Hello I am slowly moving from the manual (manually back test and manual day trading) trading arena to the programming, developing, testing, and algorithmic trading arena. When a trader moves towards developing algorithmic trading systems: 1. How do you find trading strategies to program and test? 2. Is it true that to properly find and develop a winning strategy/system, a trader may have to test nearly 100 trading ideas before moving forward with 1 proven system? I recent watched a webinar over algorithmic trading and the speaker stated it usually takes testing 100 to 200 trading ideas before find 1. 3. When you find a winning strategy, do you stop finding and testing strategies? 4. How long on average does it take you to program the trading strategy? Thanks,
If you have one strategy I mean unless it's satisfying all your goals and financial needs and it's been effective for decades, outside of that why and really even than why would you not look for other strategies that may have less drawdown, higher profit factor or even that may allow you to diversify through time? Or did I misunderstand #3?
Not asking you particularly to answer that, I just can't imagine why you wouldn't so I am asking this ahead of time incase someone decides to answer yes so they will tell me why.
Thanks trdes, I will rephrase question 3: 1. Once you find a winning strategy via your back testing and forward testing performance results and you start trading that strategy live, do you stop looking for other strategies to test, or slowly scale up that 1 winning strategy that is now live after some time? Thanks,
1. Have an idea, usually have some 50 ideas in a pile, matter of when I will get time to test all of them out, it snowballs from idea to going to the charts and manually mark trades to see what the risk is going to cost. The only systems I have bought and use besides my own are all long term, nothing worth while for day trading, so all the systems are either made within or I will read something on internet and test that and see all the flaws or why it doesn't work, sometimes that snowballs into workable system, it is more fun than crossword puzzles. 2. I never counted, it is much more of your knowledge of how price reacts, your skills. 3. No, I study much more now than test as I know what works for me and when something looks like it fit my losing percentages then start to program litely entries. I have my software where I had it made so it will accept "units", so I have a risk management "unit", then entry, then ad-ons etc, so I don't have to think up of how to risk control any system as I am always going to use same rules for day trading. 4. One system took 2.5 years of working on it 40 plus hours a week. When you start adding options to stocks/commodities, they get pretty long, some concepts may look easy but aren't when you program them. 1. I trade one lots when starting automation for six months, one lot for every signal and no averaging down, after six months I turn on averaging down for six months then add so many lots each month, I use 10k for each ES I trade so that gives good amount against risk of ruin. It is a long process, cause losing is FAST! Most systems I have made and automated grind it out, have made three I don't talk about anything inside of them, am not a hog and don't do thousands of contracts as people somehow eventually discover what you are doing, Hogs get slaughtered. But it is more fun to be able discover something, and always seems by error, of whatever I am trying to do. It's the challenge and fun.
Thanks Handle123, I like your comments about: 1. Starting with 1 contract 2. Thinking of the process as fun challenge like a puzzle 3. Its long process, cause losing is FAST. Definitely can relate to this. Good information
A result of all the stuff in my head. Trading books read. Calculus. Art. Programming. Geometry. Etc. Ideas 'pop' in. Deliberate thought refines those ideas. They are rejected or pursued further. Many ideas grow from previous ideas. The new idea may take weeks to 'pop' into my head. That hasn't been my experience. It's usually not a case, (for me, at this point) of whether the system is profitable; it usually a question of whether it's more profitable than what I've already developed. Ideas still flow into my head (so far). Out of curiosity and thoroughness, I'll usually test them to see if I can improve upon my current system. From a couple of hours to 20 hours for the initial prototype.