Topic: basic TA question (Read 1003 times)

...case closed.

I think of technical analysis and fundamental analysis as being two orthogonal strategies to trading. (Actually I might use the word trading for someone who uses TA, and investing for someone who uses fundamental analysis.) In practice, almost nobody does one or the other exclusively. Pure TA would imply making trades looking at charts, without knowing or even caring what you're trading. What would pure fundamental analysis look like (in theory), with no TA whatsoever? I'm still trying to figure out what that would mean. But it's hard not to try to time the market to some extent, even for a long term investor; so it's probably pretty uncommon for anyone to do "pure fundamental analysis." Most people do some mixture of TA and FA.

Prior to bitcoin, my investment strategy has been to diversify and hold for the long term. I hardly even paid attention to the ups and downs of the market, figuring I was just along for the ride. My job as an investor was not to pick winners and losers, but to make sure my risk stratification made sense for my individual situation. Although this isn't exactly what I did, spreading out among a few Vanguard mutual funds would have suited my investment philosophy pretty well. So I guess you could say that it was guided by the EMH.

With crypto, my strategy has changed completely. I keep up to date with all the major developments with bitcoin. I follow many of the alts. I've even met a few of the leaders in the crypto world. I know that I don't have any information that is not out there for everyone to see. However, I think the information that is out there has not been digested by anything more than a tiny fraction of the financial world. I think that my knowledge and judgement of the fundamentals has alerted me to a good investment that most do not understand, and will not understand for a long time. So when it comes to buying bitcoin and alts, I am not guided by the EMH, because (as a HODLer) I think I can beat the market.

So the question is this: why would my investment strategy be guided by EMH in the first scenario but not the second? It's not because I have any insider information. It's because I trust my analysis and judgement of the fundamentals of crypto better than the analysis and judgement of the vast majority of investors out there. Not only my judgement of the relevant technical issues, but perhaps more importantly, my judgement of the dedication and abilities of the relevant development teams. Indeed, my first purchase of bitcoin in 2011 (only a small purchase, unfortunately) was based in large measure on my impression that Gavin seemed way too smart to be throwing away his career on something that was as crazy as bitcoin seemed at first glance.

Getting back to the EMH, I now think it is possible to beat the market without having information that is not readily available, and the way to do it is to have superior judgement to know which information is the most important. If I were to start my life over as a fund manager, I think I would focus the vast majority of my efforts on the leadership of whatever I was investing in. Maybe this strategy would work better for startups or for small/medium sized companies than for large established companies, I don't know. Obviously the business plan would have to make sense too, but business plans change so I would be more interested in whether I thought the leadership would be likely to adjust the business plan as needed. So I would want to be a judge of human character more than anything else.

Are you saying that you are a believer of the efficient market hypothesis? In that theory, it says that whatever we do, all of us cannot out perform the market. In the case of Bitcoin all we need to do is to buy and hold. Bitcoin is a censorship resistant and not centrally controled. There are many markets Bitcoin can enter, get into and make them efficient. There more it does this the more the price will increase. A good example of this is the dark markets and that is only the beginning.

As a hodler and permabull, I don't make trades based on TA, partly because I don't know if I trust it, and partly because even if it is real, I don't have the time it would take to outperform the pros.

Having said that, I am open to the possibility that there may be a science beneath it that I do not appreciate. As a general rule, if I am unsure whether I believe a given claim (e.g. such and such a pattern predicts such and such a price movement, not with certainty but better than random guessing), then I am more inclined to give it credence if I can postulate some sort of underlying mechanism behind the claim.

And that is the motivation behind this thread.

In that case, then they might be really scamming us into thinking that technical analysis has predicitve value. I just realized that these authors are more concerened about selling books for peofit than making profit in any market themselves. Joining this forum has taught me a lot of things and made me realize that almost every "idea" out there could be out there to scam the clueless.

In a controlled environment, sure it would be easy to trade and see the best tf. Now, in the real world, you have thousands of traders, using 10's of time frames with seemingly infinite amounts of indicator to time to method, setups. It would be impossible to determine what time frame is used the most since (as I said in my last post) you have multiple combinations that people watch and some will agree on completely different tf's (my example where 4hr 20 sma and 1 hour 50 sma are at the same price and both appear to be the support for a bounce), but you can't determine which was the one that was watched more.

