Bitcoin Forum

hey all you TA nay-sayers out there! this is your chance. i'm opening a thread for discussion on this very important point so you guys can stop spamming mine and lucif's threads.

let's set some ground rules:

-try not to repeat points.
-TA is a hypothesis that needs to be discredited properly (rigorously), i.e. not in the manner that nanopene employed.
-'disproving' TA involves proving a negative, which is not possible. avoid the use of that word altogether and stick to the idea of evidence.

let me also open with a well-defined hypothesis:
"Since there exist time-dependent autocorrelations in price data, past performance sometimes correlates with future results."

evidence against TA would be evidence that time-dependent autocorrelations do not exist in price data.

for example, a random walk does not exhibit any autocorrelations.

some examples of things that do not count as evidence against technical analysis:

-the relationship between technical analysis and herd behavior
-individual examples of poor analysis (stop picking on lucif :P)
-the fact that certain kinds of analysis (e.g. elliot waves) are prone to apophenia.

another note: if you don't understand the central premise here -- autocorrelation in stochastic functions, read up some more before replying for everyone's sake.

-===-

and with that i'll leave you with something cool: daily all-time price data compared to the DETRENDED PRICE OSCILLATOR, just an hour or so before the crash.

http://s23.postimage.org/ybe4o6agb/bubbletop50dollars.png

do you see what i see?

Lets talk about empiricism, lets talk about real world cases of people who became filthy rich by doing only Technical Analysis.
How many TA do you find among the Forbes richest people?
So, far I find... lets see... one, two... NONE!

Lets see those who followed the value investing methods proposed by Ben Graham and David Dodd... how about their direct disciple Warren Buffet? Then followed by Carlos Slim, Li Ka-shing, David Koch, Charles Koch.
Value investing was born as a direct response against the chart reading, which is as effective as any pseudo-scientific school.

Investing is investing if you invest. Swallow that truism.
Also, data without context is meaningless.

PS: Oh, btw, don't exclude "elliot waves" as more prone to apophenia, we are prone to anything and everyone as long as we have human brains.

-facepalm-

this is not empiricism.

-===-

let's not get offtrack here, because i gave a very straightforward hypothesis in the OP which you did not even touch, but one point -- 'proneness to apophenia' doesn't correlate with invalidity. it just makes the analysis easier to screw up (read: bias). that's all i was trying to say.

also, it's not surprising that well-known methods of technical analysis do not work. this is actually a good point. this is probably due to the fact that markets are anti-inductive. the fact that markets are anti-inductive also works against the so-called "Lemming Effect" that many claim is responsible for any apparent success.

"Since there exist time-dependent autocorrelations in price data, past performance sometimes correlates with future results."

This is a general observation, but it does not support any specific technical analysis results. The fact that autocorrelations exist does not allow you to write arbitrary laws about these correlations. Your laws need to be stated as theorems and backed by proofs. Otherwise all you are saying is "observation: heads sometimes follows tales", therefore any time heads follows tales we can invoke technical analysis in support of the observation.

The problem with technical analysis is that due to its fractal nature it is unclear what time scales to consider. Any analysis that works on one time scale can be completely contradicted on another time scale.

As much as I enjoy looking at charts, the bottom line is there is nothing to it. It's just alchemy.

It is empiricism of the efficacy of the discipline as a whole, stop and think.
Empiricism is the experience gained through the trial and error to see what works and what doesn't.
Given the millions of people using TA, experts traders, none of them seem to have made it. Why?
You can use all the excuses you might want, but the conclusion is clearer than water: it is useless in the real world.

In other words, the lack of successful technical analysts making it to the Forbes list is empirical evidence of its uselessness, especially considering that charting existed since the 17th century.

Hey, lets get the facts straight. You are constantly spamming this forum with your threads advertising the so called "methods" of "technical" "analysis". This is annoying at least

Even the mere fact that you use the words "technical" and "analysis" for some kind of coffee cup reading is infuriating for anyone who has learned his way in the hard sciences and knows how technology works.

So this thread is your chance.



And here you start with your manipulation and propaganda right away.

Again, lets set things straight.
By this "disproving" approach, you are implying that TA is even anything of a method.

