Hypothesis: Technical Analysis only works in the immediate short-term

[Handle123]

On the topic of pricing already factoring in fundamentals, could one not argue that price also factors in technical analysis? It's paradoxical in a way, but I think it sort of supports the old adage that it is a self-fulfilling prophecy. I tend to think in both cases, it is the lack of consensus that leads to profitable opportunities (e.g: the market priced fundamental information incorrectly or investors are viewing the same price action incorrectly). Of course, I tend to think the process of predicting it correctly even with a minor edge is a ridiculous challenge and is why index investing is so enticing...

Anyway, I really like your mention of charting by hand. Although I do not know if I will go to this extent, I could see myself creating the charts in Excel using my own inputs instead of simply using charting software - I agree completing tasks yourself helps you learn better, both in terms of speed and retention. Thanks for your input regarding noise too, which seems to be a hotter topic than I thought lol.

Well, most of TA is calculated on Price, enough volume going same way, and if it is December 24th, I will think back that Santa is real. Yes, TA is very much like "cult" like of people doing what they believe in whether back tested or they have their lucky rock in there 2nd pocket and did three turns on my left leg. Hey, when you losing most days in beginnings you will try anything.....I once wore same socks for 34 days in a row as I didn't have a losing day during this oddity.

One thing I did long ago was to take one chart that has a little trend and chop, then, this will take years, give each bar a name, and each have to make sense to you and has to be back tested so this bar works today but is not completely wrong tomorrow, as years go by you might add a name or more, I doubt in my lifetime I will understand all the bars, but when something happens and it or several bars create a pattern that is better than 50/50, then I have something to spend more time.

I often think using fundamentals can be considered TA, numbers are technical and if makes a rule based on debt or increased sales, it can become another piece of the puzzle to help if one is into it. But I use 100% charts. I have even made a system that trades Reports, but doesn't matter at all what was reported, just based on charts. And no, not breakout entries.

Much luck to you.

 

[eurusdzn]

This post or article seems relavant to this discussion.

https://www.elitetrader.com/et/threads/article-by-pabst.125197/

 

[CyJackX]

This is the best answer I've seen thus far as to why using TA for the long-term is a good idea, although it still feels much more like a confirming tool than it feels like a predicting tool. Any thoughts on that, CyJackX? You mention it gives you idea of how all players are moving because movements are smoothed out over longer time-frames, which I agree with. Do you think that gives you enough meaningful information to predict, with any edge, future movements? Or do you feel it is a tool best used for gauging and reacting rather than anticipating? Thanks.

I mean, TA is predicated on the theory that it gives you an edge in predicting future movements.
I may be being pedantic, but I also don't understand your question. What's the difference between "gauging and reacting." and "anticipating?" Or "predicting," etc?

I would trust a TA pattern with above average volume more than I would one with low volume. But, on a smaller timeframe, a high volume bar can be spoiled later by even higher volume, causing you to wonder which one to trust. Longer timeframes are less susceptible to "random" volume spikes, as they are usually due to news events.

But keep in mind the HFTs and hedge funds are very well aware of the patterns on the 1-5 charts and they are experts at creating false breakouts to trap the twitchy day traders by swinging price violently the other way.

Still not on board with this narrative. If all the big players know the traps, there can be no trap. If retailers are the lemmings, Firm A must expose itself to the market in order to trap them. And Firm B, if hypothetically better equipped than A, then knows how it can trap Firm A, who is probably a bigger target than all retailers combined. And this isn't accounting for firms C-Z, all who want a slice of the pie. I just can't believe that any "obvious" trap for market participants would not be hedged or arbitraged away. Also still saying that all retailers combined are barely worth chasing.

 

[globalarbtrader]

Disclaimer: I'm a (relatively) fresh graduate with a major in Finance, so as might be expected, I primarily learned about fundamental analysis and corporate finance for the last four years of my life. Therefore, there might be some bias in what I say. TL;DR at bottom.

For the last few months, I've been exploring into the territory of utilizing technical analysis (TA) and price action theories to explore trading Forex (and to a less extent, equities). One of my issues with technical analysis is that while some components are easy to understand the logic behind (such as support and resistance levels or momentum), other components seem (keyword "seem" - this is strictly my opinion) less logical such as Fibonacci levels. However, I can simply disregard indicators or chart patterns that I do not feel comfortable with.

Where I feel I differ from most people is where I believe the best time-frames are for utilizing TA. I see many traders and investors alike utilize it to identify long-term trends and intermediate trends. While I agree these trends are common, I do not understand the logic behind extrapolating such trends with a method like TA.

Prices are driven by supply and demand. In the long-run, fundamental changes in the investment vehicle in question (or its environment) are what will cause the largest shifts in supply and demand. Technical analysis cannot predict surprises (nor can fundamental analysis for that matter, although I believe it can prepare you for them better by identifying specific risks and allowing you to calculate for them accordingly). Given this knowledge, why do so many use TA in their long-term analysis of the future?

I feel TA is best suited for environments where the chance of fundamental surprises to the underlying security is low, or for identifying sentiment following a surprise. Therefore, my thought is that TA is best suited for the immediate short-term (approximately a month or less) instead of the long-term future. It feels erroneous to extrapolate on price and volume alone for the long-term.

Tell me why I'm wrong - I would love to hear from other people who have more experience on the subject. Thank you.

TL;DR: My hypothesis is that TA works best in the short-term because in the long term, fundamental surprises are what drive shifts in supply and demand. Hence, when we extrapolate on the basis of price and volume alone, we are exposing ourselves to dangerous beliefs that could instantly be eradicated by fundamental changes in the underlying security (while this can also happen in short-term, it is less likely).

