Murray, I have a similar background and have found the "signal processing approach" to be fairly worthless as well. But, I have a different question for you, and I don't intend it to be an attack, it's a respectful question that I hope you can answer openly, because I think it's a fair question. If someone was able to design a "robust system", why would they seek additional revenue by marketing strategy development tools? Certainly a robust system could be scaled up over a few years to provide more money than any one person could ever need in their lifetime. Can you explain this apparent contradiction? Fletch
Software development is expensive and so is research. I develop tools , like TradersStudio and the addins for my own research. In addition I have developed many system and selling some of them does not hurt my trading because most people can't follow a system consistently. Developing a platform like TradersStudio is a seven figure project. Complex addins like my Cycle addin cost about 3-4 man months of work. It only make sense to subsidize development and research by selling copies. Marketing these tools allow me to keep my programming and research staff long term because a bad trading year does not put their jobs at risk, because of the software income. I can just put the software income back into research and development. In addition I can develop large research projects for my own use and just have my programmers work with me on it. One example of that is a program which uses Rough Sets I have developed. This program generates trading rules from data. We have spent 7-8 man months on this project so far and am not sure if I want to sell the tool, because I think it is too valuable. Doing things the way I am, I can afford to develop these type of tool for my own use.
I totally respect and agree with your response here Murray, especially "We have spent 7-8 man months on this project so far and am not sure if I want to sell the tool, because I think it is too valuable. Doing things the way I am, I can afford to develop these type of tool for my own use" Now that makes sense. Best, JJ
Will all respect, IMHO, developing a metholdogly based on logical and proven understanding of the market place and implmenting it consistently is not gambling. It's making money. It's the gamblers who have undue concern for losing their (or OP's-or maybe not OP's) money, not the traders. Best, JJ edit: Hope you're picking up your knowledge trader56, they don't (or won't) tell you this stuff in the books.
Thanks for your feedback Murray. I think your explanation is reasonable. I still have an itch to fully understand, however, why successful edges in the market can't be leveraged into a fortune in a relatively short amount of time. This is a general question and has nothing to do with you personally, although I'd be interested to hear your thoughts on the subject. Something I understand quite well is the game of poker. The reason this is not possible in poker is that as you move up in stakes, the game changes and unless you are one of the 5 or 10 best players in the world, your edge will go negative at some point as you try to "compound up". There's a built-in ceiling that limits your ability to compound your returns. While this may also be true in the markets in some theoretical sense or true on the scale of i-banks or hedge funds, it's not clear why it would be true on the scale of an individual trading his own money, who considers something like $10M to be game over. Fletch
Thanks for clarifying your position. I respect what you are saying. I just don't see it that way. Regards, JJ
There are two problems , first I like researching the markets and gaining a deeper understanding of how they work. Second you can't pull money from a trading operation to fund these large projects or a drawdown period will leave you crippled. The approach you saying works great out of a retirement account where you do not have to even take taxes out, but I am only 42.
Well event processing is not signal processing. If signal processing worked we could use standard time series forecasting methods like ARIMA and it would be game over. Signal processing does not work because the markets are not stationary. It can be used for parts of a system , just like any other indicator but not the core of one. This is true 99% of the time. Just like the currencies are the test case for the trend following world , the bond market is the test case for the signal processing world. If it going to work anywhere it will work in those markets.