Simple correlations with salf
Expense -0.221
Turnover 0.183
Salf 0.149
SSR 0.188
Multiple regression with salf for historic performance and salf as dependent variable
b t bsd Int 1.687 4.443 0.000 ExpRat -1.232 -4.353 -0.682 Turn -0.008 -3.645 -0.574 OldPerf 0.146 4.103 0.626
r2 = 0.0945
r = 0.308
Multiple regression with ssr for historic performance and salf as dependent variable
b t bsd Int 1.666 4.405 0.000 ExpRat -1.207 -4.281 -0.668 Turn -0.008 -3.415 -0.533 OldPerf 0.845 4.638 0.701
R2 = .1021
R = .3195
Test to see if assets add anything
b t bsd Int 1.854 4.663 0.000 Assets 0.000 -1.516 -0.237 ExpRat -1.300 -4.511 -0.719 Turn -0.008 -3.360 -0.524 OldPerf 0.886 4.816 0.735
r2 = .1059
r = .3255
Note that assets not significant. If anything, larger funds do worse.
Under best of circumstances (know the appropriate parameters) prediction is hard. Use version with exp, turn and ssr:
Range from -10 to 10 (536 funds out of 541)
Note that actual results would be worse since parameters change. Also, survivor bias.