Increased or decreased risk with adding strategies and boredom etc.

Discussion in 'Automated Trading' started by sheepsucker, Nov 15, 2011.

  1. Hi!

    This post is intended for those that have one or more strategies making enough trades and providing some profit.

    After I got my first strategy live and making profits, it seems most weeks, I find it extremely hard to motivate myself to add strategies or improve the existing one, until I find some problems with it. Leading me to boredom.

    I tried to code and add other strategies to have them as backup or decrease variance of returns. But instead of decreasing variance they seem to increase variance. Because the markets are so interconnected and when eg. there is sudden increased volatility all markets seem to behave bad. Like in 2008 all markets went wild, which means if you have many strategies it is very likely ALL of them will take a beating at the same time which will make the DD deeper. So risk is actually reduced by just having one active strategy.

    How have you guys solved this. What do you do once you have a working strategy? Try to improve it, create backup-strategies(in case edge would die) or try to add strategies and find uncorrelated ones or something else?
     
  2. gmst

    gmst

    Few things for you to kindly note:

    1. More strategies means more number of trades which naturally means deeper drawdowns. Remember, max DD increases monotonically with number of trades for any one system, so an analogous result will hold for system of systems also, if you are taking trades from different strategies one after another (and not concurrently).

    2. Case II: You are taking trades from multiple strategies simultaneously. Let us assume, you are willing to risk x1% on strategy1 and x2% on strategy2, if you are running these strategies alone. If you are running them as a portfolio in a concurrent fashion, and the strategies are negatively correlated, then you can risk y1% and y2% to get the same DD that you were getting while running either of the strategies (such that y1+y2> both x1 or x2). So, net net theoretically, negatively correlated strategies allow you to run a 'significantly' higher risk compared to a single strategy.

    3. Situation becomes interesting when the strategies are positively correlated. Theoretically, you should be able to run a higher risk still but in this case, y1+y2 might be just marginally greater than both x1 or x2.

    4. To sum the discussion from a practical standpoint, most of your time should go into researching the tails of the strategy pay-offs distribution and identifying what are the underlying reasons for the strategies alpha. The reason is you should know how your strategies interact not only in normal times, but in times of extreme financial stress/euphoria/low vol time/high vol time etc. This analysis will help you in modifying the combined risk on the strategies as per normal times/high vols/low vols times/market stress times etc.
     
  3. Hi,

    Thanks for your post.
    When you say more trades equal larger DD you mean more trades in a specified timeperiod? The underlying reasons I dont quite comprehend.

    If there is an edge and you get to apply that edge more often, the strategy makes more. More trades also allow you to compound faster.

    BR,
    Sheepsucker
     
  4. gmst

    gmst

    yes in a specified time period. pretty simple, more # of trades mean more chances to have losing streaks so sooner or later you will hit a DD which will be larger than your historical max DD

    Yes, exactly, more # of trades with an edge means faster compounding and faster account growth but it also means a larger DD. Hope it helps.
     
  5. A system has an edge and needs a certain number of trades to reach a certain profit.

    With regards to DD it should not make a difference if this number of trades is taken in a short or long time interval? ie. how quickly it makes this number of trades. There must be something I am missing.
     
  6. Bob111

    Bob111

    yep..weeks? how about years? but i have no problem take a good 1-2 hours nap on the middle of the day.then exercise..then drink..no boredom,but too lazy to code something new :p

    http://www.youtube.com/watch?v=puIstclBcO0&feature=related
     
  7. gmst

    gmst

    I am only saying that as number of trades increase (either within a time interval t, or in any time), the system will experience a drawdown > max historical drawdown. A thought experiment ill probably help here: Let us say your backtests have 1000 trades and your largest losing streak (you can put max DD here if you want) was 25 trades. Let us assume that you are going to keep trading this system for next 10,000 years, and it will result into 1 million trades. It should be pretty easy to see that sooner or later you will have a losing streak of > 25 trades. Hope it helps.
     
  8. hehe...
    I think I want to feel like I am doing something useful.
    At least before I have made some serious moula it somehow just doesnt feel right to sleep during the day:)
     
  9. Bob111

    Bob111

    raise a child..buy a house and spend the rest of your time fixing it.:p
    waste some time on ET(that's what i do)

    trading..it's kind of like sex with some girl you use to dream about..you know..
    once the goal was achieved-it's not fun anymore..:p
     
  10. bone

    bone

    I have some clients that take a dozen trades a month, and I have some clients who take a dozen trades a day - and the higher frequency does not necessarily equate to greater net P&L.

    In terms of boredom, I would say that taking a trading on impulse due to being bored or needing action is usually not a good idea and is in fact a really bad habit that can be very destructive. Especially on a Friday afternoon.

    Another really destructive behavior is forcing trades in order to make a certain amount of money. Seriously. Nothing good ever happens when you force a trade, or try to impose your will upon the market.
     
    #10     Nov 15, 2011