Topic: Is technical analysis bullshit? (Read 4615 times)

Wrong on two counts, I'd say:

(a) there are a number of statistically well established predictive factors (not many, but some), so I'm not sure where you're getting the idea from that nobody ever managed to find any correlations that enable market decisions/risk control better than 'guessing'.

(b) your claim that "if formal algorithm X can't do it, the human brain most certainly can't do it" is pretty off, imo. As Lo et al. put it:


(source)

TA is an art not a science.  TA makes no assumptions about time price correlation.  Where did you get this idea?

Quantitative guys dont even need to look at candlestick charts.  They just look at options prices and greeks.

Fundamental guys just look at financial statements and DD

Market profile guys look at volume profile & TPO

TA is just a method do analysis.  Just like you can do quantitative analysis based in statistics or fundamental analysis based on financial statements and DD.  They all have there own place

yes, you're missing something

Technical analysis is based on one assumption:  that there exists time-correlations in market prices.

If someone manages to formally prove the existence of these correlations, that would settle it for me.

I've seen very complex attempts at extracting these correlations, through artificial intelligence algorithms such as high-dimensional support vector methods. These algorithms can find extremely complex correlations in the data that would be very hard for us humans to grasp, or completely unintuitive. If these methods fail at detecting correlations, I have a hard time with the credibility of "toy functions" used in classical TA.

All TA is, is a "technical" analysis.  Instead of looking at the assets fundamentals, analysts looks at price action.

The patterns tell the analyst something about the market.  Some patterns arent common in reverse.  For example a pennant.  You see pennants A LOT before breakouts.  But you will rarely see the reverse.  I dont even know if there is a reverse where you have a breakout then suddenly squeezed and gradually unsqueeze then breakout

If someone were to pick a random price chart from a random market at some random point in time in the past over some defined time interval (say, one week), and then flip it so that the chart runs backwards in time -- would someone trained in TA be able to recognize that it had been flipped? If so, how?

You dont use TA to predict the future.  You use it to confirm your thesis.

It gives you context on market behavior.  My thesis is bearish.  How do I know?  Lower highs.  Pull up a daily chart and draw a trendline through the highs.  Gets rejected everytime.

If you want to trade long draw a trendline through the lows.  Bounces everytime.

Looks like a falling wedge.  Its bullish WHEN that pattern breaks and not before

Some of the people here have been pretty spot on on the bitcoin price. Check the thread that predicted 270 ish by mid april, and either a big rally after mid april or a loss of support. Very interested to see what happens.

I think of TA as being like playing poker. Meaning:
- Some people can in fact make money more or less consistently by playing.
- But it is a zero-sum game.
- A lot of it is psychology.
- In poker if you discover a "tell," then it ceases to work if the person who commits the tell learns about it; likewise in TA if a pattern becomes widely known and accepted, it ceases to be followed because everyone would front-run it.

It should only be used as a reference of an idea of where its heading, but not a 100% sure thing.

Yeah, you can mention about trend line support, and the break outs all that crap. The reality where someone can predict the future, you let me know and ask them when i`ll die to be exact.

No one knows, it should be used as a measure of this "could" be the general direction.

The same also applies in day trading, "intra day traders", who do this im sure many traders try to predict the next move.

TA is a function of random variable, thus it is random variable.

Market has a deterministic AND random components. If it was purely deterministic, we would be able to predict it thousands years ahead.

I think TA is not the same as throwing a dice. Its not random. The same way its not random if you play poker or blackjack. If you know a bit more than the others and are better then you have an advantage. The only thing you then have to overcome is the orderbook spread, fees on exchanges and of course the price has to move more often the way you predict it so you can make a profit.

Why isnt it random? Two points... there are humans behind those prices. First, the price goes down and a couple of people is selling fearing a crash. The same goes in the other direction. So if you can find out how the mass of people work you have an advantage. And there can indicators come into play. Showing if a stock is overbought and so on. Its simply an advantage against others.

Second thing is the other traders. They believe in indicators and so on. Simply see the resistance and support lines. They are in the chart. Check it. And if you see them then other traders see them too. And yes, its a self fulfilling prophecy. Traders make these indicators work. Thats it.

If you are a normal person you dont have a clue about this all. Then its random, yes. If you learned a bit about it you will find that there is an effect. Its not 100% but the small advantage you can get is the difference.

Pretty much.


There's TA and then there's TA.
If used well the BTC markets are pretty easy to trade.

Its bullshit to people who dont use it correctly.

