Why constantly optimizing fails to work. Proof. - page 2
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I had interesting experience related to this topic, someone might find it interesting.
Last year I was trading bitcoins using EMA strategy bot. I have started to do it somewhere in September, and I had to optimize parameters once per two weeks. Then after like a month I had to optimize them already once per one week, otherwise I would get losses. I did't make any researches, and basically I was updating parameters as soon as bot started to make wrong decisions. Anyway, this approach was getting me like 30% income per month. All this ended in the December, when I had to update parameters every day, and price chart was so unpredictable that EMA strategy just stopped to work for this currency. But in the beginning of 2013 it was enough to update parameters once per month... So required frequency of optimizations was just rising during the year, and then strategy become useless.
There is interesting research somebody can do:
Take current date minus few months, then find best parameters for 30 days of backtest. Then fronttest those parameters for few month forward, make profit chart for each day. At some point of time this profit chart line will go to negative area, it will show you time window that will be profitable for this EA. Repeat this tons of times for different starting dates to get average picture.
Then change backtest day amount from 30 to 60, and repeat all over again. You will get another chart with average profitable time window, and this time it will be different. Goal is to repeat all this again and again numerous times to find amount of backtest days that will give you bigger profitable fronttest time window.
I am not sure, but I think that in result you will have backtest and fronttest lengths very close to each other.
I had interesting experience related to this topic, someone might find it interesting.
Last year I was trading bitcoins using EMA strategy bot. I have started to do it somewhere in September, and I had to optimize parameters once per two weeks. Then after like a month I had to optimize them already once per one week, otherwise I would get losses. I did't make any researches, and basically I was updating parameters as soon as bot started to make wrong decisions. Anyway, this approach was getting me like 30% income per month. All this ended in the December, when I had to update parameters every day, and price chart was so unpredictable that EMA strategy just stopped to work for this currency. But in the beginning of 2013 it was enough to update parameters once per month... So required frequency of optimizations was just rising during the year, and then strategy become useless.
There is interesting research somebody can do:
Take current date minus few months, then find best parameters for 30 days of backtest. Then fronttest those parameters for few month forward, make profit chart for each day. At some point of time this profit chart line will go to negative area, it will show you time window that will be profitable for this EA. Repeat this tons of times for different starting dates to get average picture.
Then change backtest day amount from 30 to 60, and repeat all over again. You will get another chart with average profitable time window, and this time it will be different. Goal is to repeat all this again and again numerous times to find amount of backtest days that will give you bigger profitable fronttest time window.
I am not sure, but I think that in result you will have backtest and fronttest lengths very close to each other.