0.6)
 30: 11.6k (+0.5)
 31: 11.3k (+0.6)
 32: 11.1k (+0.6)
 33: 10.9k (+0.6)
 34: 10.7k (+0.6)
 35: 10.5k (+0.6)
 36: 10.3k (+0.6)
 37: 10.1k (+0.6)
 38:  9.9k (+0.6)
 39:  9.7k (+0.6)
 40:  9.6k (+0.5)
 41:  9.4k (+0.5)
 42:  9.2k (+0.5)
 43:  9.0k (+0.5)
 44:  8.8k (+0.5)
 45:  8.7k (+0.4)
 46:  8.5k (+0.5)
 47:  8.3k (+0.5)
 48:  8.1k (+0.5)
 49:  8.0k (+0.4)
 50:  7.8k (+0.4)
 51:  7.7k (+0.3)
 52:  7.5k (+0.4)
 53:  7.3k (+0.4)
 54:  7.2k (+0.3)
 55:  7.0k (+0.3)
 56:  6.8k (+0.4)
 57:  6.6k (+0.4)
 58:  6.5k (+0.3)
 59:  6.3k (+0.3)
 60:  6.2k (+0.3)
 61:  6.0k (+0.3)
 62:  5.8k (+0.3)
 63:  5.7k (+0.2)
 64:  5.5k (+0.3)
 65:  5.3k (+0.3)
 66:  5.2k (+0.2)
 67:  5.0k (+0.2)
 68:  4.9k (+0.2)
 69:  4.7k (+0.2)
 70:  4.5k (+0.2)
 71:  4.4k (+0.1)
 72:  4.2k (+0.2)
 73:  4.0k (+0.2)
 74:  3.8k (+0.2)
 75:  3.7k (+0.1)
 76:  3.5k (+0.1)
 77:  3.3k (+0.1)
 78:  3.1k (+0.2)
 79:  2.9k (+0.2)
 80:  2.7k (+0.2)
 81:  2.5k (+0.2)
 82:  2.3k (+0.1)
 83:  2.1k (+0.1)
 84:  1.9k (+0.1)
 85:  1.7k (+0.1)
 86:  1.4k (+0.1)
 87:  1.2k (+0.1)
 88:  1.0k (+0.1)
 89:  0.8k (+0.0)
 90:  0.5k (+0.1)
 91:  0.2k (+0.1)
 92:  1.1d (+0.0)
 93:  1.4d (+0.0)
 94:  1.8d (+0.0)
 95:  2.2d (+0.0)
 96:  2.6d (+0.0)
 97:  3.1d (-0.1)
 98:  3.8d (-0.1)
 99:  4.7d (-0.2)
100: 15.0d (+0.0)

