Hello, ( I may have accidentally submitted an incomplete version of my question): I am an intermediate-beginner in the process of trying to develop a trend following system. While I have gotten pretty good at eliminating some very bad trading ideas (considered good by some trading gurus) and making some obvious improvements, my dilemma is the following: 1. How do you decide when to end the research/optimization process and actually start trading? It seems that there is always another way to "improve" or change the system. 2. If you have several systems to choose from with mild to low-moderate differences in profitability, how do you choose which system to utilize: This question is harder than it sounds. I will clarify: Currently, I am researching two trend following trading based systems: A faster one and a slower one. On a given move, it appears that the faster trend system is the winner: It managed to catch the one minor breakout bar right before a big spike up!! The slower system would have still been short during the big spike up!! Loss versus profit. Also, for two moves before hand the slower system entered a trend late after 30 percent of the move was underway. Finally, you notice the slow trend system entered a move just as it was ending and reversed too late However, scroll back and discover that: A. On another trend move the slower system stayed in for the whole move, without being faked out, while the fast system made three false moves during a mid trend consolidation! B. Paradoxically, the slow system actually got was able to get into a powerful uptrend "faster than the fast system. "How? iI caught a slowly developing trend and did not fall for a fakeout bar (before the big upward spike bar). C. The Fast system had a set of three whipsaw trades in a row, that the slow trend system did not. However, on many other trends that developed more slowly the difference between the two systems was negligible. It appears for every negative price shock that occurs with a system, you can always find some variation that would have avoided the price shock ( as in sudden spike in the wrong direction) or even entered the spike in the right direction. For every choppy period or trend in a system, you can find a variation that would have done better. How does one avoid the trap of deciding to add a rule or remove a rule every time some variation would have done better at handing a specific situation? If one system had two price shock moves and another had three within a given time period, how do you know if the results ( that favor the system with only two price shocks) are material or randomness. While one should discard inferior systems, as one becomes better at making genuine improvements to systems and adding genuine improvements based on material factors does have a place, how does one know when to stop looking for a better system, accept one of the developed systems that is net profitable, and shut up and trade. To conclusion, I don't want to of course utilize an inferior system, when several choices exist, but I don't want to fall into the trap of traders who always jump around from system to system, looking for "something better." Any thoughts on handling this matter would be tremendously appreciated. Thanks very much
You need a system that when backtested for a minimum of, say, 75 trades is not only profitable after accounting for slippages and fees but also beats the buy-and-hold result. If you can't beat the buy-and-hold, then there's no point in trading, right? (But I continue to be amazed at how many newbies consider beating the buy-and-hold to be unimportant.) So once you get a system that can be net profitable and beat the buy-and-hold, then you can look at paper trading or maybe even testing a small account with it.
It comes down to a few factors, your personality, committment and maybe your age. I am in my fifties, been trading over 30 years. What is important NOW as opposed to 30 years ago is low drawdown, I don't consider what makes most profits, cause generally, most profits offers highest drawdown. I have found to stay in the bigger profitable trades, once I get to breakeven stop, I don't move the stops. Matter of fact, I go to next higher timeframe, so if entry was on daily, I go to weekly charts and this has always allowed me to not get stopped out on noise of the market. I am constantly running backtesting, methods always have to be tweaked cause your experience expands. But as you learn less and less as years go by, tweaking becomes less as well.
Robertwiz Sounds like you've done your homework and most of the heavy lifting. The nemesis of most trend following systems (generally speaking) is highly volatile choppy markets or congested markets, which lead to draw downs. Fast and slow is a vague description, especially when you talk about, "shocks and spikes." I'll assume you have a couple intra-day systems. That said, are you, or can you verify the segments of time these spikes occurred, (e.g. breaking news or economic report release). Breaking news cannot be "traded around" like a scheduled economic report release, the best we can do is protect our capital. The flip side of that is, are the "windfall profits" of these shocks and spikes included in your backtesting as profits, if, the outcome is favorable to your account. The size of your sample should tell you which system to use. Most traders consider drawdown very important, I use 2x my maximum run of losses from back and forward tests as a system stop. If the system triggers this system stop, I can still return to the market and trade.
Robertwiz <<Sounds like you've done your homework and most of the heavy lifting. The nemesis of most trend following systems (generally speaking) is highly volatile choppy markets or congested markets, which lead to draw downs. Fast and slow is a vague description, especially when you talk about, "shocks and spikes." I'll assume you have a couple intra-day systems. That said, are you, or can you verify the segments of time these spikes occurred, (e.g. breaking news or economic report release). Breaking news cannot be "traded around" like a scheduled economic report release, the best we can do is protect our capital. The flip side of that is, are the "windfall profits" of these shocks and spikes included in your backtesting as profits, if, the outcome is favorable to your account. The size of your sample should tell you which system to use. Most traders consider drawdown very important, I use 2x my maximum run of losses from back and forward tests as a system stop. If the system triggers this system stop, I can still return to the market and trade.>> It seems like the jist of your replies is that if you have the choice of several systems that are within a close range of profitability, you should choose the one that you feel subjectively comfortable with. Also, it probaly is a good idea once you find a good enough system to commit for a specific time. yes, the spikes are included in profitability. By the way, certain "nothing but the chart/trend fundamentalists" seem to be similar in mindset to efficient market fundamentalists. Just like EMH fundamentalists believe that everything is reflected in the market, certain system fundamentalists believe that everything is useless with the sole exception of systems that work on every market. In other words, attempting to use outside factors to decide when to use a system, what markets to use or to pick a specific system for a specific market are useless. However, might there be some validity when designing a mechanical intraday trading system to maybe turning off systems during economic reports, picking markets based on market dynamics, and altering a system based on the specific market. For instance, the times a market opens and closes, the existence of secondary markets and strength of such after hours, and the volume of a market might affect chart patterns and whether a market reverses at a fast speed or slower speed. Thanks
I would say choose the slower system. In my experience it is more likely to be robust. Then concentrate on developing other shorter term uncorrelated systems rather than continually tweaking your existing ones. For example I trade one very long term trend following system, one medium term trend following systems. Two short term index systems and two intraday index systems.
Designing a trend-following system is very easy. Trading it and following it is hard. Actually, talking about developing problems in this area is a distracting issue from the real problems: (1) How do you size trades? (2) What is the proper capitalization? (3) Discipline
================== Robert T Whi; Well, First[1st]never stop studying/never stop study of trends.; which i personally would consider laziness........ No offence intended. %%%%%%%%%%%%%%%%%%%%%%%% The fact you asked the question[good question] probably means you should not trade yet, Bull & bear trends are so different.......................................................... Wish i had started with 1 share. Generally slow is much better, less comissions; which is precisley why there are many , many,many more profitable investors/longer term traders than day traders. Wisdom is profitable to direct. Or another way to put it-Market Makers Edge[daytrading] book has multi-MONTH charts.................................................................... Funny but true......................................................