TA-Charting on Brownian Motion | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Suggestion for a charting sofware not charting service 3 replies

Charting tools that allow per-tick charting 0 replies

price in motion 12 replies

How can I represent motion around fibs with continuous variables? 0 replies

Attachments: TA-Charting on Brownian Motion
Exit Attachments

TA-Charting on Brownian Motion

  • #1
  • First Post: Oct 6, 2011 3:42am Oct 6, 2011 3:42am
Below are reasons to suggest that trading using a TA-charting strategy may be just as good as entering randomly. In the Brownian motion charts below, entering ANYWHERE is a good entry if you're bullish. Every point in time has the same probability of going up.

I want to emphasize I did not do any excessive choosing. 3 of the graphs I selected were consecutive, and two graphs from there I chose were from a group of ~5. So in all, these are 5 graphs I've chosen to analyze from a group of ~8.

This doesn't logically disprove that charting doesn't work in the market, but it does raise a question.


http://i925.photobucket.com/albums/a...d%20TA/bm5.png
http://i925.photobucket.com/albums/a...d%20TA/bm4.png
http://i925.photobucket.com/albums/a...d%20TA/bm3.png
http://i925.photobucket.com/albums/a...d%20TA/bm2.png
http://i925.photobucket.com/albums/a...d%20TA/bm1.png
  • #2
  • Oct 6, 2011 5:05am Oct 6, 2011 5:05am
Flip it on its head - if you assume TA does work, then what does Brownian motion tell you about collective behaviour in the markets?

Bon voyage!
  • #3
  • Oct 16, 2011 2:09pm Oct 16, 2011 2:09pm
maybe the problem could be inverted. what if you can predict random data with technical analysis?
  • #4
  • Jan 12, 2012 6:42pm Jan 12, 2012 6:42pm
Sorry for the very late response. I didn't see interest in this thread so didn't check back.

Quoting triphop
Flip it on its head - if you assume TA does work, then what does Brownian motion tell you about collective behaviour in the markets?
How do you mean?

Quoting deltatrade
maybe the problem could be inverted. what if you can predict random data with technical analysis?
This is not possible. By construction*, the data is independent and not autocorrelated (data point x is not correlated in anyway with previous data).

*You could argue that my software's random-number generator is not random. But not only would you have to prove my computer isn't random, but also prove that it's susceptible to creating linear patterns.
  • #5
  • Jan 13, 2012 5:53am Jan 13, 2012 5:53am
Quoting grimReaper
How do you mean?
Brownian motion is the string that ties institutional financial risk models, markets and algos together because it allows them to predict the randomness of movement.
It's used so extensively that it'd be something shy of a miracle if they *didn't* look the same.
But there's still the same need for big orders - big orders = price memory = TA.
  • #6
  • Jan 13, 2012 9:57am Jan 13, 2012 9:57am
Quoting grimReaper
This is not possible. By construction*, the data is independent and not autocorrelated (data point x is not correlated in anyway with previous data).

*You could argue that my software's random-number generator is not random. But not only would you have to prove my computer isn't random, but also prove that it's susceptible to creating linear patterns.
there is another way to prove that. experiments which can try to anticipate the next random numbers in the series. if you can generate candlesticks then i will tell you what to look for and make this experiment yourself.
  • #7
  • Jan 13, 2012 10:21am Jan 13, 2012 10:21am
Quoting grimReaper
Below are reasons to suggest that trading using a TA-charting strategy may be just as good as entering randomly. In the Brownian motion charts below, entering ANYWHERE is a good entry if you're bullish. Every point in time has the same probability of going up.
Assuming you're right about being bullish, then yeah - except how big of a drawdown are you willing to suffer? I'm not a big TA guy, but the reasons behind all chart analysis are to get a better entry. Would you rather go Long randomly, or would you rather enter on pullbacks or at the bottom of the range?
  • #8
  • Jan 15, 2012 7:07pm Jan 15, 2012 7:07pm
Quoting triphop
Brownian motion is the string that ties institutional financial risk models, markets and algos together because it allows them to predict the randomness of movement.
It's used so extensively that it'd be something shy of a miracle if they *didn't* look the same.
But there's still the same need for big orders - big orders = price memory = TA.
You can't predict randomness, that's a paradox. You can only assign probabilities.