Onion histogram

Percentiles
       All   5 y   1 y   6 m   1 m   1 w   1 d
  0: 68.4k 68.4k 59.0k 59.0k 52.9k 52.9k 39.6k
  1: 30.3k 31.2k 32.2k 31.0k 27.4k 25.7k 23.6k
  2: 27.4k 28.4k 28.9k 27.3k 24.3k 22.6k 20.7k
  3: 25.7k 26.7k 26.9k 25.4k 22.5k 20.8k 19.0k
  4: 24.5k 25.4k 25.4k 24.0k 21.2k 19.4k 17.8k
  5: 23.5k 24.5k 24.3k 22.9k 20.1k 18.6k 16.7k
  6: 22.6k 23.6k 23.3k 21.9k 19.3k 17.7k 15.9k
  7: 21.8k 22.8k 22.5k 21.1k 18.6k 17.0k 15.4k
  8: 21.2k 22.1k 21.6k 20.4k 17.9k 16.4k 14.9k
  9: 20.6k 21.6k 21.0k 19.7k 17.3k 15.9k 14.4k
 10: 20.1k 21.0k 20.4k 19.2k 16.8k 15.4k 13.9k
 11: 19.6k 20.5k 19.8k 18.7k 16.3k 15.0k 13.5k
 12: 19.1k 20.1k 19.4k 18.2k 15.8k 14.6k 13.1k
 13: 18.7k 19.6k 18.9k 17.8k 15.4k 14.2k 12.8k
 14: 18.3k 19.2k 18.4k 17.3k 15.0k 13.9k 12.4k
 15: 18.0k 18.8k 18.0k 16.9k 14.7k 13.5k 12.1k
 16: 17.6k 18.5k 17.6k 16.5k 14.3k 13.2k 11.9k
 17: 17.2k 18.1k 17.2k 16.1k 14.0k 12.9k 11.6k
 18: 16.9k 17.8k 16.9k 15.8k 13.7k 12.6k 11.4k
 19: 16.6k 17.4k 16.5k 15.4k 13.4k 12.3k 11.2k
 20: 16.3k 17.1k 16.2k 15.1k 13.1k 12.0k 11.0k
 21: 16.0k 16.8k 15.8k 14.8k 12.8k 11.8k 10.8k
 22: 15.8k 16.5k 15.5k 14.5k 12.6k 11.5k 10.5k
 23: 15.5k 16.2k 15.2k 14.2k 12.3k 11.3k 10.3k
 24: 15.2k 15.9k 14.9k 13.9k 12.1k 11.1k 10.1k
 25: 15.0k 15.6k 14.7k 13.7k 11.8k 10.9k  9.9k
 26: 14.7k 15.4k 14.4k 13.4k 11.6k 10.7k  9.8k
 27: 14.5k 15.1k 14.1k 13.2k 11.4k 10.5k  9.6k
 28: 14.3k 14.8k 13.8k 12.9k 11.2k 10.3k  9.4k
 29: 14.0k 14.6k 13.6k 12.7k 11.0k 10.0k  9.2k
 30: 13.8k 14.3k 13.3k 12.4k 10.8k  9.9k  9.1k
 31: 13.6k 14.1k 13.1k 12.2k 10.6k  9.7k  8.9k
 32: 13.4k 13.9k 12.9k 11.9k 10.4k  9.5k  8.7k
 33: 13.2k 13.6k 12.6k 11.7k 10.2k  9.4k  8.6k
 34: 13.0k 13.4k 12.4k 11.5k 10.0k  9.2k  8.4k
 35: 12.8k 13.2k 12.1k 11.3k  9.8k  9.0k  8.3k
 36: 12.6k 13.0k 11.9k 11.1k  9.6k  8.9k  8.2k
 37: 12.4k 12.7k 11.7k 10.8k  9.4k  8.7k  8.0k
 38: 12.2k 12.5k 11.5k 10.6k  9.2k  8.5k  7.9k
 39: 12.0k 12.3k 11.2k 10.4k  9.0k  8.4k  7.7k
 40: 11.8k 12.1k 11.0k 10.2k  8.9k  8.2k  7.6k
 41: 11.6k 11.9k 10.8k 10.0k  8.7k  8.1k  7.4k
 42: 11.4k 11.7k 10.6k  9.8k  8.5k  7.9k  7.3k
 43: 11.2k 11.5k 10.4k  9.7k  8.4k  7.8k  7.2k
 44: 11.0k 11.2k 10.2k  9.5k  8.2k  7.7k  7.0k
 45: 10.8k 11.0k 10.0k  9.3k  8.0k  7.5k  6.9k
 46: 10.6k 10.8k  9.8k  9.1k  7.9k  7.3k  6.8k
 47: 10.4k 10.6k  9.6k  8.9k  7.7k  7.2k  6.6k
 48: 10.2k 10.4k  9.4k  8.7k  7.6k  7.1k  6.5k
 49: 10.0k 10.2k  9.2k  8.5k  7.4k  6.9k  6.4k
