My losing, forward-tested Forex strategy (charts and equity curves) - let's fix it!

Discussion in 'Forex' started by 1a2b3cppp, Mar 30, 2011.

  1. I've been dicking around in an Oanda practice account for the last month or so trying to play around with some "catch big trend" strategies (which I still have no success with).

    The premise and rules for this one are easy, and are as follows:

    Direction cannot be predicted (except by ET MACD millionaires who try to sell you their shitty courses via PM), so just make sure you're "on board" when price happens to take off.

    Trade 6 pairs. Which 6? The ones with the lowest commissions.

    Open two sub-accounts or whatever they're called. One is for long positions, one is for short positions.

    Begin whenever you want (although make sure it's at a time when commissions are reasonable, so not when they're 10 pips).

    Go long all 6 pairs in the long account and short all 6 pairs in the short account. Buy position sizes such that 1 pip = 1 dollar (I like to keep it small, but you can do whatever you want). Each order gets a trailing stop of 50 pips.

    Eventually your positions will close. The rules for reopening them are as follows:

    Only reopen a position that has closed on both the long and short side. In other words, if EUR/USD is still open on the long side, don't reopen it on the short side. But if both sides have closed, then reopen both of them.

    Reopen them whenever you want, just make sure the commissions are low. In other words, if it's like 5-10 pips to open one, don't cuz that's already eating into your profits.

    The reason for this is that price will eventually take off in one direction (which cannot be predicted), and when it does, the losing side will close out and the winning side will ride to huge profits!!!!

    What usually happens:

    There are three things:

    What happens occasionally:

    Price just chops around endlessly and doesn't go anywhere. I've been in the same ZAR/JPY trade for like 2 weeks now and still neither side has closed out.

    What happens often:

    Price starts to take off, closes out one side and gets you a good +25 pips or so on the other side, and then retreats back in the initial direction, so you get a -25 pip loss on the good side and a -50 pip loss on the bad side.

    What happens rarely:

    Price takes off and your losing side closes out and your winning side makes bank. Like this:


    I'm up like 148 pips or something on that trade.

    The results:
    After almost 4 weeks of this, with both accounts starting with $10,000 the account balances are as follows:

    Long account:
    Balance: $9,943.52
    Unrealized P&L: $210.20
    Realized P&L: -$67.26

    One thing to keep in mind, this account currently has 6 positions open because I just reopened them all (except for USD/JPY which has been rockingever since 3/25 when I bought it).

    Short account:
    Balance: $9,682.33
    Unrealized P&L: $8.41
    Realized P&L: -$298.58

    This account has 3 positions open: EUR/GBP, AUD/USD, and USD/CHF.

    Its balance has been steadily declining ever since I started, with occasional.

    Equity curves:

    (both accounts started at $10,000, but for some reason I don't have that data anymore, so just assume that wherever they started, it went down to that level roughly at the same level as the decline shown on the chart from the starting level down to the current level)

    Long account:


    Don't get all excited just cuz the equity cuve is going up. Remember it started at $10,000, and also remember that there are 6 positions open right now, most of which will probably close at a loss, and even the winning +150 pip position right now will likely close at only +100 pips (since there's a 50 pip trailing stop).

    Short account:


    So net starting value was $20,000, current value is like $19,625.

    If I keep doing this, I will update the thread every so often with account balances and stuff.

    But overall it's a losing strategy (and would be even without commissions).

    Any ideas for how to make it better?

    I mean real ideas, not indicators or fib voodoo.
  2. cvds16


    imo this doesn't even come close to real trading, it's more like blind gambling with some obfuscation of two subaccounts (wth is it with this both going long and short at the same time ? that's the equivalent of no position at all).
    Learn price-action it's the only thing that really works from all the people I know that make money short term trading ... you'll need lot's of screentime though to get really good at it (some people add one indicator to it for more subtleties, but even they basically trade pa) ... I never listen anymore what the news is or try to interpret that (more clever people then myself will do that for me; I'll see it in the PA), the only thing I have written down on a piece of paper the day before is the time when big news comes out, I just make sure I am flat then and don't get back in untill at least five minutes later. Trading is basically simple and it's got to be in tune somehow with your personality, but it only took me several years to find that out and turn that into something consistant lol don't expect to get there fast, it's only after tons of screentime that the fog will lift.
  3. Sure, if you close both positions at the same time, then yes, you are right.

