You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
What should I do with my model?
- Thread starter galilean
- Start date
Quote from galilean:
There is no special handling for survisorship bias. But it's unlikely that a company will be picked up by my model when it's going to be out of business next day.
Before you trade with your system, I just want to let you know that anything can happen in the market. It is Murphy's law that govern the market.
What should I do with my model?
This question should be asked at the beginning of the design, not at this stage. Why you spent so much time for developing something that you didn't have a clear picture of what you are going to do with it.
IMO, it's very easy to develop something that work base on past data. We just look to the known data and try to find the formula of it, bingo. But what will be the formula of something that going to happen? To correct this problem, why not make some changes, if any, to the system to use current data (now + tomorrow) and if it works well then you know exactly what to do next.
This question should be asked at the beginning of the design, not at this stage. Why you spent so much time for developing something that you didn't have a clear picture of what you are going to do with it.
IMO, it's very easy to develop something that work base on past data. We just look to the known data and try to find the formula of it, bingo. But what will be the formula of something that going to happen? To correct this problem, why not make some changes, if any, to the system to use current data (now + tomorrow) and if it works well then you know exactly what to do next.
Quote from galilean:
I have spent a lot of time in the past few years on studying the stock market. I recently developed a model which seems to be quite promising. My model is for trading stocks (long only, no short, but it might the next step). Back testing has been done based on daily close price starting from year 2000, and the average annualized return is about 40% to 50%, see attached graph for reference.
What should I do with this model? Can I try to sell it to a trading company? Are they interested in this kind of model at all? Anybody interested in this model?
a couple of comments.
first your approach has been the correct one imo. specifically the decision of signaling just long entries. most probably, the edge you've locked in is due primarily to the positive drift that stocks and indexes have. i would say that a equivalent system to signal short entries would be far more difficult to accomplish and i doubt you can devise it as an extension on the existing code.
secondly, there are hundred of trading systems available out there which buy and hold for a couple of days and have a robust historical track record. most of them are widely available to any informed retail trader. obviously, these systems dont have any locked edge even though it is quite possible they'll continue to perform well in the near future, they are essentially a time-bomb ready to go off. hence this means that, most probably, and i am sorry to say this, your system doesnt have any value as "marketable" product or service. to illustrate this just check how the "sell in may and go away" has performed in the last 10 years using the nasdaq components and you'll see what i am talking about.
lastly, i would add a recommendation. why not spending some time translating your work to the emini futures? if the logic doesnt work with futures, then it is obvious it wont work with stocks either imo. benefits are two-fold, first you check the system robustness and on the other hand you'd be able to put it into action with just a fraction of the capital.
good luck
What about 2007 and YTD, that's one of the most interesting time.
This question should be asked at the beginning of the design, not at this stage. Why you spent so much time for developing something that you didn't have a clear picture of what you are going to do with it.
My idea was primarily inspired by the text book "Mechanical Trading Systems - Pairing Trader Psychology with technical analysis" Wiley 2005, by Richard Weissman.
Do traders really use some models that are backtested or they just use their sense to trade?