Saturday, March 17, 2012

What I Hate About System Trading

For any trading technique, there are disadvantages to it.  For example, if you trade in a discretionary fashion, your mood (state of mind) might influence your results - if your last 3 picks were losers, you might be hesitant to pull the trigger on the next trade.

For that reason, many people like mechanical trading (where you just follow the rules) better.

BUT, I have never heard anyone talk about the main drawback of mechanical systems: HELPLESSNESS.

Let me explain, using my recent equity curve as an example...

Before the contest started, I spend quite a bit of time developing a system, running simulations, evaluating performance, etc.  During this time, I was doing a lot of thinking, trying to develop (but not over-develop) a winning approach.  I was in control of the process.

Once I was done, and the contest started, I basically just follow the rules.  Sometimes the rules work (and equity goes up) and sometimes the rules don't work (and equity goes down).  In the short run, anything can happen - the market is in control.  But over the long run, I expect my system to prevail.  The key point here is that once the system is running, I am not thinking too much, and I am NOT in control of the process (unless I completely shut the system off).

Once the system is turned "on," the market is in control, and I am helpless.  That is the way I have felt the last couple of days - helpless.  All I can do is follow my system.  Intervening/changing the rules/making emergency "fixes" may reduce or eliminate my helpless feeling, but I can guarantee that in the long run it will hurt my performance.  Whenever I tried that approach, I always ended up worse off (so I do not do it anymore).

I have young children, and I suppose system development is similar to raising children:  you do your best to set the child (system) on the right path, but one day you have to let the child (system) go and face the real world (market) on their own.  If you did things right early on, the child (system) will thrive.

So, right now, I feel a bit helpless.  But next week, I may feel like I am back in charge.  Or not.  That uncertainty is just part of the trading game!





11 comments:

  1. In Michael Covel's book on trend following, he said that a trading system is like a piece of clothing, neither too loose or too tight.

    ReplyDelete
  2. Can you explain? I'm not sure what this means.

    ReplyDelete
  3. This is another analogy of trading systems, in contrast to them as children. I would think that children are self-learning, they can override and improve on what they have been taught.

    A tight-fitting system to historical data has results that are 'too perfect' and the wearer ('mr market') would find it too uncomfortable. A loose-fitting system has too much room for the wearer and benefits for the system user is lost.

    ReplyDelete
  4. In my limited system development experience, I noticed that when I back-tested trading system with historical data of 10 years, the results might be positive expectancy overall. But, when I look at the trades year by year, there are bound to be negative return years.

    "In the short run, anything can happen - the market is in control. But over the long run..."

    Regarding the short run and long run, what is the definition of short run and long run? For eg, short run = few days?

    ReplyDelete
  5. No set definition, as it really depends on the system. You could have a positive expectancy system, but if the standard deviation of the trades is high, it might take 30-50 trades or more before it could be considered "long run."

    Looking at my total accounts, I many losing months, so for me a month is short run. 6 months to a year is probably "long run."

    If my numbers are like many others, it is no wonder people lose. Likely many give up after 1-3 months, when their positive expectancy system doesn't show a profit.

    ReplyDelete
  6. After you got the standard deviation, how do you interpret the standard deviation? What kind of numbers would tell that it is high and what kind would be low? Are you able to show the standard deviation for the system you using for this contest as an example?

    ReplyDelete
  7. Unfortunately, I don't have any hard and fast rules for std deviation - I use it more as a qualitative measure, rather than quantitative. A big std deviation means high variability of results, which means you need a long time to evaluate it properly. That was my general point.

    Maybe another reader uses std deviation, and can share with the readers of the blog.

    ReplyDelete
  8. Broker report system was down last night, so I will post either late tonight or first thing tomorrow AM. Sneak preview: back in positive territory.

    ReplyDelete
  9. Wow, that's nice, considering the choppy actions for past few days.

    ReplyDelete