=== true edge +1.0pts over break-even · estimate error 0.0pts · pool 20 ===
  selected 14.9 bets/slate | believed edge 1.64pts vs TRUE edge 1.64pts on the bets actually taken
  full Kelly, uncapped                115.34
  full Kelly + 3% per-bet cap          88.51
  half Kelly, uncapped                 87.26
  full Kelly + 15% total cap           56.47
  quarter Kelly, uncapped              50.91
  full Kelly + 3% bet + 15% total      48.24
  flat 1% per selected bet             39.85

=== true edge +1.0pts over break-even · estimate error 1.5pts · pool 20 ===
  selected 13.6 bets/slate | believed edge 2.11pts vs TRUE edge 1.55pts on the bets actually taken
  half Kelly, uncapped                 72.21
  full Kelly + 3% per-bet cap          67.38
  full Kelly, uncapped                 49.72
  quarter Kelly, uncapped              47.39
  full Kelly + 15% total cap           45.95
  full Kelly + 3% bet + 15% total      40.44
  flat 1% per selected bet             33.91

=== true edge +1.0pts over break-even · estimate error 3.0pts · pool 20 ===
  selected 12.3 bets/slate | believed edge 3.07pts vs TRUE edge 1.42pts on the bets actually taken
  full Kelly + 3% per-bet cap          51.48
  quarter Kelly, uncapped              43.71
  half Kelly, uncapped                 42.33
  full Kelly + 15% total cap           35.87
  full Kelly + 3% bet + 15% total      34.05
  flat 1% per selected bet             28.25
  full Kelly, uncapped               -146.01   ruin 0.011%

=== true edge +0.0pts over break-even · estimate error 2.0pts · pool 20 ===
  selected 10.0 bets/slate | believed edge 2.00pts vs TRUE edge 0.72pts on the bets actually taken
  quarter Kelly, uncapped              14.79
  full Kelly + 15% total cap           14.49
  half Kelly, uncapped                 13.83
  full Kelly + 3% bet + 15% total      12.26
  full Kelly + 3% per-bet cap          10.46
  flat 1% per selected bet              9.48
  full Kelly, uncapped                -38.29

(growth = mean log wealth change per slate, x 1e-4)