 50:  9.8k  9.9k  9.0k  8.3k  7.3k  6.8k  6.3k
 51:  9.6k  9.7k  8.8k  8.2k  7.1k  6.6k  6.1k
 52:  9.4k  9.5k  8.6k  8.0k  6.9k  6.5k  6.0k
 53:  9.2k  9.3k  8.4k  7.8k  6.8k  6.4k  5.9k
 54:  9.0k  9.1k  8.2k  7.7k  6.6k  6.2k  5.7k
 55:  8.8k  8.9k  8.0k  7.5k  6.5k  6.1k  5.6k
 56:  8.6k  8.7k  7.8k  7.3k  6.3k  5.9k  5.5k
 57:  8.4k  8.5k  7.6k  7.1k  6.2k  5.8k  5.4k
 58:  8.2k  8.2k  7.5k  6.9k  6.0k  5.7k  5.2k
 59:  8.0k  8.0k  7.3k  6.7k  5.9k  5.5k  5.1k
 60:  7.8k  7.8k  7.1k  6.6k  5.7k  5.4k  5.0k
 61:  7.6k  7.6k  6.9k  6.4k  5.6k  5.3k  4.9k
 62:  7.4k  7.4k  6.7k  6.2k  5.4k  5.1k  4.7k
 63:  7.2k  7.2k  6.5k  6.0k  5.3k  5.0k  4.6k
 64:  7.0k  7.0k  6.3k  5.9k  5.1k  4.9k  4.5k
 65:  6.8k  6.7k  6.1k  5.7k  5.0k  4.7k  4.3k
 66:  6.5k  6.5k  5.9k  5.5k  4.9k  4.6k  4.2k
 67:  6.3k  6.3k  5.7k  5.3k  4.7k  4.4k  4.0k
 68:  6.1k  6.1k  5.5k  5.2k  4.6k  4.3k  3.9k
 69:  5.9k  5.9k  5.3k  5.0k  4.4k  4.1k  3.8k
 70:  5.7k  5.7k  5.1k  4.8k  4.2k  4.0k  3.6k
 71:  5.5k  5.4k  4.9k  4.6k  4.1k  3.8k  3.5k
 72:  5.3k  5.2k  4.7k  4.4k  3.9k  3.7k  3.4k
 73:  5.0k  5.0k  4.5k  4.2k  3.8k  3.6k  3.2k
 74:  4.8k  4.8k  4.3k  4.0k  3.6k  3.4k  3.1k
 75:  4.6k  4.5k  4.1k  3.9k  3.4k  3.2k  2.9k
 76:  4.4k  4.3k  3.9k  3.7k  3.3k  3.1k  2.8k
 77:  4.1k  4.1k  3.7k  3.5k  3.1k  2.9k  2.6k
 78:  3.9k  3.8k  3.5k  3.3k  2.9k  2.8k  2.5k
 79:  3.7k  3.6k  3.2k  3.1k  2.7k  2.6k  2.3k
 80:  3.4k  3.4k  3.0k  2.9k  2.6k  2.4k  2.2k
 81:  3.2k  3.1k  2.8k  2.7k  2.4k  2.3k  2.1k
 82:  2.9k  2.8k  2.6k  2.4k  2.2k  2.1k  1.9k
 83:  2.7k  2.6k  2.4k  2.2k  2.0k  1.9k  1.7k
 84:  2.4k  2.3k  2.1k  2.0k  1.8k  1.7k  1.6k
 85:  2.1k  2.1k  1.9k  1.7k  1.6k  1.5k  1.4k
 86:  1.8k  1.8k  1.6k  1.5k  1.4k  1.3k  1.2k
 87:  1.6k  1.5k  1.4k  1.3k  1.2k  1.1k  1.1k
 88:  1.3k  1.2k  1.1k  1.0k  1.0k  1.0k  0.9k
 89:  1.0k  0.9k  0.9k  0.8k  0.8k  0.8k  0.7k
 90:  0.7k  0.6k  0.6k  0.5k  0.5k  0.5k  0.5k
 91:  0.4k  0.3k  0.3k  0.2k  0.2k  0.2k  0.2k
 92:  0.0k  1.0d  1.0d  1.1d  1.1d  1.0d  1.0d
 93:  1.3d  1.3d  1.3d  1.3d  1.4d  1.3d  1.2d
 94:  1.7d  1.7d  1.6d  1.7d  1.7d  1.6d  1.5d
 95:  2.1d  2.1d  2.0d  2.1d  2.1d  2.0d  1.8d
 96:  2.5d  2.5d  2.4d  2.4d  2.5d  2.4d  2.3d
 97:  3.1d  3.0d  3.0d  3.0d  3.0d  2.9d  2.7d
 98:  3.8d  3.7d  3.6d  3.6d  3.6d  3.5d  3.2d
 99:  4.9d  4.7d  4.6d  4.6d  4.7d  4.5d  4.2d
100: 15.0d 15.0d 15.0d 15.0d 12.6d 12.6d  9.6d