This is not the way it works. Rather, you are proposing some kind of method, which you like to label "technical" and "analysis"
Now you have to show that this method indeed exists, and provide a verifiable description of it.

This works by starting from a null hypothesis, and disproving this.

So the null hypothesis might be:

• TA is not an exact method, i.e. it does not consist of any clearly delineated description of operations, which can be conducted with precision and lead to a predictable and verifiable result.
• and especially: TA is not a method yielding any reliable predictions of market behaviour with a probability significantly above picking of random hypothesis

So to disprove this null hypothesis, we need to give (1) a precise description of some method or algorithm, which can be executed in a reliable, rational, and verifiable way. And (2) we need to execute that method under controlled and verifiable conditions and yield a significant score of prediction.

Unless we succeed with that, there is no such thing as "Technical" "Analysis". It simply doesn't exist.
And thus its use should not be advertised further.

---

But maybe there is a third solution. Maybe there is something much more level-headed. Something along the lines of "a long time trader's rules of thumb". Something more of a craft, than being technology or analysis. Then we should be so prudent to ditch all that pseudo-math and pseudo-scientific furnish and handle it just as that: a bunch of rules of thumb and gut feeling. There is nothing wrong with that.

'disproving' TA involves providing an example of a prediction made by TA that turned out to be false. It happens all the time on this forum. I do not have the time to do it now, but it would be pretty easy to take a bunch of charts from the past, ask the TA people to predict what will happen next, then reveal the next week of the chart and check their answers.

That's not a hypothesis. You have a premise "there exist autocorrelations", and then you restate that premise in the form of a definition. There is no inference/conclusion, so no hypothesis here.


Of course there exists time-dependent autocorrelations in price data. Why else would people be buying on speculation? Because there's an upward channel going up, a (highly correlated) time-dependent autocorrelation. Pretty much everybody here loves TA, I don't know who you are talking about by making a general address to "TA nay-sayers". You should qualify it by bearish TA nay-sayers.



I don't know, what do you see? Are you (again) calling the top, a reversal?

There is money in TA, and I guess the HFT in Wall-Street are not a myth and they DO make money (actually lots of it).

However, let's see the following: Can TA predict the Apple Stock growth? How can TA knows that Steve Jobs will make the iPhone and iPad?

It can't. Simple. and that's why Apple stock was almost dead in the first few years.

What TA can do is analyse past trends and try to predict future trends based on these past trends. This works on a very short period of time levels, and assuming that nothing changed.

For example, in the last bubble, there have been certainly a pattern to the quick raise and full. This bubble it might be similar or different. What TA bet, is that it's similar (or the analyst bet that his prediction based on his knowledge/skills will happen).

However, when it's a small market like Bitcoin (and a relatively long period) between the two bubbles, things have certainly changed (people holding the bitcoin wealth, the mentality of the new owners, past experience with the last bubble...) which mean there will be a different pattern.

Who's TA have expected the price to jump again to $42 after that crazy drop?

Everyone knows this will end somewhere in the bottom, unjustifiable high increases in BTC value with no accompanying economic value will certainly result in a burst.

If you are talking about HFT, it is algorithmic trading, not charting (TA). In HFT there are many strategies and factors weighted in that are beyond the capabilities nor interests of a charter. I can guarantee you that if you make a HFT software based solely on TA principles, you will lose all your capital in no time.

Well, that "crazy drop" bounced exactly off the 20 day EMA, which has been tested several times during this run, so by this measure the uptrend is still intact. "Support bounces" off of rising moving averages (i.e. for securities that are in a nice uptrend like Bitcoin is) is a very common set-up for technical traders, so this bounce is not at all unexpected from technical analysis. My two cents in this debate is that technical analysis does work, but you have to understand how to use it. My sense is that most people who claim it doesn't work don't know how to use it. It takes at least 6 months of intensive study to learn technical analysis, and that would be for someone who picks it up pretty quickly. However, it is also true that really good fundamental investors make more money than technical traders. But technical traders make bank too. Personally, I take the fact that we have bounced so hard off the 20 EMA as a sign that the bears shouldn't get too excited yet. I can't speak for the OP's analysis - not familiar with it, and I take the successful 20-day EMA test as more significant, though it hardly guarantees that we will go on to new highs from here.

