In my quest to define a winning trading strategy, I'm obviously worried that all my work will be wasted, and that trading is more of an art form than anything else. So, you can backtest as much as you want, but in the end there is zero bearing on your potential future profits. Or is there? Your thoughts would be appreciated.
I suspect it depends on how you define "consistently profitable." There are plenty of systems that are highly profitable over a period of years when applied to a portfolio of commodities. They might have big drawdowns and long flat periods however.
My view is that the answer is an ever-changing weighted function of your 2nd and 3rd options... I take it that when you refer to "trading system" and "trading strategy" you are equally open to the ideas of strictly mechanical systems, hybrid mechano-discretionary methods, and highly discretionary approaches... Bearing in mind that trading success emanates from the sound fusion of consistent execution / discipline (a function of psychological make-up), flexibility (having several approachs / systems, and knowing under what conditions to use them), entries, exits, risk considerations, cost ratios (commission impact etc), a reduction in the efficacy of just one of these elements of your approach can lead to a total breakdown of the system / strategy... remember that Space Shuttle Columbia was 99.9999% in good working order, but disaster still struck... Another thing: I agree with AAAintheBeltway... what precisely constitutes "consistent profitability" is a highly debateable area of discussion... for me, consistency is to be in profit on the vast majority of DAYS... but for others, perhaps consistency could be defined in the context of MONTHS...
I agree with both of you. I didn't want to provide any specific timeframes for defining "consistent", since the word probably has different meanings for different people. My own personal definition is that, considering how much effort I am placing into my system, I would consider "consistent" to mean at least one year, since I would not be able to put the same amount of work into developing new strategies every quarter or six months. Also, refering to candletraders comment, I am talking primarily about mechanical systems, though I would also like the input from people who employ discretionary methods as well. My focus is on mechanical (i.e., rules-based) because they are much easier to backtest on publicly available data.
I've been wondering the same as I work away at system after system. Every time I read something on this board that sounds interesting, I code and test it, and I've been doing this religiously for a few months, along with every permutation and idea of my own. I've tested a lot of systems. I'm sure there are crack traders around here who have great mechanical systems, but I would guess that there are a lot fewer of them than it seems most of the time. I think the truisms about most entries being scarcely better than random are correct a lot of the time, and that it's mostly about your stop, size, and discipline. I'm beginning to believe that trading is mostly an artform. (I hope so anyway, because I don't have shit for a consistent methodology!).
Yes, but they are no fully mechanical systems, there is a definitely place for an art of adjusting the mechanical system in working. Walter
Sure, plenty of Donchian or MA based systems have tested to win year after year. As with any computer based system that "solves" the market, the so called edge will eventually be neutralized when market conditions change and you'll have to find another system to make money with.
Not to sound like a nimrod, but I've scarcely been able to generate year over year expectancy with either channel break or MA based systems... using everything, multiple timeframes (2, 3 even 5), enter at one length exit at a shorter one, stoplosses, exits, trailing exits, pyramids, sizing, etc... can you give me a plain vanilla example of a donchian or MA mech system that profits year over year on a specific instrument? (this isn't meant to be a challenge, I'm just curious, I'm probably missing something really obvious)
Or at least it always helps to use such an art, by for instance minimizing the impact of the weaker points of the system if you can identify such.
Sounds like you're trying too hard.. systems aren't meant to put up pretty numbers with any regularity unless you're curve fitting over past data .. a simple 200 day MA cross where you buy on the open after a close over the 200, or short below, etc. is probably only about 30% winners but will end up with a net profit almost year after year. Likewise, a Donchian breakout might be around 45% winners, with a 1.5 to 1 ratio.. plenty to make money over the long haul, but you can still lose 8 in a row. Again, it's all relative... a system is considered quite a success if it can average 20% a year .. whereas most daytraders on here would find that to be barely enough to live off. Don't take my word for it, but I highly doubt that there are any systematic 65%+ win systems with a 1.5 to 1 ratio out there and averages 100% net per year, unless they have been grossly curve fit and therefore rendered useless.