Seeking feedback on my rules-based system

Discussion in 'Automated Trading' started by morganpbrown, Apr 12, 2019.

  1. I started my work using five indicators simultaneously, but abandoned the effort because I wasn't getting rich enough statistics for each realization of the five indicators. I guess if I was daytrading, it would be a different story. Maybe I'll revisit the multiple indicator prediction.

    Volume (daily) is definitely an attribute that I can easily mine. It's certainly worth trying past volume as an indicator. On what timescales do you typically deal with volume? past day/week/month?
    #21     Apr 13, 2019
  2. Sprout


    Since you are using indicators, these two threads have rich content:

    You seem to like larger timeframes and sector rotation. DryUp & First Rising Volume would be of interest to you.

    Since the market is fractal, volume patterns are similar in structure regardless of timeframe. My trading fractal is the ES 5m. From what you are testing, there is noise introduced into your system whereas faster timescales filter it out. This is a minority viewpoint. An example would be the volume profile for extended session trading. Since liquidity has a functionally distinct characteristic than the price/volume relationship during regular trading hours, this introduces noise that averaging seeks to reduce. This embeds a limitation in it's approach of extracting the market's full offer even though many have relative degrees of success with it.

    The above links are based in a different paradigm and will most likely not make sense to a majority of readers, nor to you unless one can embrace out-of-the-box open-minded thinking. Backtesting produces mixed results in that the assumptions generally made in a backtest do not fully incorporate a system's complete paradigm nor does it account for a tester's implied bias. This is true for most approach backtesting from an 'outside-in' perspective, using it as a way to determine the promise of a particular strategy or tactic based on an incomplete understanding of how that indicator functions on it's own as well as among it's peers.

    Another approach would be to fully understand the components of a particular indicator and from an 'inside-out' perspective see which market conditions are appropriate for that particular indicator to perform well. In this way, one understands the strengths and weaknesses of any particular indicator as well as how to combine them into a something that is 'greater than the sum of it's parts.'

    To use driving as an example, there are a number of different weather/road conditions, terrains, vehicle capabilities, traffic patterns and intent that determine driving conditions which inform how to best drive to get from point A to point B.
    Each condition requires a slightly different skill set even though they can all be considered in the domain of driving.

    All in all, where the rubber meets the road is in forward testing both as prerequisite for proof of concept and more importantly as a process as one actively trades.
    #22     Apr 13, 2019
    HobbyTrading likes this.
  3. danielc1


    You mean 1) Having an edge, 2) Riskmanagement, are the only 2 ways to make money in trading. Right? What I do not see in the OP system is an edge. No matter how complicated it will get. It is just a lotto bias where he thinks he can 'control' the outcome, because he can choose the 'numbers'. Hence, no history will tell you exact what will happen next. The only thing you can be sure of is what you lose, when you are wrong, if you do not put up blinders. I see big blinders in OP post. But he/she will figger it out.

    You fill in the edge. Take anything you believe. Stats, indicators, planet orbits, you name it. You believe it, you trade it, you get results and think: It works because I have an edge. Noo... It works because you handle the trade at the moment and more important your risk/reward skews. "Put me in a position and I will make it a winning position" should be the book that every trader must read.
    #23     Apr 14, 2019
  4. I understand what you are saying, and it's simply wrong. You may may money trading this way, but you don't fully understand what you are doing.

    What you think: I can't predict the future because I have no edge, but if I place a trade here at 2:1 R/R I can make money.

    What is actually happening: Your 2:1 R/R trade implies a belief about the probability distribution of future outcomes of price movement. Specifically the belief is that your trade has positive expectancy given the distribution of all possible outcomes. Having positive expectancy is the definition of having an edge.

    Assuming that your goal is making money and not gambling, the very fact of you placing trades implies that you believe you have an edge, even if you don't understand what the word edge means. Then, after the fact, whether you make money or not depends on 1) Luck and 2) Whether you actually had an edge or not. Disentangling these two things in hindsight can be difficult, depending on the statistical significance of the results.
    #24     Apr 14, 2019
  5. What do you mean specifically by money handling? Optimizing the exit?