As long as we run above 32, this ain't no reversal. If it goes below that, to 29 at least, then maybe it could reach 22. But I still don't believe that, my voodoo charts analysis tells me it's just slowing down of pace, which been craaaaazy for a couple of days there. Back to a comfortable 1% per day growth for us.

I love chart patterns. I don't use any other technical indicators, only patterns. This pattern (yesterday) has the potential to indicate a spike to 55$ (next few days, maybe even today!).

CHECKMATE BULLS

Nostradamus' did.


Quite obvious and much less open to interpretation that all that TA nonsense.

http://en.wikipedia.org/wiki/James_Harris_Simons is pretty rich.... from advanced quantitative trading techniques.

Maybe, maybe not...
 http://en.wikipedia.org/wiki/Survivorship_bias

This thread is pretty pathetic. If this is really the sum total of what this forum has to offer in defense of technical analysis then it really paints it in a much worse light than I originally thought.

I'm sure it has its uses. The problem is, people tend to think "Look, something we can measure and fit a curve to, let's apply it to everything". You need to know what you're using, why it is what it is and if, why and how it applies to Y when it was derived from X.

If you want an example from a hard science outside the field, look at the black body radiation curve. Classical physics produced models that didn't match reality. Einstein came along and resolved the conflict by inventing the quantization of electromagnetic radiation. Just because something works in one scenario, doesn't mean it'll work in another.

Planck*

Planck assumed that the energy of the oscillators in the cavity was quantized, Einstein took it to the quantization of electromagnetic radiation. I did unfairly leave out Planck though so good observation.

First of all, you chose a wrong example, he doesn't use TA to manage his hedge fund.

What the hell arepo?

Why do you talk about the difficulty of disproving TA, when you provide nothing to prove it.

That's like the classic argument for religion, "you can't disprove god doesn't exist". Then again, I guess TA does require blind faith as well.

There is his prediction track record. I just browsed his posting history. Some semi random quotes. Note the date, and if you have any doubt, look up the charts:








etc, etc. And yet bitcoin doubled and tripled during this period despite his endless strings of predictions of imminent crashes, bulltraps and what not.
Proof enough for me that reversing Alepo's predictions so far has demonstrated a statistically higher chance of paying off than throwing dice.

Pretty much, this.

ITT: Someone who's very bad at technical analysis, so bad that virtually all of his predictions have been dead wrong - tries to prove to us the virtues of TA.

Now, Arepo, I have no doubt that somewhere in the world, hidden in a mahogany-trimmed office full of computer screens with charts and books about fractals and logarithms, there are a few geniuses who manage to make significant profits off of TA.

You ain't one of those guys ;-)

And you can argue the virtues of TA all you want but where are the results, bro?

Some people say "Judge by results, not theory." I think we should judge by both results AND theory. But right now I haven't seen enough results to make me even remotely interested in spending 6 months studying the theory so that I can be as bad as you at it. If the learning curve is incredibly steep just to be worse than a dice at predicting the market, and we are struggling to come up with even a couple names who have used TA to attain anywhere near the wealth of the value investors in the top wealth brackets, well... good luck trying to "prove" your system.

To show that TA works I don't think "controlled and verified" conditions would be a feasible experiment. You'd have to round up every Bitcoin trader in the world and somehow control them, because they are the "conditions".

When executing an experiment itself affects the outcome of future attempts of the same experiment, that experiment is not very valuable. For instance, let us assume there was something to TA and we could identify the patterns we wanted to test.

If we ran many "controlled" experiments with real money, other traders would react by changing their strategy to take our particularly rigorous and predictable behaviour into account. If we ran the tests with play money we'd lose the vital component of how the trader's own actions affect the market.

So rather than shooting for a "controlled and verified" test, perhaps just taking a statistical sampling of the average success rate of TA traders versus the norm could be a reasonable test. Setting such a study up would require finding a large group of traders willing to be studied though.

You could test with play money to see how TA would work for investors not big enough to (significantly) affect the market. TA probably isn't as applicable to the big players anyway as they will be adopting more active strategies.

You have a point here, and this brings up the question if this is a matter of science and specifically of exact mathematical science. But there isn't just math and physics, there are also the moral sciences and there is technology and engineering. Each of those has a way of precision.