Its hard to predict the future.  But you can spot the same patterns often.

Technical Analysis itself is a very vague term and might even include moon cycles according to some, but not others. I consider the term crash to be a similarly vague term that might mean 20% to some, 10% to others, or simply a straight line going down to a few.

The price went down by close to $40 on March 19th and further today, which when taken as a whole I regard as a crash, especially if it continues further down.

People, you want to discuss whether technical analysis is useful or not in case of Bitcoin, but it seems some of the basic terms are misunderstood.
Crash? what do you mean by crash? usually, in stockmarkets a correction of more than 20% is considered a crash. Which crash are you talking about? You mean the most recent correction? It is not a crash as of now.

The crash started a day before Friday's solar eclipse with its associated full super moon. If it had started one day later we could attribute it to the full moon and include moon cycles as TA on that list.

fixed
 

Support/resistance?

Indicators?

Elliot Wave?

Moon cycles?

 

Have you heard about feedback loop mechanisms in controlled systems? Every set of new information is integrated to refine a result. The faster this happens i.e. as in real time systems, the better.

I never get the debate. Why not just do both TA and fundamentals. It surely can't hurt to have more than one tool on your box?

because teaching TA to a willing group of buyers is not that bad either. The probability of making money from teaching TA is way better than trading volatile markets. Why not?

Check Steve Nison. trader turned tutor and he seems to do ok.

Ah, well... I wouldn't say invalidate, but it's certainly something you have to consider and be prepared for violent jumps. People are often relying on the same indicators and lines, arrive at the same conclusions and thus act accordingly...

agree. bitcoin is realatively young market and with small market cap. thus, it is easy to manipulate at any given point of time, invalidating the TA.

It works worse for Bitcoin in a way, I believe. Bitcoin has pretty unique features and patterns. On the other hand, it's a pretty low market cap, still and can be influence easily. I guess it's mostly a self-fulfilling-prophecy, though!

I'd suggest not to make pretty wide ranging assumptions based on nothing but a "hunch".

Seriously though, take a quick look at the article if you feel like it. In a way, it supports your point. In another one, mine as well.

T.A. and Bitcoin

might as well toss a coin

You can fit a statistical model to data and use it for predictions, but this is already beyond TA. This is not extrapolation, this is prediction. Building a model requires to have fundamental knowledge about the system.

Some processes indeed can be predicted by simple models such as Markov model quite well.

well, in all honesty, think carefully what TA is aiming to do?
To me it is to model mathematically psychology of people trading any given asset. That changes from asset to asset as there are different groups involved in each case and thus the same set of indicators that works for one may not work for the other. Recently, also most of the markets have HFT algos programmed to make the most of each market, and taking the advantage of these who stick too strictly to their charts. Also, different markets have different grade of volatility, which is another factor to take into account that make influence price behaviour. These are just some of the aspects that are involved  here, the list is definitely longer, on why it seems the TA may not to work. It does however provide you with some information on how market reacted in the past. It is up to you and your experience to evaluate the usefulness of such hints.  

I liken TA to astrology.

Both have been practiced worldwide for a long time by many people, although astrology has been around much longer.

Both are based on sketchy "science" and could better be described as craft.

Both have an almost cult-like following.

I'll read the horoscope in the newspaper for entertainment but I won't base important decisions on it. Same goes for TA.

Just my BTC0.00000002.

What do you all define as TA?

Support/resistance?

Indicators?

Elliot Wave?

Moon cycles?

There is a lot of "TA" out there. In fact, you can invent your own and call it TA. It is a very loose definition.

I'd rather worry about fiat money die before that. Now when interest rate is going negative, it is not faraway from a mass drop in fiat money's confidence, and when that fear spread out quickly, fiat money will lose most of its value in a very short time

Fiat money and bond are all created out of thin air, the value of these things are even more intangible than bitcoin, if they can hold its value, bitcoin will definitely too

You use the big words, but you don't really know what they mean, huh?

Whether you can extrapolate (in a loose sense) from past market data to future outcomes runs down to the question whether a market is a Markov process (or rather: can be modeled by one) or not. Some processes can be shown to be likely to satisfy that property (in which case, yes, you can't extrapolate), some processes don't, then you can go hunting for predictive factors (which take historic data as input.)

Hint: even most of mainstream economy doesn't believe markets in general satisfy the Markov property. They just think (most) markets are also efficient enough to price in the information potential almost immediately. By that interpretation (which I don't share), TA still doesn't work, but not for the reason you seem to think it doesn't.

they are based on probability in fact, they don't really predict anything , it's more of a talk about the "movement"/trend


what if bitcoin die, or stay at bad value indefinitely?

