Running hundreds of trading strategies concurrently
- Thread starter fan27
- Start date
From this I have the impression that you are using the "brute force method".
- make for a certain strategy all possible combinations within certain limits and get all theoretical possibilities. That's how you get huge amounts of combinations (called strategies in your case).
- take those who give you the best results.
Mostly people who are strong in math or programming used this way of working. I have seen many attempts but they all ended in failures. The good systems from week 1 didn't work after 1 week anymore and the newly generated best parameters for the next week also stopped working after a short time. It was like a donkey running to catch a carrot in front of his nose. Always running but never able to catch the carrot.
Or perhaps I am mislead but the huge numbers and are you using a completely different way of working...
Years ago I tried to find for a specific system the optimal parameters. When I used brute force I ended up with millions of different combinations... If you use a few indicators with have each a few parameters and a modest range, your number of combinations can explode and run in the billions of combinations.
Imo, there appears from my experience a handful of signals which are highly reliable and they don't appear to go away with time. For example, a quality stock has a different chart shape to a dog stock.
Quality is easier to trade, dogs will lose money in a heartbeat.
If you study the two, it will give you some ideas, namely it's in volatility.
Myself, I can't think of more than a dozen parameters to use in an algo, plus some filters to weed out the rubbish stocks.
Not talking about shorting or Options, only mentioning long stocks here, not futures either.
Quality is easier to trade, dogs will lose money in a heartbeat.
If you study the two, it will give you some ideas, namely it's in volatility.
Myself, I can't think of more than a dozen parameters to use in an algo, plus some filters to weed out the rubbish stocks.
Not talking about shorting or Options, only mentioning long stocks here, not futures either.
Strategy parameters are not optimized. The strategy count is based on different components that can make up a strategy. See thread "a new way to backtest".From this I have the impression that you are using the "brute force method".
Problem is that this is based on past data and will give other results in future. In this way of working there is no edge IMHO.
- make for a certain strategy all possible combinations within certain limits and get all theoretical possibilities. That's how you get huge amounts of combinations (called strategies in your case).
- take those who give you the best results.
Mostly people who are strong in math or programming used this way of working. I have seen many attempts but they all ended in failures. The good systems from week 1 didn't work after 1 week anymore and the newly generated best parameters for the next week also stopped working after a short time. It was like a donkey running to catch a carrot in front of his nose. Always running but never able to catch the carrot.
Or perhaps I am mislead but the huge numbers and are you using a completely different way of working...
Years ago I tried to find for a specific system the optimal parameters. When I used brute force I ended up with millions of different combinations... If you use a few indicators with have each a few parameters and a modest range, your number of combinations can explode and run in the billions of combinations.
you mean you have systems running on hundreds of different instruments
not 1200 trading systems running 1 instrument
right?
i have systems running indexes that have options and the stocks in those indexes on daily.
several baskets that run different models that are of traders who i follow their methods.
some indexes have more than one system running on them but they are few.
--- most important ----
then i run several systems on most all the futures contracts and their cash indexes and their stocks. all total i analyze about 120,000 thousand symbols using a total of 1271 trading systems.
m
so i monitor about half of the global markets and create what i call a "thump"
The 'thump', what is this if you don't mind being asked?
When I as it were, thump the market, usually it means it hit numerous iron ore stocks at once, or copper or gold etc with two trading accounts. However my size doesn't affect anything, just a gnat's bite.
Last edited: