11:20 I'm not sure, but why do you discard a selection for the other managers? That makes each investor dependent on what the others choose, which no investor would agree to. 3 random investors have a chance of selecting apple, for instance, and there is a non-trivial chance that all three would choose apple. I do not see your logic in constraining the simulation this way?
Thank you for your comment Michael. You've made an astute observation. I set up the portfolios to be "mutually exclusive". I probably should have made any company eligible for selection by any active fund manager within any particular iteration as it is in the real world, but I suspected in terms of the outcome it wouldn't matter. This morning I reran the simulation with the lookup ranks adjusted to 1-10, 1-30, & 1-60. Here are the results for comparison: Mutually Exclusive Portfolios (Original Simulation) P10 P30 P60 P100 Min (0.215) (0.117) (0.038) 0.042 Max 0.312 0.170 0.105 0.042 Mean 0.034 0.034 0.035 0.042 Std Dev 0.073 0.037 0.020 0.000 Skew 0.095 0.024 0.028 (1.000) Kurtosis (0.080) 0.020 (0.034) (2.001) Non Mutually Exclusive Portfolios (Rerun Simulation) P10 P30 P60 P100 Min (0.212) (0.098) (0.038) 0.042 Max 0.371 0.172 0.105 0.042 Mean 0.034 0.034 0.034 0.042 Std Dev 0.072 0.037 0.020 0.000 Skew 0.128 0.037 (0.027) (1.000) Kurtosis 0.059 (0.099) 0.017 (2.001) The max and min values for the P10 and P30 managers improved a bit, and the std deviation improved slightly for the P10 managers. I'm thinking this is a typical simulation run variation not driven by whether the setup is mutually exclusive but I welcome your thoughts.
thanks!
11:20 I'm not sure, but why do you discard a selection for the other managers? That makes each investor dependent on what the others choose, which no investor would agree to. 3 random investors have a chance of selecting apple, for instance, and there is a non-trivial chance that all three would choose apple. I do not see your logic in constraining the simulation this way?
Thank you for your comment Michael. You've made an astute observation. I set up the portfolios to be "mutually exclusive". I probably should have made any company eligible for selection by any active fund manager within any particular iteration as it is in the real world, but I suspected in terms of the outcome it wouldn't matter. This morning I reran the simulation with the lookup ranks adjusted to 1-10, 1-30, & 1-60. Here are the results for comparison:
Mutually Exclusive Portfolios (Original Simulation)
P10 P30 P60 P100
Min (0.215) (0.117) (0.038) 0.042
Max 0.312 0.170 0.105 0.042
Mean 0.034 0.034 0.035 0.042
Std Dev 0.073 0.037 0.020 0.000
Skew 0.095 0.024 0.028 (1.000)
Kurtosis (0.080) 0.020 (0.034) (2.001)
Non Mutually Exclusive Portfolios (Rerun Simulation)
P10 P30 P60 P100
Min (0.212) (0.098) (0.038) 0.042
Max 0.371 0.172 0.105 0.042
Mean 0.034 0.034 0.034 0.042
Std Dev 0.072 0.037 0.020 0.000
Skew 0.128 0.037 (0.027) (1.000)
Kurtosis 0.059 (0.099) 0.017 (2.001)
The max and min values for the P10 and P30 managers improved a bit, and the std deviation improved slightly for the P10 managers. I'm thinking this is a typical simulation run variation not driven by whether the setup is mutually exclusive but I welcome your thoughts.
@@spreadsheetgeek4070 Interesting! Let me think on this a bit after work.