Don't get me wrong! I would love to see this data plotted on a graph, but I just don't see it as feasible without seeing what everyone uses most and then it may be different for these 5 stocks than for Bitcoin. Just as every asset class has different systems that work best, the stocks within each class also has things that work where others won't. Bitcoin is no different in this respect. It has its own nuances that make some systems work and others not so much.

@the naysayers
TA has never been a science. It is a practice as medicine is. You form a hypothesis with the info you have available to you, then plan a trade based on that info. As with anything in the world, it's not for everyone, and some excel when others can't. Not everyone can program, or play in the world cup, or race in nascar or make a platinum album. It doesn't mean it is voodoo. Profitability comes to those that understand it and can decipher the info the best. No, it is never guaranteed, but one can create an edge over random walks.

This would actually be a fun website to build. Give everybody X virtual dollars and allow them to buy and sell shares in Virtual Companies A, B, C, D, and E at whatever prices they want. Anybody can play. You have to pay an entry fee, but after a determined amount of time, if you have more than X in your account, you get a reward. Then run the experiments like I describe above.

This could be designed in a way to ensure the website makes money, i.e. 95% of entry fee money gets paid out as reward. Plus, take the data and publish academic papers on TA! lol.

probably not, but it does not make TA a science, just another voodoo, basically, of postfactum explanation for the events that already happened.
TA has very little predictive value, otherwise everyone using it would be rich in opposite to other investors who would be poor, as I said before.
TA tries to explain what happened from a certain point of view plus provide future probabilistic analysis and look for some patterns, that's all.
explain from the TA perspective why first bitcoin bubble stopped at $32, second at $266 and third at 1165? Impossible.

Not sure I follow .... what are you saying would be impossible to calculate? I guess you're saying it would be impossible to calculate which time frame a given trader is using. I agree it would be impossible to calculate with any certainty. Granted, no one trader sticks to one timeframe only. You might focus your attention on the 30 minutes, but that doesn't mean you don't at least glance at the 5 minute and the one hour and the one day charts. So I imagine it would be more like a Gaussian curve than a dirac delta function, with the majority of your decisions based on what's happening in the time frame indicated by the center of the Gaussian curve.

So with the above caveats in mind, the first question is this: is there merit to the idea that a trader who is doing a trade focused on time frame T_i (e.g. T_i = 30 minute chart) is (for all practical purposes) trying to anticipate price movements that are determined for the most part by traders who are engaged in trades focused on a longer time frame, i.e. focused on time frame T_j (e.g. T_j = 1 day chart) where T_j > T_i ?

It kinda seems intuitive to me that the answer would be yes, in which case we could begin to wonder whether there are any useful or interesting implications of this fact.

Here's an idea. Suppose somebody did an experiment with a large number of traders competing against each other using a handful of virtual stocks on a virtual exchange. That is, the only people trading on the exchange are traders who are part of the experiment, and the stocks they are trading exist only in this exchange. Now, suppose you give the traders access to current price, current order book, and 30 minute charts but no way to look at any other charts. The people running the experiment, however, can look at whatever charts they are interested in. I would hypothesize that if you look at the 1 minute or 5 minute charts, which are available to the people running the experiment but NOT available to any of the traders, then you would see exceptionally clear, clean, easy to interpret TA patterns, which give exceptionally clear buy and sell signals. But if you repeat the experiment, and this time allow the traders to access any chart that they are interested in as they are trading, then you will still see TA patterns, but they will be more ambiguous.

If true, then this could provide an experimental tool for academics to discover and demonstrate "idealized" TA patterns, much cleaner than what you would generally see in the real world.
 

It becomes difficult to process that kind of data when different people use different combinations of indicators/lines/methods which they trust to tell them to buy or sell. Maybe one of the big miners can do it?!  It is sheer coincidence that, for ex., the 4hr 20 and the 1 hr 50sma are both hard support in a dip. Which one is the one that was used? While it would give an advantage to he who has the chart in front of them, it sounds nearly impossible to calculate with any certainty.

OK, I think that answers my question. Sounds like it does make sense to talk about how much total capital is allocated at any given point in time towards trading within a given time scale. The reason I'm asking is that I'm wondering whether knowledge of this distribution could be useful.

To elaborate ...