Veteran histogram

Percentiles (difference compared to general population)
     1000+         5000+
  0: 26.4k (+32.6) 23.1k (+35.9)
  1: 16.0k (+13.0) 13.1k (+15.9)
  2: 14.7k (+11.1) 12.1k (+13.7)
  3: 13.7k (+10.3) 11.5k (+12.5)
  4: 13.0k  (+9.5) 11.0k (+11.5)
  5: 12.5k  (+8.9) 10.6k (+10.8)
  6: 12.1k  (+8.4) 10.1k (+10.4)
  7: 11.8k  (+8.0) 10.0k  (+9.8)
  8: 11.5k  (+7.7)  9.8k  (+9.4)
  9: 11.3k  (+7.3)  9.7k  (+8.9)
 10: 11.0k  (+7.0)  9.3k  (+8.7)
 11: 10.8k  (+6.7)  9.2k  (+8.3)
 12: 10.5k  (+6.5)  9.0k  (+8.0)
 13: 10.3k  (+6.3)  8.9k  (+7.7)
 14: 10.1k  (+6.0)  8.8k  (+7.3)
 15:  9.9k  (+5.9)  8.6k  (+7.2)
 16:  9.7k  (+5.7)  8.5k  (+6.9)
 17:  9.6k  (+5.4)  8.4k  (+6.6)
 18:  9.4k  (+5.3)  8.3k  (+6.4)
 19:  9.2k  (+5.2)  8.2k  (+6.2)
 20:  9.1k  (+5.0)  8.1k  (+6.0)
 21:  8.9k  (+4.9)  8.0k  (+5.8)
 22:  8.8k  (+4.7)  7.8k  (+5.7)
 23:  8.7k  (+4.5)  7.6k  (+5.6)
 24:  8.5k  (+4.5)  7.5k  (+5.5)
 25:  8.4k  (+4.4)  7.4k  (+5.4)
 26:  8.3k  (+4.2)  7.3k  (+5.2)
 27:  8.2k  (+4.0)  7.2k  (+5.0)
 28:  8.0k  (+4.0)  7.1k  (+4.9)
 29:  7.9k  (+3.9)  7.0k  (+4.8)
 30:  7.8k  (+3.8)  6.9k  (+4.7)
 31:  7.7k  (+3.6)  6.8k  (+4.5)
 32:  7.6k  (+3.5)  6.7k  (+4.4)
 33:  7.5k  (+3.4)  6.6k  (+4.3)
 34:  7.4k  (+3.3)  6.5k  (+4.2)
 35:  7.3k  (+3.2)  6.4k  (+4.1)
 36:  7.2k  (+3.1)  6.3k  (+4.0)
 37:  7.0k  (+3.1)  6.2k  (+3.9)
 38:  6.9k  (+3.0)  6.1k  (+3.8)
 39:  6.8k  (+2.9)  6.0k  (+3.7)
 40:  6.7k  (+2.9)  6.0k  (+3.6)
 41:  6.6k  (+2.8)  5.9k  (+3.5)
 42:  6.5k  (+2.7)  5.9k  (+3.3)
 43:  6.4k  (+2.6)  5.7k  (+3.3)
 44:  6.3k  (+2.5)  5.7k  (+3.1)
 45:  6.2k  (+2.5)  5.6k  (+3.1)
 46:  6.1k  (+2.4)  5.5k  (+3.0)
 47:  6.0k  (+2.3)  5.4k  (+2.9)
 48:  5.9k  (+2.2)  5.3k  (+2.8)
