Anyone thinking about doing doing some technical analysis may well become less enthusiastic about the idea if they first played around with generating some random time-series with statistics similar to the real thing and then dwelling on the fact that they look pretty close to the real thing. GARCH(1,1) isn't a bad place to start (although it's not perfect). I just put up some R code on github for this here: https://github.com/mhowlett/garch11 which I was experimenting with in the past. I can't remember the state of it exactly, but I think it works.
Technical analysis is far from useless. As an active trader, it's the best tool I have come across. I'm sorry, but to say that it is useless is to admit pure ignorance, or to admit your failure to grasp the concepts and understand how to apply them visually and correctly.
People do apply techniques incorrectly, yes. They can get confirmation bias from it, yes. However, that does not invalidate successful traders who trade primarily on technicals. Keep in mind that a "tool" is just that - but not used the same way by everyone - and "this means that" isn't a blanket rule that always works as described. Everything is part of a larger structural puzzle. Also, HFT algos will eat your lunch if your hold period is too short. I don't hold a position for more than a week, MAYBE two, but never less than 45 minutes - and as a retail trader, that makes a huge difference in P/L ratios over time.
Through my trading experience I have continuously used less fundamentals and more wonky technical techniques I have come up with. It has given me a massive edge over whoever is on the other side of the trade, and has pulled me quite a long way out of debt and into making a nice living doing it on the side. The side of the story that you don't see in the endless "get rich quick" trading schemes that promise to teach you all of those super awesome secrets is that anyone who knows what they're doing isn't spending any time teaching it to the masses.
But, to get back on topic: Well done with this site! It's a great beginner guide to understand what a lot of the basics actually mean. Babypips falls very short.
If that is the case, it would create a market inefficiency which some other market participant could come along and exploit (anticipate the TAs moves and act preemptively). Your statement is basically saying there is no one out there smart enough to do this.
Not necessarily! Suppose a bunch of noise traders are piling into a stock because all seven green lights are on in the software they bought for 29.95 from an infomercial. Let's take as an assumption that this algorithm has no idea what it's doing. The smart money thing to do would be to buy now (since investors are going to keep buying as they notice their text alerts or sound alarms or whatever tells them to buy the stock) then sell again knowing full well you are driving the price further from fundamentals in the short term. The best response to past irrationality is to trade against it. The best response to future irrationality is to trade with it. That irrational behavior might be forecast-able should not be a huge surprise.
Momentum is a well established empirical regularity in stock prices. Knowing this, you trade with the momentum at then trade against it later. This does not help momentum go away.
What do you think a good portion of HFT algos do? Of course people are exploiting the order flow that comes as a result of generic "signals".
But my point is that technical ANALYSIS is the utilization of TOOLS. People apply tools in different ways. Especially with fibonacci, there is certainly not one way to trade using fibonacci retracements, extensions, time, harmonics, etc.. I know traders who use it completely differently than I do. What matters is how it's applied to a chart and how trading decisions are made based on how the trader interprets what they're seeing.
It sounds like you guys are describing things like MA crosses to be useless. How they're normally applied by new traders is "the red crossed the green, so I sold", then they get mad when they lose money on the trade (and blame technical analysis). Yet, a moving average is a tool which can be tweaked in numerous ways - how you tweak/apply/interpret/trade using that tool is an entirely other question. Nobody in the right mind should be trading using only one tool, or by applying an "out of the box" charting tool then just blindly making decisions based on that.
Liken this to a software context: "Sensitive data can get stolen, so software is BAD". Yet, it's really just how it's written, up/downsides that are inherent of the language it's written in, but mainly HOW it was designed and how businesses utilized or tweaked it. You can't generalize the entire software industry because of security holes.
If you reduce technical analysis to generic "super awesome signal, works every time!" marketing ploys, you'll miss one of the best edges you could possibly have in trading/investing.
Note: This disagreement is a perfect example of why volume exists on each side of the market.
Technical analysis doesn't have any science backing it up. However, as most pop books are some form of swing, candlestick, Bollinger bands, mean reversion, it creates a self-fulfilling prophecy.
If you're going to play the markets, it's quantitative analysis or go broke. And that is a literal go broke!
Oh wait we're economists - free markets are sacrosanct and they are efficient - if there is an arbitrage opportunity to short surely the rational investors will price it correctly. Not.
Just because I said that technical analysis is bullshit does not mean that I think EMT holds.
I can also say that using the times bulls shit in Colorado to predict stock prices is bullshit. That once again does not mean that I think EMT holds. It just means that the technique I called out has no predictive value.
The claim "technical analysis doesn't work" is the weak form EMH.
So yes, you are being inconsistent, unless I've mistakenly interpreted "technical analysis is bullshit" as "technical analysis doesn't work". Is it your belief that technical analysis does work, but is still bullshit?
Actually, I don't believe this is correct. The weak form EMH claims that any information provided by the price history of a security should already be incorporated into the current price. You can claim that Technical Analysis (construed as patterns in charts and the like) isn't valid and that weak form EMH doesn't hold, if you believe that charting fails to correctly incorporate the relevant information from the price history.
Like most things in life, it isn't quite that simple. There are real inefficiencies in the market which can be highlighted by TA. TA can also be considered a form of Statistical Arbitrage and vice versa.
A simple example: There are many traders out there using TA, if sufficient numbers of them start following the same signals then real market impact occurs. You can use TA to establish what signals they are using and then front run them. This shows that TA has real power in some circumstances and there are many more out there.
I could go into a lot more detail, but am pressed for time at the moment.
> Like most things in life, it isn't quite that simple.
Like most things in life, technical analysis is bullshit.
> There are real inefficiencies in the market which can be highlighted by TA. TA can also be considered a form of Statistical Arbitrage and vice versa.
