perhaps an overly simplistic comment. What would happen if everone coded their favorite subroutine for Buy / Sell and returned it to the program as a bias rather than a random number?
that is to say replace this line
int Dir = MathRand()%2;
int Dir = MySignal();
if (--some condition--) return(OP_SELL);
if(--opposite condition--) return(OP_BUY);
I believe most programmer's on here favorite sub-routine depend heavily upon their Trade-Manipulation and Lot-Size-Manipulation. But you're more than welcome to try and post an equity curve if you like, you don't even have to post your code. In my experience tho, when you force most signals to a define sl/tp and flat-lotsize they tend to produce result similar to random the more data it's tested upon. Forcing signals to a set sl/tp is the best way I can think of in gaging if the initial forecast is accurate.
If the system's depending on the fact that the market cannot go in one direction forever and there-fore taking advantage of that... I wouldn't consider that forecasting. Rather a form of statistical arbitrage like what Zzuegg tried to do. Also peggy back systems like grid, martingale, pyramid etc... would not be considered independent events. They're basically using the results of the initial orders as a form of indicator.
Well those are my Opinions anyways. :)
I find this approach very nice: it is the same proffesional approach
that casinos and brokers use against their customers. Just a few things
1) Have you tried finding a edge using simple indicators? -51% is enough-
- What if you only launched positions in the direction pointed by the moving averages? (ie, ma10, ma30 and macd)
- What if you used Stochastics and RSI to filter trades?
2) Regarding money management and martingale systems
Martingales are dangerous and you will end up bankrupt if you use it. But what if you used a fractional inverse martingale?
This is, for each winning trade you would reinvest 25% or 50% of the
profit in the next trade, and so forth. Winning streaks will make you
money very fast on the house, and losing streaks will lose the average
bet. This gives you a edge against the market and with only 51% winning
trades you would rip the broker apart in notime.
3) Further thoughts on reverse martingales and trading systems
I'm actually thinking that the best way to develop a profitable EA is
to think of an out-of-proportion stupid system that winds up busting
your account: the less winning trades you get, the better. For instance:
go long every time the high of the previous bar is reached: TP: 15. SL:
80. This system will eventually lose all your money in a straigh down
line. Do you see it? Nice! Now, do the opposite and apply a fractional inverse martingale forcing the broker to trade with that stupid system against you in a compulsive gambler fashion.
Go short every time the last high of the previous bar is reached (SL:
15. TP: 80) and apply an inverse martingale every time you win. I will
try this soon and let you know.
You would do better hiding SL and TP from the broker using this scheme, it should have the direct opposite results than the initial losing system.
I just want to show you a backtest I did with a very simple EA that uses Bill Williams TradeZone2.4 indicator to add to positions if they move in our favour.
Very simply: go short on any red candle, and long on any blue candle. Add another position if it closes above last high or low. Nothing else. Trail stop 2candles.
Reinvesting the brokers money is a good thing. I haven't done an inverse martingale yet but it looks promising indeed. I'll share when finished.
I messed around with Ubzen's very simple template to exploit something that I thought should and always exist in any market.
Pretty nice. Did you use Strategy-Optimizer for your strategy?
Did you run the above with random trades or an algorithm?