 49:  5.8k  (+2.2)  5.2k  (+2.8)
 50:  5.7k  (+2.1)  5.2k  (+2.6)
 51:  5.6k  (+2.1)  5.1k  (+2.6)
 52:  5.5k  (+2.0)  4.9k  (+2.6)
 53:  5.4k  (+1.9)  4.9k  (+2.4)
 54:  5.3k  (+1.9)  4.9k  (+2.3)
 55:  5.2k  (+1.8)  4.8k  (+2.2)
 56:  5.1k  (+1.7)  4.8k  (+2.0)
 57:  5.0k  (+1.6)  4.6k  (+2.0)
 58:  4.9k  (+1.6)  4.6k  (+1.9)
 59:  4.8k  (+1.5)  4.5k  (+1.8)
 60:  4.7k  (+1.5)  4.5k  (+1.7)
 61:  4.6k  (+1.4)  4.3k  (+1.7)
 62:  4.5k  (+1.3)  4.2k  (+1.6)
 63:  4.3k  (+1.4)  4.2k  (+1.5)
 64:  4.3k  (+1.2)  4.1k  (+1.4)
 65:  4.2k  (+1.1)  3.9k  (+1.4)
 66:  4.0k  (+1.2)  3.8k  (+1.4)
 67:  3.9k  (+1.1)  3.7k  (+1.3)
 68:  3.8k  (+1.1)  3.6k  (+1.3)
 69:  3.7k  (+1.0)  3.5k  (+1.2)
 70:  3.6k  (+0.9)  3.3k  (+1.2)
 71:  3.4k  (+1.0)  3.3k  (+1.1)
 72:  3.3k  (+0.9)  3.2k  (+1.0)
 73:  3.2k  (+0.8)  3.0k  (+1.0)
 74:  3.1k  (+0.7)  3.0k  (+0.8)
 75:  2.9k  (+0.8)  2.9k  (+0.8)
 76:  2.8k  (+0.7)  2.8k  (+0.7)
 77:  2.7k  (+0.6)  2.7k  (+0.6)
 78:  2.6k  (+0.5)  2.7k  (+0.4)
 79:  2.5k  (+0.4)  2.5k  (+0.4)
 80:  2.4k  (+0.3)  2.4k  (+0.3)
 81:  2.2k  (+0.3)  2.3k  (+0.2)
 82:  2.1k  (+0.2)  2.2k  (+0.1)
 83:  1.9k  (+0.2)  2.0k  (+0.1)
 84:  1.7k  (+0.2)  1.8k  (+0.1)
 85:  1.6k  (+0.1)  1.7k  (+0.0)
 86:  1.4k  (+0.0)  1.6k  (-0.2)
 87:  1.2k  (+0.0)  1.4k  (-0.2)
 88:  1.1k  (-0.1)  1.1k  (-0.1)
 89:  1.0k  (-0.2)  1.0k  (-0.2)
 90:  0.8k  (-0.3)  0.9k  (-0.4)
 91:  0.6k  (-0.4)  0.7k  (-0.5)
 92:  0.3k  (-0.4)  0.5k  (-0.6)
 93:  0.0k  (-0.4)  0.3k  (-0.7)
 94:  1.3d  (-0.5)  0.1k  (-0.9)
 95:  1.6d  (-0.6)  1.2d  (-1.0)
 96:  1.9d  (-0.7)  1.4d  (-1.2)
 97:  2.3d  (-0.8)  2.0d  (-1.1)
 98:  3.0d  (-0.8)  2.3d  (-1.5)
 99:  4.0d  (-0.7)  3.2d  (-1.5)
100:  7.4d  (-7.6)  7.2d  (-7.8)