But you make an interesting point that financial institutions may be using the same models which share similar assumptions.

Quoting deltatrade
there is another way to prove that. experiments which can try to anticipate the next random numbers in the series. if you can generate candlesticks then i will tell you what to look for and make this experiment yourself.
Here's an interesting candlestick chart. It's NT's simulated data feed with neutral direction, but I can make my own candles. And it's confirmed here that NT's simulated data feed is random.

http://i925.photobucket.com/albums/a...ndomdatata.png

Quoting Rob Mondave
Assuming you're right about being bullish, then yeah - except how big of a drawdown are you willing to suffer? I'm not a big TA guy, but the reasons behind all chart analysis are to get a better entry. Would you rather go Long randomly, or would you rather enter on pullbacks or at the bottom of the range?
But continuing with the assumption that the model is right, the probability a trade with X ticks profit target and Y ticks stop loss is just as successful if you buy the highs or lows.

Even if you wait for a pullback, you still don't know if it'll continue downward or start going up. Buying a pullback is just as successful as buying the highs in this model.

It's like gambler's fallacy. Just because red hit a lot, doesn't mean black or red is more likely in the future.
  • #9
  • Jan 15, 2012 9:36pm Jan 15, 2012 9:36pm
You seem to be coming from a very specific perspective, and it's obvious you're not a trend-follower. In a way, you answered your own question:

Quoting grimReaper
Even if you wait for a pullback, you still don't know if it'll continue downward or start going up. Buying a pullback is just as successful as buying the highs in this model.

It's like gambler's fallacy. Just because red hit a lot, doesn't mean black or red is more likely in the future.
That's where Technical Analysis, and separately, Chart Analysis, are doing their job. You or I might be right about buying into the uptrend, but TA/CA will keep us from having a 300-pip move against us before it continues up.

But I'm curious: since you're not a trend-follower, do you have a reason for this thread beyond an academic study? If you're trying to find out why successful trend-followers are successful, you can just ask. Otherwise, you're explaining to a bumble bee why he can't fly.
  • #10
  • Jan 16, 2012 12:18am Jan 16, 2012 12:18am
Quoting Rob Mondave
You seem to be coming from a very specific perspective, and it's obvious you're not a trend-follower.
I day trade index futures using S/R as I've described, and am happy with that approach, but that doesn't mean I can't be skeptical about it. Just because we perceive something as orderly, doesn't mean it can't be random.

Quoting Rob Mondave
In a way, you answered your own question:

That's where Technical Analysis, and separately, Chart Analysis, are doing their job. You or I might be right about buying into the uptrend, but TA/CA will keep us from having a 300-pip move against us before it continues up.
What question? In the model, entering on the pullback is just as good entering on the highs. You might be more comfortable putting a x-tick stop loss behind a recent swing or area of support when entering on a pb, but it's just as effective as entering on the highs and putting a stop loss of the same amount, again, in the model.

Would you use TA for entry in the first charts I showed?

Quoting Rob Mondave
do you have a reason for this thread beyond an academic study?
Like? Btw, there's a 2000pg thread called "Technical Analysis Fallacy," I'm not the only one to bring up the topic.

Many things in TA are worthy of being skeptical about, e.g indicators, fibs, ganns fanns, pitchforks etc, which I'm convinced either have no validity or effectiveness, at least in the markets I trade. S/R does have validity, but I'm still approaching it with an open mind.

As I said, I'm happy with my approach to the market, but just because something may appear to be orderly, doesn't mean it's not random. If you look at the candlestick chart above, I'd be convinced that was from real market.

