The last conversations have inspired me, so I've been collecting data to analyze and I want to share some conclusions:
- About the need to calculate profit factor or expectancy, I agree with CavaliereVerde
All the darwins with D-Score>75 have profit factor>1 (except ABH), so you don't need to calculate it, it's enough with a high D-Score but it shows that a good darwin needs profit factor>1 (or expectancy>0, it's exactly the same)
The average of the profit factor of the darwins with D-Score>75 is 1,52.
- About the profit factor as an indicator, I don't agree with NemesisCraze
There's no correlation between profit factor and D-Score (0,05), so you can say hte higher profit factor, the better.
All that darwins have Ex>9,40 and Mc>9,20 but these attributes have no correlation with D-Score (-0,03 and 0,06)
The only attribute with a strong correlation with D-Score is Pf (0,50), it shows that Pf is the most important attribute in D-Score; the average Pf for these darwins is 8,85 and the minimum is 6,70
Attributes like Ex, Mc, Rs, Ra, R+, R- and Dc have no importance in the D-Score, you can set a standard rule and forget about them (Ex>9,40; Mc>9,20; Rs>5,2; Ra>6,60; R+>3,10; R->3,70; Dc>3,80)
Attributes like Os, Cs, La and Cp show a large standard deviation and this means that it is not true that the highest is the best, there are very good darwins with low Os, Cs, La or Cp
La has a strong positive correlation with Cp (0,66 ) but a strong negative correlation with Os (-0,65) and Cs (-0,79) and Cp has a strong negative correlation with Os (-0,50) and Cs (-0,69); this means that among the best darwins there are two different types:
A) darwins with high La and Cp but with low Os and Cs
B) darwins with low La and Cp but with high Os and Cs
La, Os and Cs are strongly correlated with average winnings/average losings (La: 0,63; Os: -0,50; Cs: -0,79); if average winnings is much higher than average losings, then La and Cp are high but and Os and Cs are low (all the darwins with La<5 have Cp<1,50, Os>7,20 and Cs>8,20 and average winnings/average winnings<0,84; all the darwins with Cp>2 have La>5,10)
IMO, there are attributes with less weight in D-Score that can be set to be sure of having a high D-Score:
Another rule you can set is VaR underlying strategy<25 (all darwins with D-Score>75 have VaR<25 except NSC and OSM).
I think CavaliereVerde's rule should be mandatory: do not invest in darwins with return> 5% in the last 3 months.
Another mandatories: divergence>-0,50 and a minimum capital (>50000).
There are clearly two types of good darwins (high La and Cp; high Os and Cs and low Cp), so I don't trust darwins that don't fit into one of these types.
Based on the data, I build a general filter for any darwin (already shown) and specific filter for each type of darwin:
Examples: STV, FEG, THA, HFD, BPX, QUA, OLE, ICX.
Examples: HZY, OVL, ABH, SKJ, CBH
I'm sure I forget some mandatories and this portfolio needs exit rules.
Please, help me.