Well, in his defense and by his own humble admission, nitro is much farther along on both the scientific and spiritual continuum than the rest of us hoi polloi.
Not shooting anywhere near far enough. You will have to trust me on this. The stuff that is discussed on ET is the equivalent to what people were discussing in mathematics in the 17th century as it relates to what some firms do. nitro
I trade a universal style and trade systems as well as discretion and they both work for me. I have no prejudice towards either, except that it is way more fun for me to do research as it relates to systemic trading. nitro
i responded to "The people that could answer that question in earnest don't come and post on ET." since i know some people here are trading sharpes in excess of 3.0, some even above 4.0, which, in my eyes, is quite something. "stuff that is discussed on ET is the equivalent to what people were discussing in mathematics in the 17th" ... well sure, 99.99% of all posts and posters fall into that camp ...
Generally I design my systems to be loose, similar to the old time NASCAR driver, Cale Yaroborough. (he liked his cars suspension set up loose for what he perceived to be better handling and what he preferred). The reason for looseness is two-fold...first and foremost, it allows for execution errors as I am a manual trader, not using automation. Secondly, I find that strings of losses are reduced with this method. A system should be tradeable & profitable even if its loose. Projection of yields should be predictable with your edge and your reports to your investors or yourself should be fairly accurate. Currently, in the scalping system that I trade, I find that exits are more challenging than entries. Michael B.
Funny you mention that, there are 'optimal' ways to design controllers for race-cars that are optimized via simulation where the driver races around computer modeled tracks and detailed physics simulations to find all the optimal parameters of the car configuration. Application of automatic diffentiation to race car performance optimisation A formal method for the evaluation of the minimum time vehicle manoeuvre is described. The problem is treated as one of optimal control and is solved using a direct transcription method. The resulting nonlinear programming problem is solved using the sequential quadratic programming algorithm SNOPT for constrained optimisation. The automatic differentiation software tool ADopt is used for the evaluation of the first-order derivatives of objective and constraint functions with respect to the control variables. The implementation of automatic differentiation is more robust and ten times as fast compared to the use of a finite difference determination of the Jacobian. No, smooth curve simulations are not directly applicable to high frequency trading due to price discreteness (at high frequencies, the brownian motion assumption is completely wrong).
Deringer, you are suggesting what is already known. The issue here is understanding the assumptions made during the strategy design phase that lead to over-optimization / curve fitting. RoughTrader