Indeed, personally I rather see TA as an proposed / alleged method of technology or engineering. And it is quite common to run field tests of various kinds to verify methods and procedures in technology. In this context I had in mind, "controlled and verified" means that the conditions are controlled to an extent so that the experiment could be repeated, which of course includes some management regarding the knowledge of the subjects involved into the test.

In situations where self-suggestion of the test subject is important, often some kind of blind test or even double blind test is applied. Transferred to our situation, this would mean to conjure up some trading strategies, which look superficially correct, but are wrong or ineffective, based on the methodology of TA. And then, in the practical test, the traders executing these placebo strategies need to do significantly worse than those employing the real TA based strategies. And, as a point of reference, some strategies based on random choices need to be included as well.

This is a good example why TA "analysis" constantly raises all kinds of red warning signs.
This statement quoted above uses an impression, assumption or conclusion as a basis, not an unbiased fact. Then, on top of such shakey foundations, TA folks typically execute a lot of math formulas and graph vodoo.

What is wrong in the statement above?

First, the subject in question has not been constituted properly (anyone observing that "crazy drop" should have noticed that the reversal and spike up coincided with the effective drop-out of the Mt.Gox data feed, leaving most of the Bitcoin economy blindfolded. Also you should have noted that the down move ended in a cluster of existing orders, which would have been a noticeable resistance to break. Yet the claimant doesn't take that into account or even mentions it.).

Secondly, the hypothesis to be verified by this "observation" might have been chosen to fit, and there are no safeguard measures against self-deception. (why 20 day EMA, why not 10 or 30 days. And what exactly is the scope of "several times during this run"?)

Third, there is no actual statement or claim, but rather a conditional, self-referral claim ("so by this measure the uptrend is still intact"). But for a person not trained in logical dissection or scientific reasoning, this looks like a real statement. And especially this is what raises the suspicion of manipulation.

Oxymoron. The guy who invented the term "Social science" should have had his arse kicked all the way down the road and back.

The reason for the 20 EMA and not the 10 or 30 is that the lowest dips during this run since January have been bouncing off of it - it's a pattern price has been following. That is a plain fact, not a biased one. I couldn't have noted the cluster of orders because I don't have access to that information, since I'm holding long-term and don't have an account on any exchange. I'm not surprised, though that there was a cluster of orders right at the 20 EMA, since that is a good spot to go long on a dip given the established pattern. What it does is give one a low risk entry, because one can sell quickly if the dip doesn't turn at the EMA, thus limiting one's loss. This is how technical analysis is used by traders - they use TA to find "low risk" set-ups, where low risk means that if things don't go as expected, one can stop out quickly, but if they do go as expected, one will have a significantly larger gain than the potential loss. It basically gives one a "line in the sand," and TA may work largely because traders tend to use the same lines in the sand - but it doesn't matter why so long as it can be used effectively to find low risk trades. Also, a moving average is mathematically a way to gauge an uptrend relative to a time-frame. This is a technical point about the price action that can be used by a trader who is trying to stay with the trend and not sell too early for emotional reasons ("Gosh, price is too high! Sell!" or "OMG this is a big dip! Price must me crashing! Sell!"). If there is a hypothesis here, it is only a trading hypothesis. A reasonable strategy for a trader to take here is to say, "I'm going to stay with this run so long as we don't breach the 20 EMA, since this has held so far." So, one might place a stop on one's trade a little below the 20 EMA, and keep raising the stop as the EMA rises. This is just a way to make a reasonable line in the sand for oneself in order to stay with the trade. There is no hypothesis here to the effect that BTC (or any other security) will always bounce off of a rising 20 EMA, nor is there a hypothesis that we crash into oblivion once it is breached. But a rising 20 day EMA is a support level, so it is a spot to potentially go long, or a spot to set a stop below, since breaking support represents weakness in price momentum.     

Seeing the lack of rebuttals, I think he just created this thread as a sandbox for us for all the bitching.
Genius move, haha


Social science is real science.
Experiments follows the scientific method, they ARE falsifiable, they are quantifiable, they can be empirical, and they are peer reviewed.
It also relies on observation, and there are a very eclectic range of subdisciplines.
Social Psychology and Behavioral Economics are my favorite because of their counterintuitive discoveries, and I exploit them effectively to push sales every day, to influence people and to seduce women.