TA works as good or bad as any other self-fulfilling prophecy. It's 90% psychology and 10% statistical probabilities.

It was meant to be a joke.

EDIT: By I'll try again:
Chess is zero sum game, therefore chess theory is BS.

Yes, you might say financial markets are a chess game where the players, dimensions of the board, and movement rules for pieces change after every move.  A strategy that worked last time to gain a piece is in no way guaranteed to work again.  The odds of it working again are inversely proportional to the intelligence and alertness of the players who were present last time, are present once again, and still have significant funds.

Looks like he was being sarcastic and pretty sneaky about it. Probably one of them chess playing wankers, they can be crafty as hell.

I can't really understand how you compare chess with lotteries.

The same story with chess. If some winning chess strategy existed, everybody would learn it and become the champion, which is not happening. Therefore winning in chess is pure luck, like winning a lottery.  
[/quote]

This makes no sense really..

The same story with chess. If some winning chess strategy existed, everybody would learn it and become the champion, which is not happening. Therefore winning in chess is pure luck, like winning a lottery.  

It´s a tool and like with other relatively complex tools you need to learn and train to be able to understand and use them. Otherwise you may very well be clueless as to their functionality - I guess.

My father used to work as an economist, he studied macroeconomics in uni yet he calls bullshit on technical analysis.

I've tried doing TA myself and I can tell you that it's really hard to come to a logical conclusion solely relying on statistical evidence. Many people that do TA professionally base their predictions on real world events, they study the news more than the stats but that's something they'll never tell you.

There's too much hypocrisy when it comes to people that claim to be good at TA. I personally wouldn't say that it's bullshit. I wouldn't base my trading movements on an analysis alone especially with bitcoin. In the case that you're starting to think that TA TA is bullshit, then you should believe that bitcoin TA is bullshit*2.

You know how volatile and open bitcoin is, there's no way to predict long or short term movements solely on stats.

Because extrapolations lie. Try it, learn it hard way.

Buy and hold is much much more useful, especially for bitcoin, no amount of TA can defeat a long term buy and hold strategy 

And why is that, exactly? If past market patterns such as price/indicator divergences have repeatedly yielded a statistically significant amount of valid predictive signals, why should one assume it would suddenly stop doing so in the future?

Granted, no TA pattern always has the same guaranteed outcome, but there are many (such as the aforementioned divergences) that historically have been shown to have predictive value beyond random chance, and when used in conjunction with other relatively reliable patterns, greatly increases the probability of a projected outcome as further confirmation rules out more uncertainty caused by the inherently probabilistic signals.

Of course things can also change, and previously reliable patterns may at some time end up not working as well (or at all), which will be noticed and taken into account by the analyst in order to adapt to a dynamic market. But until such time I do not see why it is "fundamentally wrong" to assume past data of market behavior can be extrapolated into the future.

That is very true, and it has to be clearly said that for now we had very little fundamental developments that could support the price of Bitcoin. For now the two past bubbles were made on the pure speculation, with not much fundamentals involved.

Insider information = best TA

When the Euro was crashing a TA expert said the indicators were stressed to almost their maximum and were useless in that situation. Ironically, that's exactly when you want to predict what the price will do the most. Do you think in extreme crashes that most TA indicators are useless?

If you have above average intelligence, can learn quickly and have the patience to watch how and why the market moves, yes you can learn how to apply TA correctly and make money. There are professionals doing it every day.

Most people, however, are not open minded, of below average intelligence and are not patient. For them, TA does not work. Trading attracts some of the brightest minds around and the market will quickly take your money if you don't know what you're doing.

Price is demonstrably not a random walk. This implies that there exists some TA that yields results.

For example, one of the best-studied deviations from a random walk is the phenomenon of 'drift', or trending behavior.

I resemble that remark!

Blah blah blah....

You can't extrapolate past statistical data onto future. Period.

Nonsense!
In trading, there are only 2 outcomes to price movement from the current price rather than 6 or 12 like with dice... Right there odds are more in our favor. Then we have other data like volume. This tells us where more force is being applied, up or down, plus a multitude of other methods for trying to determine what direction comes next. They aren't even close to the same thing.

