Technical analysis :useless junk science

[ssrrkk]

I would say this: you haven't really optimized the MAs for this particular scenario. since there is no cost in stuttering in and out of a trade, it is better to make the MAs closer in duration and shorter. Then what happens is when the price takes a big trend, you capture most of those big trends as profits. But you don't pay when the price moves sideways, causing you to enter and exit with no profit. Also by making the MAs shorter, you will have less of a chance that an entry/exit results in a loss. Here I would agree with you that less lag will be better because of this tracking error. But simply making the MA periods smaller will help you here.

Quote from jcl:

At first, I made a mistake with the code, I got the crossover the wrong way around - sorry for that. Here's the fixed code, with the MA plots:

function run()
{
	Spread = 0;
	Slippage = 0;
	Commission = 0;
	RollLong = 0;
	RollShort = 0;

	DataSplit = 80;
	
	var *Close = series(priceClose());
	var *MA1 = series(SMA(Close,optimize(100,10,1000)));
	var *MA2 = series(SMA(Close,optimize(300,30,3000)));
	Stop = 2*ATR(100);
	
	if(crossOver(MA1,MA2))
		enterLong();
	else if(crossUnder(MA1,MA2))
		enterShort();
	
	plot("MA1",*MA1,0,RED);
	plot("MA2",*MA2,0,BLUE);
}

And here's the chart:

As you see, the fixed version ends indeed with a win, not with a loss, but it is obviously not a profitable strategy - the drawdown is larger than the profit, and the positive result was just random. The Sharpe Ratio is 0.39.

 

[ssrrkk]

Or another way to reduce tracking error is to do like my example: don't use two SMAs, but simply use the even more primitive method of a single MA crossing over with price. That creates the beautiful profit curve I have attached earlier because of reduced lag.

Quote from ssrrkk:

I would say this: you haven't really optimized the MAs for this particular scenario. since there is no cost in stuttering in and out of a trade, it is better to make the MAs closer in duration and shorter. Then what happens is when the price takes a big trend, you capture most of those big trends as profits. But you don't pay when the price moves sideways, causing you to enter and exit with no profit. Also by making the MAs shorter, you will have less of a chance that an entry/exit results in a loss. Here I would agree with you that less lag will be better because of this tracking error. But simply making the MA periods smaller will help you here.

 

[oilfxpro]

Quote from d08:

You really lack any kind of logical thinking.
If someone has something that really works, they will NEVER sell it (unless they're insane).
What you're suggesting is as if Apple should sell iPhone manufacturing rights along will patents for pennies - they're not insane so they manufacture it themselves and hold on to the profits.

Apply logical thinking yourself , a software is held on servers with security and everything, licensed software users provide their account connection details with brokers .The licensed user logs in and sets all trading parameter on software via interface,without access to the codes.The software manufacturer keeps the software secrets. profits and enjoys license fees annually like apple or Microsoft.

By your statement , most of the software sellers are selling scams.Just admit it.

 

[oilfxpro]

Quote from ssrrkk:

You seem to be missing the point. I am not arguing that delays are irrelevant or whatever. I am saying that simple cross-overs don't work when you add commission and slippage. But they actually do in the idealized world where there is no cost when you stutter step and enter in and out of a trade because of high frequency fluctuations and delayed entry exits. It is so easy to confirm this if you just ran it. Just run without costs and a simple cross-over is profitable. The results are so robust, almost any simulator or R-script can reproduce this.

The secret is to have m/a filters with trend lines etc, and to let profits run , cut losses and re-enter trades.This can not be done with fixed tsl and most of this can't be coded to work perfectly.

 

[jcl]

Quote from ssrrkk:

Or another way to reduce tracking error is to do like my example: don't use two SMAs, but simply use the even more primitive method of a single MA crossing over with price. That creates the beautiful profit curve I have attached earlier because of reduced lag.

I've tried to replicate your algo, but I get huge losses with S&P500. Do you use a 1-minute timeframe? Is this your algo?

function run()
{
	Spread = 0;
	Slippage = 0;
	Commission = 0;
	RollLong = 0;
	RollShort = 0;

	BarPeriod = 1;
	asset("SPX500");
	
	var *Close = series(priceClose());
	var *MA = series(SMA(Close,30));
	
	if(crossOver(Close,MA))
		enterLong();
	else if(crossUnder(Close,MA))
		enterShort();
	
	plot("MA",*MA,0,RED);
}

I'm no R expert, but maybe the difference arises if you calculated the profit from the crossover point, while Zorro uses the bar after the signal. This makes a huge difference when trading on extremely short time frames, such as 1 minute.

