Paying your dues

new_trader

Legendary member
Messages
6,770
Likes
1,656
I have been studying the markets as a long term investor for over 10 years but only decided to start daytrading about 12 months ago. Since then, I have spent US$1000 on historical data and hundreds of hours analysing and backtesting it with a bespoke program that I wrote myself using Visual Basic 6. That alone took 100's of hours of writing and debugging.

So far I have lost almost US$3000 trading my system and have resorted to paper trading it. I have NOT lost faith in my system, I just think I need more confidence before continuing. I took the advice given in another thread and decided to back test it further by purchasing more historical data. This restored both my enthusiasm and confidence, as it has now been tested with around 20 years of historical data and the greatest majority of that is out of sample to my first back test. The first time I tested 2002-2005, I have now tested 1985-2006 (end of March) and it is robust. But I have never had the guidance of a mentor (apart from the support/advice here) and I have never had experience on a trading floor. I can’t help thinking that there is something elusive to the markets and I haven’t paid my dues yet.
 
There is always an entrance fee.
I suggest you have a look at the historical results on a short time frame. I have a great looking crude oil system that makes money consistently even over 25 years of data I get a great equity curve including the last 3 years since I wrote it. If you look at the entire curve it looks like a lovely smooth gradient all the way...but..I do not trade this system and I am unlikely to ever start as if I drill down to, for example, 3 month segments I see that there are multi-month periods of drawdown along the way. It is very hard to maintain the faith when you have such a system. A few months is a very long time in trading your own account.
You need to be 100% sure that what you have developed suits your own psychology and funds at risk. If the $3000 you have lost is well within range of past performance then the system is probably still working but your expectations were wrong regarding how soon you would make a return. Always assume sods law and that the start will be right at the start of a draw period. If $3000 is the worst loss the system has ever suffered then maybe it was just very unlucky or it has been curve fit in some respect or something has changed structurally in the market you are trading .
I am not sure whether developing in VB and paying up for historical data is the best way to test these ideas maybe you should look at cheaper alternatives such as Tradestation if you want to test a lot of ideas and adjustments around your core system. Certainly would have been far cheaper than the method you have chosen and it is designed to give results that are relevant to understanding exactly how a system performs over different time frames.
 
twalker said:
There is always an entrance fee.
I suggest you have a look at the historical results on a short time frame. I have a great looking crude oil system that makes money consistently even over 25 years of data I get a great equity curve including the last 3 years since I wrote it. If you look at the entire curve it looks like a lovely smooth gradient all the way...but..I do not trade this system and I am unlikely to ever start as if I drill down to, for example, 3 month segments I see that there are multi-month periods of drawdown along the way. It is very hard to maintain the faith when you have such a system. A few months is a very long time in trading your own account.
You need to be 100% sure that what you have developed suits your own psychology and funds at risk. If the $3000 you have lost is well within range of past performance then the system is probably still working but your expectations were wrong regarding how soon you would make a return. Always assume sods law and that the start will be right at the start of a draw period. If $3000 is the worst loss the system has ever suffered then maybe it was just very unlucky or it has been curve fit in some respect or something has changed structurally in the market you are trading .
I am not sure whether developing in VB and paying up for historical data is the best way to test these ideas maybe you should look at cheaper alternatives such as Tradestation if you want to test a lot of ideas and adjustments around your core system. Certainly would have been far cheaper than the method you have chosen and it is designed to give results that are relevant to understanding exactly how a system performs over different time frames.

My program writes a txt file with every trade the system makes so that I can import it into Excel. I then use filters to see performance over different periods. I did look into buying Tradestation but it looked like the more expensive option? It was priced over US$2000
 
Top