T/A Fails Time and Time Again - Why?

[NoVoodooHere]

I am finding this thread quite entertaining, yet puzzling, and, also, to a lesser extent, disturbing

https://en.wikipedia.org/wiki/True-believer_syndrome

 

[kut2k2]

I think that when time is eliminated from the display of market data we find that the data is much more stable and reveals itself nearly a constant of definable motions. Those motions cataloged can perfectly define all market movement with no surprises.

Now applying math to the data we can get constant and even predictable results.

Sounds like a unicorn hunt to me. Good luck.

 

[eurusdzn]

Mr. Brown, being an amateur without a strong math background, generally speaking, can you elaborate and possibly give a concrete example.
I am thinking of unordered daily return data over many years that does well with a Monte Carlo run.

 

[MarkBrown]

Mr. Brown, being an amateur without a strong math background, generally speaking, can you elaborate and possibly give a concrete example.

market data based on time intervals is infinite in its variability. how can a one discover reoccurring anomalies when applying mathematical indicators. Tommy can't be fitted for pants because he is constantly growing out like expanding foam shot from a can of weather sealant.

so now lets think of applying indicators to a well defined assembly of limited data sizes. Tommy the plant supervisor has declared that all widgets will no longer be made to order. widgets will be produced in 10 common sizes only. sales analysis will show what the most popular widget sizes are and production will shift to meet that demand.

 

[eurusdzn]

Thanks , and that analogy is understandable. But , still , the analogy seems to make good sense
with or without "time" as a factor.

 

[i960]

As a strong believer and user of volume-based charts I can assure you there is no lack of surprises when time is removed from the picture. Not a Grail.

 

[Mysteron]

I am an advocate of technical analysis trading. Continually T/A shows great promise and can have it's run of success which is why it perpetuates the notion that it does work. Then it seems just as you leverage up, it fails and wipes you out. Only to get back in sync and start working again as your account has been drained. I went thru this for many years boom and bust.

Sophisticated math became my pursuit and for years. If the math wasn't complicated to the point of being a classified national security algorithm, then it was not worthy enough to pursue. That was the thought process and having exhausted all those avenues as well it was still hit and miss. Make no mistake we are not just talking about obtaining and edge to be profitable but seeking perfection of understanding price movement.

Luckily some of the discovered algorithms had runs of profitability which helped to finance research and keep the discovery process alive.

Finally the culprit was discovered, the one ingredient which kept poisoning the process. Math itself is thought to be perfect, then why when applied to market data did it fail. The answer is that the math was being tainted with the input of time. Consider taking the time out of the data and now math can do it's job consistently.

Mark

Taking the view that significant price changes and new trends often start as a result of events whose timing in known in advance then it doesn't make sense not to include time in a trading strategy, whether it be discretionary or automated.

 

[MarkBrown]

Thanks , and that analogy is understandable. But , still , the analogy seems to make good sense
with or without "time" as a factor.

looks like i assumed that it would be understood that Tommy stopped making custom widgets because his widget company was going broke satisfying each individuals desires.

Tommy raised prices and customers stopped buying, yet demand was there for a breadth of standard products which met the needs of most his customers he found.

Indicators are chasing around trying to satisfy traders who are applying them on every time based bar size imaginable. It must be maddening to the poor indicator when he then gets the blame that T/A does not always work and often fails.

 

[MarkBrown]

Taking the view that significant price changes and new trends often start as a result of events whose timing in known in advance then it doesn't make sense not to include time in a trading strategy, whether it be discretionary or automated.

in the realm of trading i can only answer this by saying in my opinion math is a gift from god and time is a destructive device set in place by someone else with bad intentions.

 

[MarkBrown]

It's not time, it's the math. Time is just the dimension in which change takes place. And the markets are constantly changing. Unless the mathematics is designed to be adaptive to market changes, it will eventually fail no matter how sophisticated it is.

consider this math is a rock steady constant it will always be the same. to make math adaptive is to spoil what is natural and more valuable than the data it's applied to. let's consider making the data adaptive to our desire rather than expect the math to adapt.

data with time is like concrete without forms it does what it wishes in the time interval observed. we all know that concrete needs to be poured into forms which define it's dimensions irregardless of what the concrete wants to do on it's own.

why then do we allow data to do what it wishes without constraints and expect math to adapt?