Please, don't ridicule things simply because you don't understand. You'll end up ridiculing yourself.

So in order for a TA analyst (I won't judge the TA science per se, I'll keep it with the people) should make correct predictions most of the times (or at least, correct enough to break even).

So here is my suggestion: Every TA in Btctalk keep a note about his predictions in a text file (sure you'll post about it here, but I don't want to go through hundreds of posts) and then present them to the btctalk society to judge his TA skills.

The file format is quite simple

dd/mm/yyyy: (predicted price for the day) | (predicted future trend)

Who's in?

It could be real science, but as currently practiced it is not. In social science everything is related to everything else, so you know the null hypothesis of "not related" or "A has no effect on B" is false to begin with. I'm sure there is some good work out there so please show me a paper that attempts to falsify a real prediction similar to "the speed of light is between 298,000,000 and 300,000,000 meters per second".

This is similar to TA. To judge a particular method's effectiveness people need to make predictions of "the price will be within this range at this time and date".

We should do better than that, and create a speculation game/competition. Everyone starts with 100 virtual bitcoins (is that an oxymoron too?) and trades them using mtgox prices. After a few months, we see who comes out ahead.

That sounds like a fun idea, but also sounds like a lot of work and would be hard to keep everyone "honest".

In any case, I'd play and just sit on my 100 coins :P

why would it be hard to keep them honest?
And yeah, it would be a bit of work, especially if you want to do it right, but a basic implementation might not be that hard. I think I will look in to it anyway.

Doubtful.
Chartists are limited in two ways:
1) All the techniques that they learn, it is basic and known by millions of wannabies who learn the same MACD indicators. EVEN if it worked, there is no advantage to be profitable. And because everyone shares the same knowledge, the market reacts based on that preconception, and it works only because everyone agrees that it will happen, thus happens. This is known in psychology as self-fulfilling prophecy.

2) They are human. What I mean by this is that they can't process all the indicators in real time, they must simplify and most of them rely solely on graphical patterns, neither mathematics nor market analyses.
Those who rely on math, end up generating algorithms, which are obviously kept secret if it works, therefore they have an edge they can profit from. If everyone knew THE a successful strategy, the edge would be lost.
Charters (TA's) don't research the market for indicators, they apply tools that had existed for centuries. And because every average guy know what you know, you can't beat them unless they make a huge mistake. At the end, without solid math, it becomes a cheap martingale (the betting system, so popular among amateur gamblers).

The way I see it, Technical Analysis is to alchemy, what Algorithmic Trading is to chemical engineering.
In algos you go beyond what the drawn pattern, you gather waaaaaaaaay more data in search of patterns that may anticipate the market.


I would like to know what you consider to be a Social Science, you are encompassing a whole spectrum of disciplines.
Are you talking about Political Science, Sociology, Social Anthropology, Social Psychology, Social Neuroscience or Economy?
The way that people generalize "Social Sciences" is not only rude, but very ignorant... especially when they have a distorted concept of science even among those who claim to love hard sciences. And I blame the schools and colleges for not teaching the history of science.
I would understand from high schoolers, but seeing this kind of behavior from adults is very frustrating at the least.

First of all we must define that science is the pursuit of objetive understanding of the natural world, first and utmost.
Then IF this understanding is true, it should be testable, reproducible, therefore, it should have predictive power.
And IF this understanding is false, it should be provable as false.
This is the key requirement to be scientific.
Science+Time refine this knowledge. High accuracy is not an excluding requirement to be scientific (or to discredited), but it is a refinement of the result of this testing and retesting process over time to confirm, improve or reject an hypothesis or a theory. It is really not the result but the process that matters.

Currently all social sciences (excepting political sciences), have predictive power. Some disciplines have more, others less.
Obviously the "harder" it gets (such as the neuropsychological branch), the more accurate it will be.
And yet, all the corpus of knowledge from all the disciplines are extremely useful and relevant for our everyday life, and they are actively exploited commercially. It is so pervasive that it is transparent in our daily lives, but right now you are being manipulated by the biggest marketing firms.

You should ask yourself why did you buy the brand what you bought. You might think you had free will, think again.

