Getting away from trying to automate a strategy

Discussion in 'Strategy Development' started by BillySimas, Apr 14, 2010.

  1. I've been using TradeStation for awhile and I have built a few automated strategies that are profitable, but not enough for me to have enough confidence to go live with them. Every time I try to base a strategy on a specific set of rules and then backtest it and look at the results trade by trade, I realize that price action is too often dictated by simple support and resistance levels, not indicators or whatever the basis of my strategy is. I've always told myself that psychologically, I would really require clear evidence of a mathematical edge before I'm able to trade, but making up rules based on support and resistance levels is very difficult from a programming perspective. It really is more of an art than a science, which is annoying to me because I like to keep things objective and logical. I used to read TA books when I got into trading about 10 years ago and I loved it and now I think it's mostly nonsense and far too subjective. I want to eliminate as much of the subjectivity in my trading as possible, but the more I study price action, the more I find that there are way too many variables for me to program into an automated strategy. I feel like I might as well just trust my own instincts and trade on a discretionary basis using my own pre-defined set of rules and price points that are particular to each individual trade instead of coming up with some blanketed strategy. Is there anyone else that feels the same way or has had the same experience?
  2. pipdaddy


    I've thought of trying a semi-automatic system. Where i use discretionary entries and have the program manage the trade from there on. The biggest problem i have with that type of system is that it's not really backtestable so you don't know how it will do till you do it.
  3. I attempted to automate my trading by using macros ( I use Amibroker software ) but gave up.
    I've been trading over 20 years and have excellent support in that I have a highly skilled guy write programs for me, ie Amibroker formulas.
    I've designed numerous strategies and formulas, backtests, T/A charts etc in conjunction with a professional guy who is a topgun in Amibroker.

    In the end it boiled down to this, time, energy, broker commissions, slippage, data costs etc increased the more one traded. As well, in my mind I thought, the more I traded, the more my broker could take advantage of my moves and front run my stops. That's what seemed to happen, might be my mind playing tricks, but was stopped out prematurely numerous times.

    So, in order to get my advantage back, I slowed down on my trading, now I place orders manually, not as often, and play quality stocks with those which have lower $ daily turnover and lower volatility.
    I take more care choosing my stocks, I have a watchlist of stocks 'I dont wish to trade' which is much bigger than those I wish to trade.

    This is a game where learning is a constant, never stops. That's what I love about it.
  4. I agree, you're constantly learning from your mistakes and seeing what works for you and what doesn't. You figure out what suits your personality, what time frames and risk/reward parameters you're comfortable with, and it's basically like trial and error. I keep telling myself that I'll be able to hammer out a concrete set of rules that I can stick to, but the market is constantly changing and what works for one period of time will not work at all in the next. If you have the right frame of mind and understand the way the game works, I think you get to a point where the only true determinant of success that you have is your P&L. I'll just trade small size until I show consistency and go from there, thankfully I haven't wasted a lot of time or money yet.
  5. Whether one trades minute by minute, daily or weekly, the charts look similar. So I now use a weekly time frame which obviously has wider stops, but at the same time has more room for the stock to breathe.
    I looked at monthly time frames too, but thought that was too wide in terms of stops.
    I play multiple positions simultaneously, ie might be in 20-30 stocks at any one time. I keep position sizes small.
  6. You will not find anything like what you're looking for in pure speculative markets such as the ones you're likely trading in.

    "Clear", "Mathematical" edges come from one (or a combination of) a few spots:

    1.)You are a spreader and you're essentially arbing different markets
    1a.)Your systems are faster than everybody else trying to do this

    2.)You are a "High Frequency" trader - by this I mean, you engage in order-flow front running.

    3.)You have material, non-public information upon which you can act.

    Otherwise, there's no "one thing" that's going to give you a specific measurable edge.

    What I suggest is to examine your rules in aggregate - that is, your signal generator might not have a ton of edge (if any), nor will your order management have any. However, used in combination and used according to strict, automated rulesets, you might be able to generate significant alpha.
  7. To the OP, I have been where you are now.

    If you havent found what you wanted to find within your indicator systems ( and you wont ) at least you moved on to the right path: price.

    I too have stumbled along the price road for a longggg time. It is very difficult and takes a very long time to code a way to exploit what you find. But if you make the journey you will find the edges. It's the only way, give up or keep on truckin. I wish you the best.

    My tip to you would be once you find something keep it simple. A stable 1.5-2.0 PF that has no optimization is much much better than your optimized indicator code. To trust your results you must code price. Its the only way.