Define S(T) as the sum of all capital allocated and "in play" right now to trades in time scale T. T is defined as the length of time of your chosen candle for that trade (e.g. T = 5 minutes, T = 1 hour, T = 1 day, etc). (Alternatively, I could define T as the size of the window that commands your attention for that trade, iow T = the time spanned by 140 candles.) S(T) is summed over all capital that is in play in the here and now; that means either a trader who has money in play (or is contemplating putting it into play), and also would refer to any active bot whose algorithm is based on data from the charts with candle width T.

Suppose you knew the shape of S(T). What would that tell you? Suppose, purely hypothetically, you knew that S was flat across all T except for a dip at S(one hour). Or suppose S(T) was usually flat, but S(5 minutes) or S(30 minutes) takes a big dip when it is midnight in China, because Chinese traders have a particular preference for those time frames and don't keep trades open when they sleep.

If nobody is trading on the 30 minute charts (hypothetically speaking), is it possible that that would influence the reliability of TA using the one minute charts? If S(T2) is very low, could that influence the reliability of the TA you are using for your trades using time scale T1 where T1 < T2 ?

TA is basically psychology. OK then, whose psychology? I postulate that by looking at the 5 minute charts, you are attempting to peer into the minds of the traders who are looking at the 10 minute or the 30 minute or the hourly charts, in the hopes of predicting what they are about to do just before they do it. TA tells you how to front-run the traders who are operating on the time scale just above yours.

It's like you are a native American Indian with your ear or your hand on the railroad, feeling the vibrations of the oncoming train. When you stare at the 5 minute charts, you are feeling the rumblings and vibrations so you can anticipate bigger moves made by the traders who are staring at the 30 minute charts.

So if nobody is looking at the 30 minute charts, then it is pointless to stare at the 5 minute charts, and your TA won't work. But if LOTS of people are staring at the 30 minute charts, then TA should work especially well on the 5 minute charts. Especially if nobody ELSE is staring at the 5 minute charts, which would mean you have less competition.

So if you knew the shape of S(T), then you would predict TA to be the most effective at at time scales T where S(T) is low, but then gets larger for larger time scales (larger T). Wherever the slope of S(T) is maximal (dS / dT), that's where you want to be doing your TA.

How does one go about gathering knowledge of the shape of S(T)? I dunno. Maybe a Fourier analysis of the price would be the place to start.

Whatdya think? I'm totally making this up as I go along, in case you couldn't tell lol ...

 

If that is the case then would it be valid to say that all those authors who wrote all the books about technical analysis are scamming the trading newbies like us into thinking that we too can be winners in trading only if we learn it?

When I say "they are scamming us" what I mean is that those authors know that what they are writing about does not really work.

For me it's about the time I can afford to devote to trading at the time in question. At the same time, I may have a long term trade (cold storage), A medium term trade (My wave B since the lows), and one or more short term "daytrades" where I may hold from hours to days. Now, consider this, a 5 minute chart is good for upto hours in a trade. a 30 minute chart is a good time frame for a day. Then a daily chart if I plan to stay in for up to weeks or maybe months. Figure on around 140 candles visible and depending on the amount of time included in those candles should be about the middle road for that candle size (in time). As you deviate from that 140 candles, it can become less accurate in a sense because now you are looking back to irrelevant data for the time frame.

So, if I have time to sit and trade today but not tomorrow, I would check most time frames just to get the overall feel and then trade directly on 5min to 1 hour charts depending on the volatility/price action.

That is a topic for a different thread ... I'm interested in whether traders themselves consciously operate -- in the context of a given trade -- within the parameters of a particular time scale. In theory, in practice, both, neither.

Or perhaps it's a question that hasn't occurred to them ...

TA does not work (mostly), otherwise all TA experts would be insanely rich already which is typically not the case.

To all the TA experts out there (masterluc, RyNinDaCleM, chainsaw, chessnut, and the many others I am leaving out!), I am wondering whether it makes sense to talk about the time frame that a given trader is working in, in the context of a single trade.
- When trading, do you consciously pick a timeframe? i.e. do you pick a candlestick time interval and stick with it for the duration of the trade?
- Suppose a trader makes a buy based on a pattern in the hourly charts, but then makes a sell based on a pattern based in the 5-minute charts. (Or vice versa). Would that be considered poor practice? Would it be evidence of an undisciplined trader? Or would you encourage a trader to look at multiple timeframes within a single trade? Or does it matter one way or another?