Estimation Theory Trading

[jem]

Quote from Roanoke:

The thesis of this thread apparently has eluded most of ET judging from the mostly negative responses (those few positive responses are much valued). So let me put it another way that most of ET can understand.

Trading is all about character, or the lack thereof. Traders who fail generally cannot follow rules because they are persons of low to no character. But being onesuch myself, I think we moral bankrupts deserve to win just as much as those of high moral fibre.

Our saving grace is that we CAN take orders when closely supervised. My control theory systems provide that supervison and give those orders. Red light short. Green light long. Orange light stand aside (yellow being bit difficult to see on the screen, and this command being the most important in terms of not losing money).

So ye of low character (both ye and we know who you are) may rally around me in this effort to make trading profitable for reform school graduates as well as boy scouts.

Frequently trading seems to be about character.. but it is really about edge matched to bankroll.

 

[eagle]

You forget another important variable, his mood.

Quote from Roanoke:

So I treat price as the raw signal and design feedback outputs to drive the trader to the desired state.

 

[Mike805]

Quote from Roanoke:

My experience is that treating price series as if they were time series measured from physical phenomena is a useful approach if used with caution, understanding that price is not bound by the constraints of physicality. In regards to "robust and tradebadle", does a young mathematical buck like you have any hints for a broken-down old innumerate like me? Best regards, and thanks for the reference.

The pertinent question here is where do you want to focus your efforts? Modeling price series versus "engineering" - if you will - a good system are two very different pursuits IMO. While a good model can be created by anyone with some elementary mathematical knowledge, a good trading system on the otherhand uses an entirely different set of rules and assumptions. In any case the two can work in concert given an appropriate fundamental understanding of the market principle one is trying to exploit.

From the looks of your model (and please correct me if I am wrong), it attempts to time market turns or inflection points. Immediately I would question why your model attempts to "time" something that is very often catalyst - and hence randomly - driven. Because of my narrow views of what does and does not work in the markets, the idea of timing any market event seems to me to be a contradiction in logic. Some have mentioned I am set in my ways regarding this, but, again I have done the research and have come up empty.

The following is all opinion and should be taken with a grain of salt:
A trading system should aim to respond to low/high prices irrelevant of time. Time should definetly incorporated into the trading side/money management aspect of the model, but, in terms of identifying the actual points of inefficiency a timing model is too suspect to curve fitting and general volatility as well as LAG - in other words: current price rules all, historical recent price should be used as a second order input. I run systems that are essentially permutations on two very simple concepts - are prices moving away from value or are they reverting to value? That is it.... I do have some mathematically sophiticated volatility models that help define extrema points, but these models only aid in assessing risk, in as such they act as a fifth wheel in many cases as well.