 

[ssrrkk]

I use one minute data intraday. One thing I don't understand is when you enterLong, are you closing your short and entering long at that point? It seems you might need to double enter long or short when you flip to the other side. I am not sure if you are doing this correctly in your sim. Also I take care to close out any open trades each day at 3:58 PM (this is a stock market sim). If you look even closer, I even enter a long at the high of the bar, and enter short at the low of the bar, making it less favorable for me. The idea is that this considers some slippage. So I actually included some slippage effects here.

Quote from jcl:

I've tried to replicate your algo, but I get huge losses. Do you use a 1-minute timeframe? Is this your algo?

function run()
{
	Spread = 0;
	Slippage = 0;
	Commission = 0;
	RollLong = 0;
	RollShort = 0;

	BarPeriod = 1;
	asset("SPX500");
	
	var *Close = series(priceClose());
	var *MA = series(SMA(Close,30));
	
	if(crossOver(Close,MA))
		enterLong();
	else if(crossUnder(Close,MA))
		enterShort();
	
	plot("MA",*MA,0,RED);
}

 

[ssrrkk]

Quote from jcl:


I'm no R expert, but maybe the difference arises if you calculated the profit from the crossover point, while Zorro uses the bar after the signal. This makes a huge difference when trading on extremely short time frames, such as 1 minute.

Oh yes, I don't use delayed entry exits. As I mentioned, I use the low of the same bar for short entries and long exits (sell low, which is unfavorable), and I use the high of the same bar for long entries and short exits (buy high, again unfavorable, but more realistic).

 

[jcl]

Quote from ssrrkk:

I use one minute data intraday. One thing I don't understand is when you enterLong, are you closing your short and entering long at that point? It seems you might need to double enter long or short when you flip to the other side. I am not sure if you are doing this correctly in your sim. Also I take care to close out any open trades each day at 3:58 PM (this is a stock market sim). If you look even closer, I even enter a long at the high of the bar, and enter short at the low of the bar, making it less favorable for me. The idea is that this considers some slippage. So I actually included some slippage effects here.

Yes, enterLong closes any open short trades before entering the long position. Trades are always entered and closed at the begin of the next bar as long as the slippage is set to zero. But I'm trading 24 hours here, maybe that's the difference - I'll make later another test with trading only during NYSE opening hours.

Quote from ssrrkk:

Oh yes, I don't use delayed entry exits. As I mentioned, I use the low of the same bar for short entries and long exits (sell low, which is unfavorable), and I use the high of the same bar for long entries and short exits (buy high, again unfavorable, but more realistic).

Doesn't this cause peeking bias when your signal is generated from the close?

 

[d08]

Quote from oilfxpro:

Apply logical thinking yourself , a software is held on servers with security and everything, licensed software users provide their account connection details with brokers .The licensed user logs in and sets all trading parameter on software via interface,without access to the codes.The software manufacturer keeps the software secrets. profits and enjoys license fees annually like apple or Microsoft.

By your statement , most of the software sellers are selling scams.Just admit it.

It's not hard to reverse engineer any system with enough time and a large sample of trades. You don't need to see any of the code to do it and you cannot patent or protect a trading system at all.

If someone has something that works and works well, they'd never sell it. By definition, buying a system is buying something that doesn't work well. And by working well, I mean having significant edge.

 

[ssrrkk]

No it assumes instantaneous entry i.e., precisely no slippage. If you notice, the opening price of the next bar is usually equal to the closing price of the previous bar. And for those bars that show a difference, the distribution of differences is centered at zero, i.e., there is no systematic tendency for the open of the next bar to be up or down from the prev bar close. So if you enter instantaneously, i.e., no slippage, then you can enter at the close. But like I said, I systematically enter at a worse price than the close and systematically exit again at a worse price so I am making it even more stringent.

Quote from jcl:

Yes, enterLong closes any open short trades before entering the long position. Trades are always entered and closed at the begin of the next bar as long as the slippage is set to zero. But I'm trading 24 hours here, maybe that's the difference - I'll make later another test with trading only during NYSE opening hours.



Doesn't this cause peeking bias when your signal is generated from the close?