I use fundamentals to say that as long as Bitcoin doesn't die, the far future should go way up, and that is it. The fundamentals don't mean shit in a mostly speculative market. $2->$1200->150 in 2.5 years is a clear example that fundamentals aren't at play here. One more example would be the numerous large-scale businesses that began accepting Bitcoin over the last year, and only sparked pump and dumps.

Note, this is not a dig at the fundamentals, only that they aren't currently important for the price discovery or used by anyone but bag holders (forced or otherwise) or long term investors (who fall into "otherwise").

Markets with meaningful volume and depth tend to behave in more or less the same manner. There´s profit taking, going long and short, retesting highs and lows et cetera. So, technical analysis is useful - to an extent. Nothing is written in stone. There can be sudden black swans and all sorts of surprises.

TA is just like homeopathy, the amount of science behind it is so tiny that it doesn't produce any visible effects.

Oh and I can also give examples of when TA is bullshit:
That is in low volume markets. Like most altcoins. Don't try to go with TA there. It will most likely not work. Altcoins are just thin markets insider info pump and dumps.

TA is quite helpful in bitcoin trading.
How could anything that gives you more information about what is going on not be helpful?

I agree on using TA with success is a lot about experience, and some specific kind of intelligence (dedication) too.

It is definitely not for everyone. You could see TA as a way to get more insight in the market.
More information is always good. Still you need to know what to do with this information of course...

TA provides you with very important statistics about the past of the market. Projecting it onto future is fundamentally wrong. It is like predicting an outcome of dice throw based on past results. You need to have fundamental data in addition to past statistics to predict a future market.

I didn't say everyone would profit. If everyone is making money, no one is making money. The point I was making with that sentence was that not everyone can effectively apply TA and it comes down to a LOT of variables in understanding what the charts are telling you, why the indicators do what they do, what could come next and what effect the market participants will have on such future outcomes. Simply reading an explanation about the MACD or RSI on investopedia isn't going to make you know how to use them properly. You need to know the formula, for one, but more importantly, why one dump in price can make it go down deep while another dump may only make it go down a little. Basically, there are other underlying factors than just overbought/oversold. Something that takes experience to understand. Something that many people don't have patience for before writing off the whole idea of TA as "bulshit".

True, TA is a tool (sometimes highly overrated)
Fundamentals are more important, especially with something historic like Bitcoin.

There is not one single indicator that will scream BUY or SELL for you, but if you know how to interpret some of the many indicators they could definitely be helpful. Basically the theory behind it is mathematically proven. Still you need to know what to do with this data.

Three words - self fulfilling prophecy. If enough people believe, then it becomes valid.

This sounds exactly like astrology.

I have frequently seen 5 different predictions from 5 different TA guys who are all using the same form of TA. If it was an exact science all their predictions should be identical, and correct. It's obviously not an exact science, and none of them get it right all the time.

More or less I would say that. They are just trying to predict price movement using analysis tools and I think the more time spent trying to understand, the more confusing it gets and most of the time it's not even accurate. I remember seeing ads about paying for the guy to teach how to analyze charts but I was thinking, why try to earn from the participants fee when the guy could game the market by himself.

This is not possible. Trading is a zero sum game? You can't have everyone making money out of it. After costs, most people are guaranteed to lose money. This is true regardless of how many people there are playing the game, surely?

That's what you got from that post?  Lol.

But it can't even work for probabilities of something happening- again, if it was accurate, you could trade on the information (you can arrive at a consistent profit if your 51% chance of a price rise is correct), eliminating the profit and it wouldn't work anymore. You genuinely might as well toss a coin.

It works for me! It's definitely not a 100% guarantee but any edge over the next guy is a definite plus. Also, it's not about predictions, per-se, but more about finding more probable outcomes. For instance, indicators can show a building bias before it is noticeable through forum sentiment (entry points). They can also show weakness from lack of strength and dwindling momentum before the trend ends (exit points).

TA isn't for everyone. If it were, everyone would trade stocks and no one would go to their normal job and brokers/advisers would be out of work. Take Raystonn for instance. He has already proven that he doesn't understand it.

Yes, it's bullshit.  It's just as much bullshit as fundamental analysis, astrology, and central banking.

Predicting the market is objectively impossible. TA is based around % chances of something happening. It's a bit like predicting the weather.

I havn't done any research on this subject, but it seems that it has to be?

If it did work (by work I mean, allow you to predict future short term price movements from past data with more than 50% accuracy) then everyone would do it, and the profit would be eliminated, so it would no longer work. Any method of predicting future (short term) price movements cannot consistently work by this logic, without insider information.

Am I missing something?