Why do I see "Trends" in Randomly Generated Data?

[ReaM]

Quote from epetrov:

Which reminds me that traders must understand and use the market drivers, and using TA in order to "organise" the trades by themselves ??

There are pure TA traders and Pure fundies. From experience, I can't see how them two can mix. You can use fundamentals to avoid certain days, where there is too much fear going on, but thats it. And with TA, you have your entry point and you just enter those no matter what.

With fundamental A you can't use triangles and stuff like that, because you need to enter earlier when the liquidity is there.

 

[intradaybill]

Quote from epetrov:

Trend means by definition non random movement. How trend can be random then?

Good questions that royally describes the current confusion in the statistics filed.

Sure you can get a trend tossing a coin. What is that? Isn't that a trend? Isn't that trend generated by a random process? What is the difference between a trend generated by a random process and a deterministic trend?

 

[ReaM]

Quote from intradaybill:

Good questions that royally describes the current confusion in the statistics filed.

Sure you can get a trend tossing a coin. What is that? Isn't that a trend? Isn't that trend generated by a random process? What is the difference between a trend generated by a random process and a deterministic trend?

There is no difference between a computer generated trend and trends we know from markets. In fact, they look identical.

 

[epetrov]

Quote from intradaybill:

Good questions that royally describes the current confusion in the statistics filed.

Sure you can get a trend tossing a coin. What is that? Isn't that a trend? Isn't that trend generated by a random process? What is the difference between a trend generated by a random process and a deterministic trend?

Here is the point.
Can I tade a "trend" generated by tossing a coin with TA?
Then a second question is which trend I can trade and which not?

 

[intradaybill]

Quote from ReaM:

There is no difference between a computer generated trend and trends we know from markets. In fact, they look identical.

True but you are missing the point completely.

 

[intradaybill]

Quote from epetrov:

Here is the point.
Can I tade a "trend" generated by tossing a coin with TA?
Then a second question is which trend I can trade and which not?

(1) Yes, but your decision will be also random
(2) 7 million $ question

 

[logic_man]

Without reading the whole thread, I would say that so long as you have a specific definition of what a trend is, i.e. what are its quantitative attributes, you will see them in the market. You will also see them in randomly-generated data when that data shares those attributes.

Once you find a quantitative definition of a trend which you can regularly exploit profitably, trade-wise, you will always have trading opportunities. If the definition is generic enough, you can apply it to multiple markets and increase the number of opportunities. As long as you only trade when those quantitative criteria are satisfied, you give yourself a chance at winning the game.

The biggest hurdle, obviously, is finding that quantitative definition because it's difficult to know where to even start to look.

 

[the1]

Tracking 2 std dev's is just scratching the surface in statistical analysis. This is the program I completed and it goes waaaaaay past measuring how far a data set is moving from the mean.

http://www.iit.edu/mathematical_finance/academics/index.shtml#program_requirements

Quote from jem:

I fail to see how statistical trading is different from t/a in many cases.

If I am looking at a couple of pairs and figure they should regress to the me after a two standard deviation move apart. How is that different than plotting their difference on a chart and putting bollinger bands around it.

If I do this on big groups of stocks...

I would have been a very successful hedge fund a decade ago.

And this worked great for daytrading in the 90s as well... same concept on an intraday bases.

Arbing stocks back into line with the correct index on 1 minute or 5 minute charts. I was one of a few dozen traders who made 6 figures a year doing that in up and down markets. (Although sometimes I also looked for a short squeeze which would take the stocks even more out of line.)

 

[jem]

Quote from ReaM:

I think you are having a hindsight bias. Yeah, asking builders. You do know there are traders out there who ask this and they fail?

I know also someone who made money off health stocks. But those are 1% of people, the rest failed somewhere else. You just pick a winner and show him into my face, here, I have a winner, it works. That's BS man.

I never told trends don't exist, where did I say it? I stated they do exist, but they are random. You can't know for sure which will be the trend in the future, because some little piece of information can end that trend in an instant. Like imagine all of a sudden huge oil reserves will be found somewhere deeper in earth of amount of 10x the current estimated reserves of the earth and the crude light is $110. That's a good buy for the trend.

its hindsight bias... but for the fact I traded using t/a during that time period and made over 6 figures a year doing it with t/a.

 

[jem]

Quote from the1:

Tracking 2 std dev's is just scratching the surface in statistical analysis. This is the program I completed and it goes waaaaaay past measuring how far a data set is moving from the mean.

http://www.iit.edu/mathematical_finance/academics/index.shtml#program_requirements

I was just giving an example that stats can be represented in t/a.
I would I have liked to have taken that course.