is this EA back test results any good? - page 3

Ubzen
5300
Ubzen  
Man, from what I learned from card-counting. "Hang in there". Sigma (standard deviation) does not favor the brave it favors the patient. If you're playing a winning game, then its not if you're gonna win but when you're gonna win. Easy for me to say now ... but don't let money bring you down. If this is your first experience with watching money fluctuate then I say develop a risk-world-view. Become James-Bond (or something), have fun with it but keep your Standard of Living out of it.
zzuegg
1503
zzuegg  
ubzen:
have fun with it but keep your Standard of Living out of it.
true true
Ubzen
5300
Ubzen  

@Zzuegg, thats why I said 98% ;). A trend have to develop before you can say it's a Trend IMO. For this reason u as well as I have searched for predictive methods. Also, just to clarify, by trend I mean price-action trend and not pattens which repeat themselves.

My grand-mother can spot a trend. Once you do, the question becomes how long would it last and how much should I risk. If you risk too much the intra day movement would put you in Over-betting risk exposures. If you risk too little, you're not gonna make much, if and only if it lasts. The best systems in the world would only ketch 50% of the trend move. The part that says oh-yea it's a trend and oh-no the trend is over cannot be determined by technical analysis alone.

But don't think fundamental analysis is all heaven and is less useful for entry/exits than even TA. IMO.

Added: I'm not a Mathematician or anything. But I'll guess even a randomly generated chart would show signs of what we call Trend. Like clouds creating faces.

[Deleted]  

No;) That's what I thought. when you try it though, you get randomly generated nothing:)

Thanks for the kind words.

mbirrell
232
mbirrell  
I am going to keep trying and I refuse to give up experimenting with systems. I believe if you make lots of small trades based apon statically proven patterns the odds move to your favour. The trick is to find that pattern in price movements. I believe a profitable system would have trader intervention most likely as the market is very complex in the way of infinite possibilities that affect price - some of which are not predictable - e.g. tsunami as we have seen recently. This is why it can't be predicted 100%. I also believe the longer you are in a trade the more exposure you have to risk. If I could predict the movement of 3 bars that would be enough for me to make a living off.
[Deleted]  
lol having a highly probable solution for 3 bars in advance (we talking about any timeframe?) we'd be bloody rich:))) if you really figure something out, i promise to help with tehcnical problems, as probably everybody else here.
zzuegg
1503
zzuegg  

Systems based on statistics by definition have the same characteristics as overoptimized systems with the advantage that the adapt automatically.

In theory you should gain a nice edge by combining statistical informations such as how many candles goes in the same directions as the previous. (the attached indicator does that)

If the value is higher then 0 you should trade only in the same direction as the previous candle, if it is below 0 should countertrade. In theory, by analyzing all timeframes simultaneosly the probabillity should increase, and so does you edge. However, as good it sounds, it never worked for me. Maybe by adding statistics about the size of bars in each direction allows you to extract good stoploss and takeprofit values.

//z

Files:
Ubzen
5300
Ubzen  

zzuegg... However, as good it sounds, it never worked for me.

Did it not work live or in the back-tester?

zzuegg
1503
zzuegg  
Oh in tester it works profitable, but also not pretty good. On forward demoing it was negative on average
mbirrell
232
mbirrell  
forexCoder:
lol having a highly probable solution for 3 bars in advance (we talking about any timeframe?) we'd be bloody rich:))) if you really figure something out, i promise to help with tehcnical problems, as probably everybody else here.
I bet you would.... the problem is I wouldn't need help to code it if I figured it out.