A kitchen-sink approach to support and resistance

[Wappers2]

I remember knocking something up like this with the help of a bond trader buddy way back when. The system got *immense* and basically boiled down to a multi-dimensional matrix of combinations that (we hoped) would reveal extremely high probability opportunities based on the statistical analysis of years of data for each indicator.

ooof. we ran it for a while on most of the major indices, and,... basically it didnt work. It was a very pretty display tho - we put it into a 'chequer-board' pattern to make it easier to read.

Personally, I suspect the reason it failed was that the combination of probabilities of multiple indicators was not being calculated correctly (e.g. if indicator A = 75% accurate, and indicator B = "90% accurate", if A AND B together, is it 75% x 90% = 67.5% or is the probability of the combination NOT being correct 1- 75% * 1-90% = 2.5% (i.e. its 97.5% ON!))
Harry Trader? Any ideas for this guy?

 

[chaos]

Quote from Wappers2:

I remember knocking something up like this with the help of a bond trader buddy way back when. The system got *immense* and basically boiled down to a multi-dimensional matrix of combinations that (we hoped) would reveal extremely high probability opportunities based on the statistical analysis of years of data for each indicator.

Hi Wrappers, thanks for your input. When you say immense, are you referring to the coding, data or both?

A currency trader I know has created a program along the lines of the one I have described. Although he hasn't disclosed to me the exact specifics of his system, he uses many of the trend indicators i mentioned in my first post. These combine to give him significant positive expectancy, and the system is so robust he hasn't had to so much as tweak it in over half a year.

From my research, it's apparent to me that many other successful currency traders have developed similar mechanical systems. Perhaps their success in this is due to the unique nature of the currency markets - when the underlying is as huge and complex as an entire nation state, it's no surprise that trends tend to be exaggerated and drawn out.

Regards,

Chaos

 

[chaos]

Quote from dbphoenix:



I assume you mean a change in direction from directionlessness. If a trend is in place, it can change direction to sideways. From here it can continue the previous trend or it can reverse. "Change" should not be equated with "reversal".

--Db

Hi DB, as always, you make a very good point.

By the way, your keeping it simple thread has taught me a great deal about trends and breakout trading. Thanks for all your hard work. Just a thought - you might want to consider adapting your method to trade the US dollar crosses between 7:00 and 9:00 AM when the U.S. market gets going for the day. If you haven't done so, take a look at some charts. The price fluxuations are incredible.

Cheers,

Chaos

 

[Wappers2]

Quote from chaos:



Hi Wrappers, thanks for your input. When you say immense, are you referring to the coding, data or both?

(snip)

It was only a spreasheet dude, but it started to turn into a 3 megabyte calculation, which made it a little unwieldy...

The question still remains, if statistically, the prob of indicator A being right = 90%, and the prob of B = 80%, does the simultaneous firing of both indicators mean that the chances of them BOTH being wrong = (1-A)*(1-B) ????

Damn, glad I daytrade sometimes...

 

[maxpi]

Quote from dbphoenix:



I assume you mean a change in direction from directionlessness. If a trend is in place, it can change direction to sideways. From here it can continue the previous trend or it can reverse. "Change" should not be equated with "reversal".

--Db

I was just half-joking, but still, every trend starts and ends with a change in direction, could be from up to down, down to up, sideways to up, sideways to down. A change of direction would not indicate the start of a trend, but every trend has one. If you are quantifying a trend after the fact you would id the start and the finish. There could be changes of direction in the middle of a trend from up to sideways or down to sideways and sideways to up and sideways to down but that would really be the end points of a minor trend and a consolidation during the major trend.

Max