Quoting Rob Mondave
If you're trying to find out why successful trend-followers are successful, you can just ask.
I'm open to anything you might have to share. But your judging me wrong if you think I'm a trader who busted his account and is pointing fingers.
  • #11
  • Jan 16, 2012 5:00am Jan 16, 2012 5:00am
what to look for.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: random2.png Size: 44 KB
Click to Enlarge Name: random1.JPG Size: 108 KB
  • #12
  • Jan 16, 2012 5:06am Jan 16, 2012 5:06am
Quoting grimReaper
You can't predict randomness, that's a paradox. You can only assign probabilities.

But you make an interesting point that financial institutions may be using the same models which share similar assumptions.



Here's an interesting candlestick chart. It's NT's simulated data feed with neutral direction, but I can make my own candles. And it's confirmed here that NT's simulated data feed is random.

http://i925.photobucket.com/albums/a...ndomdatata.png



But...
Quite right, I should have said 'manage' not predict. But that model is not the reason TA exists. It's the reason your charts look similar. TA can work because of price memory, but it sounds like you're veering into the 'is price random' arguments which have been done to death.
  • #13
  • Jan 16, 2012 11:49am Jan 16, 2012 11:49am
Quoting grimReaper
I day trade index futures using S/R as I've described, and am happy with that approach, but that doesn't mean I can't be skeptical about it. Just because we perceive something as orderly, doesn't mean it can't be random.
I admire you if you're successfully daytrading (I'm guessing) ES. I've tried very hard for a long time with ES, SPY/QQQ/single stocks and Forex, but can't make it happen. But what does work for me is following trends on the Forex daily chart.

Trend following and Supply/Demand reversals work because of both concrete and perceived reasons. The back-and-forth between A and B might be somewhat random, but there are legitimate, non-random reasons for the overall direction. A great example is the 5-year-long uptrend in EUR/USD - that didn't occur randomly, it was because of a combination of Fundamental money and Speculative money, based on both real and perceived value.

In order, I use Chart Analysis, Fundamental Analysis and Indicator Analysis. I use two moving averages (20 and 50 Simple) plus the Bollinger Bands, and while I place the least emphasis on these indicators, I don't think I could trade as well without them. They help me to determine the strength of the trend and when to enter and exit.

As for timing the entry, again, why would I buy randomly and risk the price first moving substantially against me, when I can wait for a pullback and trend continuation? At the very least, I could be wrong about the trend continuing and I'd rather know I was wrong within 100 pips than 300. The Fundamental bias and the existing trend give me confidence that it will continue, and I can just wait for money to start flowing into the trend again.
  • #14
  • Jan 16, 2012 2:39pm Jan 16, 2012 2:39pm
Quoting deltatrade
what to look for.
Can you explain why I should look for these patterns in random data?

Quoting triphop
but it sounds like you're veering into the 'is price random' arguments which have been done to death.
Not necessarily. I'm saying/implying one reason to believe price is not random is b/c s/r and trendlines, but lo and behold this "organization" plus other charting patterns can also be seen in random data.
  • #15
  • Jan 16, 2012 4:24pm Jan 16, 2012 4:24pm
Quoting grimReaper
You can't predict randomness, that's a paradox. You can only assign probabilities.
And that's a paradox too. By definition a random sequence has no greater probability in one direction or another.
  • #16
  • Last Post: Jan 16, 2012 4:42pm Jan 16, 2012 4:42pm
Quoting Ted1983
And that's a paradox too. By definition a random sequence has no greater probability in one direction or another.
When I say random, I don't mean a 50-50 chance of going up or down. In fact, in the charts of the OP, it's Brownian drift, i.e Brownian motion with a slight upward bias. We can assign probabilities to where price can end up and it's not 50-50. The point is, going long anywhere in those charts or that model is good entry, charting or TA won't help you, only hurt you if it signals a short setup.
0 traders viewing now