P.S: This thread is about the use of using TA by itself. I believe TA can only be used standalone in the short-term, whereas it requires assistance from FA in the intermediate/long-term to identify and prepare for specific risks.

Just to add two points to this:

Firstly at least using the kinds of technical rules that I use (trend following rules which can be systematically tested rather than magic charting voodoo which cannot) the efficacy of these starts to fall off once you have holding periods of less than a couple of weeks. This happens in the backtest about 30 years ago. It's a very clear pattern- the shorter the holding period the worse they do. Admittedly I haven't tested these on an intraday basis, but it would at least be surprising if they flipped back to profitability magically against the prevailing pattern.

Fundamentals seem to work better over the long term, but here long term means holding periods of at least 6 months and preferably years. To give you a concrete example if I build a model which uses relative momentum and another which uses relative dividend yield, then across global equities the former models does best with a one month rebalancing and the latter with a one year rebalancing.

The second point relates to transaction costs. If you are running a very fast trend following system you need to cross the spread every time you trade (otherwise you'll miss the trend). Doing this several times a day means you need to generate huge pre-cost Sharpe Ratios before you start making any money. If your holding for longer then you can passively execute most of the time and capture the spread.

In contrast if your fast intraday system operates on mean reversion (which characterises most fundamentals systems but also market making) then you can passively execute and capture the spread unless you've hit a stop.

In summary I'd say for very short time frames and very long fundamentals / mean reversion is best. In the middle trend following is best. The size of the middle section depends on the asset class. In equities the sweet spot for trend following appears to be narrower, with holding periods of at least a month up to a year. In futures it's a little wider.

GAT

 

[Investor]

This post or article seems relavant to this discussion.

https://www.elitetrader.com/et/threads/article-by-pabst.125197/

eurusdzn, This article was a great read; thank you for digging it up for me.

Handle, I appreciate your insight as always. It seems to me you believe in trading the best setups (and ones that make sense to you), which I imagine requires a great deal of patience so keep on keeping on.

CyJackX - gauging and reacting is like confirming and lagging indicators. Anticipating is like a predicting indicator... In other words, do you think TA is better for confirming suspicions and reacting to changing markets, or do you feel it can be used to adequately predict long-term price direction. It seems you believe the latter, which is good because I tend to think there is much more value there.

GAT, thanks for your input. Your specific advice is valuable and it is interesting to me the results you've found in your back-testing. I tend to agree that shorter-term strategies will rack up a ton of transaction costs that make profitable trading much more difficult than it might be for intermediate to longer term strategies (not to say it's impossible, just more difficult). Your last paragraph is something I will keep in mind as I keep my eye on the markets. Thanks again!

 

[Money Trust]

Hate to be all over you but there is no way an intra day chart holds equal weighting to price crossing a previous quarters high which is a signal nearly every hedge fund and institution is monitoring world wide. Intra day is noisy - arbitrage, day traders, hedging, etc. Only the closing price reveals the players that are holding.

That's your opinion and nothing more.

 

[learner2007]

In this thread, as in 1000 other TA related threads, 'to predict' is so often mentioned as being what TA is supposed to do. It's used by so many to 'predict' where the Dow will be at year's end, or how much Apple will be on your dog's birthday.
Dow Theory was created to identify the direction of current business conditions and to 'predict' the continued direction of such until a change could be identified. It was not created as an aid in predicting the extent of the price movement of a stock.
Those who built on Dow Theory and developed old school TA did use the word 'forecast', but not in the sense of prediction(of where a share price will be in the future). They used that word to mean the direction of the established trend and its continuation. Their objective was to identify the establishment of a trend and enter a trade at an opportune point within that trend in order to ride it, and not to buy because of a 'prediction' that the stock was going to $80.
'Prediction' is one of the biggest pitfalls that a trader or investor can employ. And, of course, when the prediction isn't reached, the first words are, "Ah,that TA stuff doesn't work".

 

[Mtrader]

'Prediction' is one of the biggest pitfalls that a trader or investor can employ. And, of course, when the prediction isn't reached, the first words are, "Ah,that TA stuff doesn't work".

Prediction should be replaced by "probability to have an expected outcome". And outcome does not mean clear result, but rather the direction of the move.

As there are never certainties you should work with probabilities. Probabilities should improve your chances to make money without any guarantee, as probabilities are probabilities and not certainties.

 

[learner2007]

Prediction should be replaced by "probability to have an expected outcome". And outcome does not mean clear result, but rather the direction of the move.

As there are never certainties you should work with probabilities. Probabilities should improve your chances to make money without any guarantee, as probabilities are probabilities and not certainties.

Right! And the probability of continuation or reversal is built into the dependability of the given set up. But as is so often seen, the prediction, before the establishment, creates an objective that blinds the trader to a change in future conditions, in addition to most of those predictions being based on little more than guesswork rather than proper analysis.

 

[bashatrader]

Investor,

You are correct and negative reactions to your hypothesis usually come from failed traders in a state of denial.

The old TA is out for good, failed and done. New TA is machine learning and we are trying to forecast next period returns of one to a few days. This is what we do in the hedge fund I work for. Average monthly return is >+10%. We are happy to take the money of old TA traders.

One of the top experts on why old TA does not work is Michael Harris. He is author of Price Action lab Blog. He has written an article about how old TA was promoted by the financial industry for own interest. Search Google for the article with title 'Technical Analysis Has Contributed To A Massive Wealth Redistribution'. The whole truth is there.

Deep learning networks is the way to go now. Most else is dead and unprofitable.