Yes, please show me a paper from any of the fields you listed that falsifies a real prediction. As I said, as practiced, the things getting falsified are worthless because we know them to be false before trying to falsify them. It would be the same as if TA was deemed accurate because it predicted the price would not be exactly the same tomorrow at this second as it is now. So it is just a 50-50 chance of guessing right up or down.

That is a grave statement and a deep misunderstanding of the scientific process.
First of all, what is your profession? I just want to know what kind of audience I am responding to.
Are you a graduate student in hard sciences, an academician or just an average joe fanatic of the sciences?
If you are one of the first, I am appalled at the lack of epistemological understanding. (this is something I also realized among the grad students in my school)

You can never know what is false unless it is tested. If your mentality was widespread, all counterintuitive hypotheses would be rejected from the get go. That kind of prejudgement is very harming for scientific discoveries, which makes me think that you are not actually related with any scientific discipline, either that or you are too new to this.

If you want to know about researches in social sciences, simply subscribe to social scientific journals yourself. If you work in academia, you should have free access to all of them. You have plenty of social disciplines that use quantitative research.
You can start with SSRN and then the Journal of experimental social psychology (ISSN: 0022-1031), Personality and social psychology review (ISSN:1088-8683), Journal of personality and social psychology (ISSN: 0022-3514), Experimental Economics (ISSN: 1386-4157).

Finally, I invite you to read this article published in Science:
http://www.ucd.ie/geary/static/publications/workingpapers/gearywp200935.pdf
Have fun.

You are severely misunderstanding me. Take any two groups of people and compare them along any measure and you will find they are different if you look close enough. I "know" this as much as I can know anything. This is what occurs in the social sciences. Here is a good description of the problem from back in 1967:

http://www.psych.ucsb.edu/~janusonis/meehl1967.pdf

I just noticed this edit. From that paper:


This type of thinking is exactly what I am talking about. "Higher" this, "Lower" that. Its 50-50 all the way. The falsified hypothesis is one of zero effect of A on B. In social science A always affects B in some way, so that hypothesis is worthless. This is what makes it not really "science", the data may be good, but the way it is interpreted is not scientific. No matter how many confounds there are it should be possible to estimate some interval of the result of A on B that can be narrowed over time as more data is collected. This does seem common in the social sciences for whatever reason.

Have you ever seen a social science experiment replicated exactly?

Note: this all goes for biomedical science as well. The only difference is it is often easier to control confounds.

Funny, I needed this paper to troll my professors :)
To be honest, I have my own criticisms as well, but I wouldn't go as far as saying that soft sciences are not real science.
It is science, not pseudo, not fringe, it IS science. At least psychology is a field that has been serious about it, and within psychology, neuropsychology, behavioral neuroscience, comparative psychology, neurobiology are the hardest of all in the spectrum of the psychological subdisciplines. In fact, there is nothing really "soft" in them, they all are very well versed on NHST (well, they should be), it is a requirement for research for that line of study.
I will read that paper with more time to dissect it carefully later, I love it, thank you.

Now, responding to your remark: "compare them along any measure and you will find they are different if you look close enough".There are always differences, that's granted. What matters is if it is statistically significant. To be sure there are different experimental designs to filter out, changing criterion, reversal and other that are very well known in medical research, double-blind testing, control groups. We use Analysis of Variance to tackle that problem, which is the same statistical tool used in any other "hard" science.

The only difference with the hard sciences, is that we are much younger and growing.
Btw, we are way offtopic.

PS: Let me share this paper: http://www.statpower.net/Steiger%20Biblio/Steiger04b.pdf

That's basically the same thing (he should add the quantity invested of the 100btc, maybe in %).


* 100 is the base. You start with 100.

So where are all these TA analysts  8)

I would suggest to stop using that. ANOVAs rely on assumptions (eg normality) that are either known to be false when describing human behaviour or impossible to prove. Do permutation testing (https://en.wikipedia.org/wiki/Resampling_%28statistics%29) instead.

Well that will depend on the type and nature of the experiments that are performed, but thanks for the suggestions ;)
Are you a statistician?

No, just someone who deals with similar situations and realized ANOVAs, etc are not the correct tool for that job. People just use them because its the only thing they were ever taught.

Lets stop this OT convo tho.