Quote from mahras2:
I would like to develop some discussion on the optimal number of degrees of freedom a system may have which would not be curve fitted. What has your experience been in playing with different degrees of freedom? How did you go about obtaining the right balance of being optimized so that it can extract profits the system is capable of but not so much that its curve fitted and will fall apart in the future?
This is a tough discussion to deal with.
Bounding the discussion under the five issues you wish to be handled is a remarkable and adroit consideration.
1. Optimum # of degrees of freedom (DF's)
2. Play with the range of DF's,
3. Balancing Df's and/with curve fitting,
4. Optimizing extracting profits, and
5. Dealing with change (markets) over time.
I extracted 28 comments of respondents over the last 8 pages (See Word Attachment) which scope and bound the discussion heretofore. They bound the opportunity to four levels of performance as depicted by 8 persons, five of which are actual practitioners of your scope of inquiry.
The solution you need lies outside of the constraints imposed by your respondent's general approach. Their Sharpe ratios scope and bound the solution they assimilate. What they offer is an automated speculation box whose guts is based on the maths of the usual stat and prob theorems. Too bad.
I spent the time to deal with what is required to replace the limitations in the collective response you received. The literature that parallels this commentary is not replete mostly because it deals with money and that makes it kind of religious, worshipful and mystic.
So there are two classes (kinds) of black boxes to deal with this tough situation. They are incompatible.
My results are.
1. The optimum DF's as inputs are 6 containing four correlated pairs deploying 5 of the 6. Four output DF's emerge.
2. When you play with the opportunity a serious choice must be made early on. probability and statistics type math has to be dumped because of limitations of performance. The alternative is more complex and elegant. Five throughput stages are imposed between the input and output stages and interim DF's are created. The stages steer and focus the DF's to deal in a timely manner with the everchanging complex opportunity. Bottom line is a nominal 70 DF's (not concurrent) having 61 displayed signals.
3. It is required to take this off the table. This is accomplished by changing mathematical basis of handling everything using a different functionality. The goal does not turn out to be balancing but, instead, to achieve efficiency and effectiveness of extraction in harmony with EMH. This is probably the basis for most disguntledness of the B people here.
4. The range of Sharpe ratios of the four levels represented here by respondents is well below what is possible and so far you just have the conventional standards represented vaguely. To shift to Sharpe ratios that meet the potential offered, the alternative maths required make use of "drawdown" as an opportunity instead of an econometric "substitution" effect. By eliminating this probability and statistical "trap", the ratios go to the "too good to be true" range.
5. Taking 3 off the table contributes greatly to not being "locked in" to the contemporary scene. One of the keys to this is to consider the DF's data flow in the context of the system of numeration used. It appears that by deploying a "phantom" base for explanation purposes is most helpful. By defining an order of magnitude with 5 symbols you have the equivalent of a base five system in operation. This nut will be very hard to crack by most. By changing bases numerically the blow up potential of a large number of DF's is avoided.
All of the above is about 15 years old meaning it has been in use since prior to the invention of the eminis.
I have written this stuff up in an extensive paper by culling my experience and the associated information into the five categories you suggest. I don't expect my comments to generate any discussion; if any facets are of interest, then I will continue.
As has been seen in financial history, things change not by osmosis but by change of concept. The automated box the respondent group is in has limitations that are not going to be resolved by iterative refinement. It is more a "sea change" type of fix that has to be put in place. The current purveyors are going to be very persistent and maybe frustrated but that is how it always has been. The vacuum tube was replaced by the transistor and the transistor was replaced by the chip and the chip will be replaced by nano tech, etc... That is how it is.
You can use the attachment to collect the situation. I just plugged the citations into the paper I have here as a way of pointing up the contrast between the paradigm here and the successor paradigm. The successor paradigm is based upon risk minimization instead of what you see here.