    Anytime being long and short at the same time gets brought up on this forum, people are always like "ZOMG that's teh same thing as no positon at all!!!"

    If it was equivalent to no position at all, then the (net) equity curve would just be a smooth downward line, declining at each data point by the amount paid in commission.

    Oh that reminds, here's the net equity:


    I think the missing data is because Oanda only keeps a certain amount of transactions on record on the demo accounts. When I scroll back to the beginning it should start at $10,000 but it doesn't cuz the first few weeks of data are missing. So that's why there's kind of that gap there.

    This strategy was based around the same idea as straddles, except without the time decay. You know price will eventually go in one direction, and hopefully it goes far enough in that direction to overcome the loss in the other direction.

    Well at least you didn't try to sell me something or tell me to use fibonaccis :D

    So you're saying the only way to trade profitably is to be able to predict direction?

  4. Direction is the only way to make money- you need to have a good prediction either "it" is UP, DOWN or SIDEWAY.

    "it" could be price, volatility and etc.
  5. nLepwa


    No need for the two accounts.

    The strategy works but the edge is very small.

    You need to keep comissions and slippage under control. The strategy is almost impossible to trade for retail.
    You need to diversify accross a lot of assets (200+) and maximize capital utilization accross the assets.. It works better if you don't re-enter right away but use a buffer zone in which you don't trade. You need to adjust the buffer size according to volatility.

  6. cvds16


    your position is like having a stop-entry away from price, so yes untill that point it's basically NO position, if you think this is anything like a straddle you really need reading up on options (but seeing your questions about options that's nothing new to me :D ).
    This is all about direction: no free lunches. Even stradle trading is about direction: the direction of vol (who do you think is on the other side ? someone giving away money for free ?) , but I guess that's beyond you here at this moment. You are really have to going to try harder than you are doing now if you want this to work.
    Like 10 years ago there were 'almost' free lunches available in options but you had to know your math and be prepared to put lots of time into it. It was like trading in five dimensions for me. My advice don't go there cause they have totally disapeared, it won't be worth it. Just go for straight simple things.
  7. This system is a higher-cost equivalent of starting flat and setting two stops to enter 50 ticks away, each with an attached trailing stop to exit.

    This sort of system has been around forever and it used to work - the Turtles range breakout system worked on a similar principle. But over the last 20-30 years all these systems have slowly stopped working. Since it's the most trivial form of technical system to implement, everyone's got one and there's too much incentive to hunt all the stops they generate. In other words, from a MM perspective there's too much incentive to generate chop now.

    Furthermore, your example of this class of strategy isn't even a good example of the breed - you're picking your stop positions arbitrarily, your trailing stop distance arbitrarily, and what time you're in the market arbitrarily. None of those are best practices.

    All in all I would have been amazed if you'd been successful.
  8. cvds16


  9. cvds16


    to be sure you really get it: you have to take a directional bet.
    btw, you could easily ask a demo of (you can get one for 7 days for free the last time I heard) and backtest this strategy like described above: entry away from current price and trailing stop. I have tested mechanical things like this in hundreds of ways in the past, I know other people who have done the same thing too (one of them a professional programmer) the end result is always the same: the edge is really not worth the effort. Only discretionary trading with interpreting will provide a real edge, combine that with tight stops and letting winners run.
  10. Actually, yeah, you're right. It is equivalent to waiting for a 50 pip movement in one direction and then entering.

    Except I don't want to sit around waiting for that to happen nor can I write code to do that for me, so I just entered both directions and let it do its own thing on its own time.

    Can't chop be eliminated by going to higher time frames but still trading the small size? Maybe bigger stops are in order.

    I mean, ok, obviously there is chop everywhere, but what chops you out on a 5 minute chart is nothing on a 4 hour chart... but then that 4 hour chart has its own "chop." Charts are fractal and you can't tell the timeframe of one just from looking unless it is labeled.

    Operating under the assumption that price is random (or at least cannot be predicted), there's no point in trying to figure out the best entry times or stop distances. Random.

    If you disagree with price not being predictable, then you are correct that it would make sense to try and optimize those things.

    Me too :D
    #10     Mar 30, 2011