New accounts graph

Estimated weekly game ids

My life is kinda pointless. Maybe you seeing these graphs that I made will add a little meaning to it. A sort of short-lived memento that I existed.

43 Likes

Huh… 91% of our community is kyu ranked. I knew the ratio was bad but didn’t know it was that bad… Less than 2% of accounts are over 4d and that includes the bots.

7 Likes

…and existed well! These graphs are awesome :sunglasses:

8 Likes

I don’t think it’s bad. 81% of EGF ranked (=active) players are kyu. That the proportion of kyu players is higher on go servers is expected, since people who play competitions IRL are generally more serious than those who only play on the internet.

10 Likes

Sorry to hear that.
I love these charts.
Take care!

14 Likes

Makes me terribly sad to read this :people_hugging:
I hope it was just a sarcastic remark … I for one don’t believe that any one person’s life is “pointless”.

10 Likes

As long as you can contribute something positive to your own life and/or someone else’s life, I think that is enough for your life to not be pointless.
Already your post of this histogram contributes something positive to my life. I enjoy inspecting it.

11 Likes

It’s just a period in life when I wonder. What’s the point of all the effort and perseverance if I won’t be happy or create anything to be proud of. It’s like, eh, what is this.

Anyway, let’s direct the attention to “Accurate percentiles (change since the last time)”. It appears everything is in plus. Either the strength distribution stays the same, and ranks are getting weaker so people get higher number (which is doubtful). Or people got stronger, and one needs to be higher rank to stay in the same percentile. Perhaps all the people from covid days are improving while we don’t get as much fresh meat these days (so the weak tail isn’t replenished to the same level). Or maybe I messed up somewhere.

7 Likes

Have you not made many squirrels happier in your life-time (just for starters) and also people who like to see happy squirrels? :smiley:

2 Likes

I doubt that you messed up. As I have pointed out before, the current OGS population is always just a sample of the very much larger population of people that could play go. It’s only because it’s a sample that the distribution looks normal. I think the underlying distribution is some kind of power law from which you shouldn’t expect to be able to extract consistent samples.

1 Like

Stated slightly differently, perhaps new users are in general not all beginners (e.g. experienced players signing on to OGS from other servers?). It is interesting that the median ranks are climbing. Bummer, thought I was hot stuff for reaching SDK :stuck_out_tongue:

I suppose that’s readily testable (new player rank after X games, vs cohort).

1 Like

Seems I never actually posted this


30+ years of depression here, probably since age of six actually, so … make it almost 60 years, even though I wasn’t aware of it for the first 30.

The “big” happiness evades me, so I’ve switched to the small things.

Today, when I went for a long walk with the dog, I smiled at a thin kid of about ten years who wore a sad face (and who reminded me of myself at that age) … he smiled back.

I imagine that I was not the only one who walked on just a little happier.

21 Likes

Your graphs got called out on a BenKyo stream just now. International and eternal fame is yours.

6 Likes

I think OGS is in the unique position to finally answer the question of “how many games do you need to play to reach shodan?” Of course, quality of play matters. But it would be very interesting to plot two plots:

  1. Simple regression rank vs. number of games to look at the average number of games needed to reach a specific rank
  2. Violin plots (mean, median, quantiles) of games needed for each rank

Data preparation is important here, as we probably need to filter out outliers. I can’t code it for myself, but I’m happy to participate if someone is willing to do the coding. I can help with data preprocessing and statistics in both R and Python.

4 Likes

People also play many games outside OGS, and may have multiple accounts.

2 Likes

Sure, but is this a strong enough argument from dropping any research on the biggest non-Asian Go server with the most data available?

1 Like

I suspect that most people who are 1d or stronger have played more than 50% of their games outside OGS. The proportion may vary a lot, maybe between 50% and 95%. I don’t know what conclusion you can reach, just by looking at OGS data.

5 Likes

I’m always interested in looking at this kind of data.

But I have to agree with @jlt that you can’t hope to answer the question “How many games does it take to reach shodan?” by looking at OGS data.

It’s extremely difficult to find opponents for live games at the dan level on OGS, and without dan opponents it’s hard to become shodan.

Lightning games and correspondance games also don’t sound sufficient to improve. At some point you need to sit down and play serious games. And those are even harder to find on OGS.

I’d be surprised if there is more than a handful of shodan who became shodan while playing mostly on OGS.

1 Like

I think it’s worth something to get a result. No one is paying us for academic rigor so it’s not expected.

I had tried something like this. But you need to see what the beginning rank was. So you take a player’s rating history, mark down points of reaching rank and see the distance between them. And average this over many players. But there’s a problem with defining what reaching a rank means, plus retroactively recalculated ranks are crap*. And it’s all very annoying and it never went anywhere, I think.

*Last change was in 2021 so we don’t have a clean rank record for very long.

6 Likes

Sure, but I still don’t see a problem. To build a complex model, you start with a small one. And on top of small models, we can play with assumptions, such as “people with >10k have around 60% games played outside” to build complex models.

Again, I agree, but I don’t see a problem. After building an initial model, we can run a some rough vote among the community the figure out approximate percentages and correlations between player ranks and the percentage of games played out-OGS.

Absolutely agree, thank you.

Maybe several models, one before, one after the change would work.

1 Like