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

Quote from marketsurfer:

The question should be, can a trader highly skilled in money management make money from a random chart? There are runs that both make money and those that will lose money randomly--- cut losses and let winners run applied to randomness--- will this work? if not why not?

Can an experiment be devised to prove this one way or another? (Other than the market itself) LOL !
More...

No. If the price behavior is a martingale in the strongest sense (meaning that the expected future price is the same as the current price when you take *all* available information into account) then a trader will not be able to make money except by luck. Of course the lucky traders will be certain they succeeded due to skill, and the unlucky traders will be certain that if only they can overcome their psychological defects they too will be successful traders.

If there are indeed consistently profitable traders then that does not mean they can make money off a random chart, it means that real prices deviate in some way from being truly random.
 
Quote from achilles28:

LOL The only clown I see is you. Parading yourself around as some guru lecturing everyone about martingaling. Look at your name for fucks sake "fxoilpro". You're a gambler. Not much more. And where are you returns ? The ones that prove you're a big hot shot pro? As far as speculative participation, those stats are a matter of public record. Look them up if you don't believe me. Then come back here and apologize. And maybe I'll forgive you.
More...

You wouldn't understand good money management, because you are a gambler.

If your 2 % risk per trade was used as martingale lot1 ,2and 4 = 7 lots max within 2 % of total capital risked , it would be considered good strategy (not martingale as you describe)
 
Quote from Rahula:

I've created an excel file I call "random chart generator". The weird thing is that most of these charts look exactly like market charts and you can apply all kinds of TA tools like moving averages, and trendlines to these charts.

If you don't believe me, see for yourself. Type in =RAND()*(-1-1)+1 into excel in column A. Then create a rolling sum of column A in column B. Then create a chart based on column B values and add a moving average.

Now what to make of this? I suppose when most people see a 7,7,7,7,7,7 result in the lottery they see a 'trend' even though it is just as random as 23,42,11,14,33,8 lottery result - both events have an equal probability of happening.

Now if the markets are random and the market closed up,up,up,up,up,up,up,up,up,up,up,up 12 days in a row - most people would likely be talking about it as a n-sigma event or a "black swan" even though it is just as unique and random an outcome as a up, up, down, up, down, down, down, up, down, up, down, down 12 day series of closes.

Now doesn't a random market assumption mean that both mean reversion and trend trading strategies don't have any science to back it up and at the end of the day it all comes down to luck?

I know, scary thread.
More...

to NYOBscalper's correct comments I might add that not only are markets obviously not random but neither are the numbers generated by Excel's "random" number generator.

One more thing I might point out is that in the lexicon of statistics "random" does not mean that the data itself is random, but rather that the order in which the datum are observed is random.
 
Marketsurfer made some interesting comments...
which I think boil down to this.

Money management should not beat random markets if there are transaction costs.

1. But, I know markets are not random. Arb funds are just paid by making them look almost random to non skilled traders.

many of you know I have been arguing with surf about this for years. But, my years of significant profits were almost a decade ago.


I closed out most of the accounts for a trader in my family two weeks ago.

He ran 800 thou up to about 5 million in profits over the last 15 years. He was mostly a swing trader holding 10 to 30 positions. He made trades just about every day.

I went back over the account history in one of the accounts and the consistent growth over the last decade was remarkable. Watching that account grow from 150,000 to 870,000 seemed like I was looking at papertrading results... but they were real. I told the man who is dying how truly impressed I was.
 
Markets ain't random. At least, not all the time.

To begin with, markets do have "memory".

Well, if anything it's the memory of eg all these hundreds of traders that have shorted the market at a level they believed would be a short term top and who see prices retrace suspiciously - if not dangerously - close to their break even level.

And yeah markets are "predictable", every now and then.

Like for instance in the example above when prices end up breaking out thru that supposed "top" and quickly trigger the massive amount of stop loss orders, which then turn into market orders, placed right above that "top".

And that people, fuels a very predictable and powerful run up. :cool:

And you know what ? Happens all the time, day in day out ! :p
 
Quote from Rahula:

I've created an excel file I call "random chart generator". The weird thing is that most of these charts look exactly like market charts and you can apply all kinds of TA tools like moving averages, and trendlines to these charts.

If you don't believe me, see for yourself. Type in =RAND()*(-1-1)+1 into excel in column A. Then create a rolling sum of column A in column B. Then create a chart based on column B values and add a moving average.

Now what to make of this? I suppose when most people see a 7,7,7,7,7,7 result in the lottery they see a 'trend' even though it is just as random as 23,42,11,14,33,8 lottery result - both events have an equal probability of happening.

Now if the markets are random and the market closed up,up,up,up,up,up,up,up,up,up,up,up 12 days in a row - most people would likely be talking about it as a n-sigma event or a "black swan" even though it is just as unique and random an outcome as a up, up, down, up, down, down, down, up, down, up, down, down 12 day series of closes.

Now doesn't a random market assumption mean that both mean reversion and trend trading strategies don't have any science to back it up and at the end of the day it all comes down to luck?

I know, scary thread.
More...

What's so surprising ?

If you never see" trends" in "random" data, data is not "random".

In random data given, for any (arbitrarily large) given N, there is always (eventually) a sequence containing more than N of equal consecutive outcomes.

Tom
 
Well, what you could do is attach a random number to every trading day on a given instrument you think you know well (it must be a larger sample to make it relevant) and then sort the numbers (in any direction) so the percentage changes become randomly ordered and see if you can consistantly pick which chart is random and which is real trading.

I think that is more fair than saying "a stock chart looks like a random brownian motion (which it does by the way)".

Also them being non-random does not automatically make them possible to predict.

You could also do this for intrada y charts by the way, just randomly re-order the bars (of any length) and see if you can tell which is which.
 
A number of academics have spent much time generating random charts with various characteristics such as tendency to have runs. They have gotten good enough at it that experienced traders can't tell the difference.

But then what?

No one is going to give up trading based on that evidence, and anecdotal evidence is iffy at best. Being in the top 10% of mutual funds one year does not predict being in the top 10% the next year. Individual traders who made mega money for 10 years in a row blow-up, and give it all back.

I believe that competence, thoughtfulness, and valor all contribute to profitability. If people who are really focused on Return on Investment are not going to spend their time hustling for slivers.
 
You must log in or register to reply here.
Back
Top