    I haven't yet optimized the exit. My testing assumes that I sell in 30 days, unless I'm stopped out.

    The first books I read on rules-based trading were written by Kevin Davey. Not sure what the group here thinks of him, but his views made so much sense to me. Must be something common to nerds who code in fortran (which he did!). Anyway, he wrote that he knows successful traders who ONLY optimize the exit (in other words, give them any position and they'll profit). He also knows other successful traders who do ZERO optimization of the exit!

    I'm certainly open to optimizing the exit. I just haven't done it yet. It seems like an optimization problem which is every bit as involved as optimizing the entry...yet I'm sure that there are many exit strategies that are far better than simply selling on day 30.
    #25     Apr 14, 2019
  6. Indeed. Optimizing the exit is a crucial aspect of a trading system. You need to consider the question "at what point do I no longer want to have this position?" The answer to that question will lead to rules related to when to close it. Your entry rules are not based on the time or calendar. So using "30 days later" as exit rule seems not suitable to me, as it is in no way related to your entry rules.
    #26     Apr 14, 2019
    morganpbrown likes this.
  7. danielc1


    I do understand positive expectancy and maybe I'm wrong about it how it is over rated by many traders that are searching for that edge in the wrong place of their positive expectancy system, namely the entry of their system. I can not put a number on it, but I think it will be 5 or maybe 10 percent of your system and how you would make money by timing the entry. Biggest part is knowing how not to lose or how not to lose so much, when you are wrong. It is more about exits, risk management and how you handle the trade when you are in it. I will even go further, that will chill many traders bones: Positive expectancy systems created with past data is a poor mans way of being a trader. It is settling for the crums of what a market can bring to you. When I figgered that out, I did not even care anymore about my entry.
    #27     Apr 15, 2019
  8. Edge encompasses the whole trade, not just the entry. If you are such a good trader that you can be profitable with random entries, that just means you have an edge all the time (on average). If that is true, congratulations because you will be a billionaire very soon if you increase your trading enough.

    You also seem to conflate edge with risk management, as "knowing when you are wrong" is just a change in expectancy as new data is revealed. At each point in time during a trade you are in the same situation, you need to have some idea what your edge is in order to make rational decisions, otherwise you are just gambling.

    What is the difference between what you can do in a trade, and what a system can do? Do you believe that "systems created with past data" are not able to consider the whole path from entry to exit? I assure you that they can.
    #28     Apr 15, 2019
    morganpbrown likes this.
  9. You know...the exit for my approach could simply be the entry. In other words, if I buy a security because the indicator signal flashes positive, then I could simply sell that security if the indicator signal flashes negative, rather than holding the security for 30 days.

    If it's not good enough to buy more, then why would I keep holding it?

    Will test and report back.
    #29     Apr 15, 2019
  10. danielc1


    Haha, I wish... increase trading enough is also increasing risk to a level that I'm not comfortable with. Some examples of trading process in my world: Let's say I want to own an index. I have several choices to do this. I can buy etf, future, options and so on. I can split orders over a period of time, I can buy all at once, and so on. When I own it at whatever price, I have again several options what I can do. I can add, sell, write options, buy options, add future contract, go short a future contract, look for inverse possibilities or non correlations, and so on. That is a decision I make when I'm in the trade. How much I risk is before entering the trade. Now when you risk is based on owning it and not on movement of the underlying, you get in to a whole other ball game. My goal is getting the cost of ownership down over time. I'm not putting down other ways of trading, but I do believe that many people make it difficult with pin pointing when they are going to enter and exit, based on past data. My exit is based on return of capital X investment duration.

    I'm much further away from gambling then any other rule based trading system.
    #30     Apr 16, 2019
    morganpbrown and nooby_mcnoob like this.