Stat arb is not what is commonly known as technical analysis. They are distinct.
> A simple example: There are many traders out there using TA, if sufficient numbers of them start following the same signals then real market impact occurs. You can use TA to establish what signals they are using and then front run them. This shows that TA has real power in some circumstances and there are many more out there.
Keynesian beauty contests do not provide consistent alpha. If they did, we'd see fewer VCs flame out.
Stat arb is not what is commonly known as technical analysis. They are distinct.
No. Technical analysis is forecasting prices based on past data. Stat arb is a particular form of technical analysis, based mainly on looking for mean reversion in sets of securities.
Any strategy lies somewhere on the continuum of strategies ranging from purely technical (looking at past prices only) to purely fundamental (looking at fundamental data only).
I have been a trader for 15 years, I have seen TA absolutely working and I have seen it failing. You can take the opinion of someone that has walked the walk, or not. I have given you a specific example where TA does work, so it can't all be bullshit right?
The example I gave is a logical one, which just happens to also work on occasion in the real world. It does not rely on what sort of trader I am, it is a thought scenario, which can and does work in the real world.
Logically, if a large number of people are using a specific TA signal to make trades, their trades will move the market to a degree. Then if one knows what that signal is, one can reliably make money using the same TA signal.
So even if the actual TA signal is a completely random walk, an alert person can make money from that TA signal if they understand that the market will move because some powerful players are using the same signal for whatever reason.
TLDR: A bullshit signal can be used to reliably make money.
How often do these occasions arise? If I used this technique every day, would I last a week before going broke? Give us some indication that this is more valuable than flipping a coin.
If you dont believe in technical analysis at all, what do you use to trade? Are you relying purely on fundamentals, following trends? Or are there other types of analysis that can be applied?
The only important indicator on a chart is price, and volume in some cases to confirm price. But all decisions need to be based on price action. Chart analysis is just a tool to read the emotions of other traders, that’s it.
Think about it, you see a 3 day rally with wide range bars and on day 4 you get a narrow range bar where the open and close are the same (or very close). What does that tell you? Momentum has stopped and people are undecided. There’s nothing subjective about that. After this the stock can move either way, but you are getting an indicator that something’s going to happen and you have to monitor closely. You can tighten the stops or exit completely. The decision is up to you. But you are using the chart to make an educated decision.
Another example, after a 3 day pullback on day 4 you see a narrow range bar with a very long tail. This is called a Hammer [http://en.wikipedia.org/wiki/Hammer_(candlestick_pattern)]. This tells you that at some point during the day the sellers were in control but something happened and buyers took control and raised the price. This was a war and the buyers won. This usually changes momentum and leads to a rally. An explanation for this could be that at some point during the day many stop loss orders from long term investors were triggered and short-term buyers see it as an opportunity to buy and raise the price.
I just gave you two examples where Technical Analysis can be effective. Does it work all the time? Of course not. But you have a better picture of what’s happening, therefore your odds are higher and you can use this information to lower your risk.
Finally, you have to remember that for every buyer there’s also a seller and vice versa. When you buy a stock (long) you have to ask yourself. Is the person selling me the stock profiting or taking a loss? Is the other person a beginner, professional trader or institution? You have to find scenarios where you buy the stock from the beginners taking a loss (even if it sounds cruel). Given enough time and hard work you can develop experience necessary to spot where all these people buy and sell and use this information to your advantage.
Backtesting[1] is really critical when deciding on whether to use a given technical indicator.
I've read too many books and articles, and heard too many anecdotes that make some technical indicator sound fantastic. But until you backtest it (and then forward test it) against a large amount of data, and make sure you're not curve-fitting, I don't think you can have much confidence in it actually working in the real world in the long run (though you may get lucky and have it work for a brief period of time).
For backtesting, the best tool I've found is AmiBroker[2]. It's orders of magnitude faster than any other backtesting tool that I've found, and lets me backtest mountains of data against any indicator I can dream up within seconds or minutes, and then go on to the next one. It's pretty awesome (though kind of quirky in that its programming language is array-based). Anyway, highly recommended.
Can you provide any useful information other than a cool sound bite? Is there a better way to analyze equities? There is a lot of computer trading. Must be some analysis happening. I'm sure a lot of us here would like to be pointed in the right direction.
Technical Analysis can be useful, because it enables a systematic approach to trading - eliminating subjectivity when making trading decisions. Consistently exploiting some edge with a well defined set of rules and conservative risk management is key.
The guide opens with an analogy suggesting that picking the best stocks to invest in is equivalent to picking the best restaurant to eat at, and that technical analysis is equivalent to looking for the stall with the most people at it.
If technical analysis is equivalent to looking for the restaurant with the most people at it, then the restaurant that you're trying to find is the one that, over the period that you intend to eat there, will gain the most new customers, relative to the number of customers it had when you walked in, not the one that's best to eat at.
The flawed analogy exactly explains the problem with technical analysis.
> "If technical analysis is equivalent to looking for the restaurant with the most people at it, then the restaurant that you're trying to find is the one that, over the period that you intend to eat there, will gain the most new customers,"
Which pretty much is what you want to achieve in certain markets.
For instance, if you are trading with CFDs and want to close your positions in the short term, you want to predict the right direction in an instrument with high volatility, which tends to be correlated with the amount of people trading in the same instrument (restaurant with the most people for a given period of time)
> The results were positive with an overwhelming statistical confidence for each of the patterns using the data set of all S&P 500 stocks daily for the five year period 1992-1996.
The fact that line even has to be used is a massive red flag.
> overwhelming statistical confidence for each of the patterns using the data set of all S&P 500 stocks daily for the five year period 1992-1996.
And tech VCs investing in anything from 1995-2000 were positive with an overwhelming statistical confidence that their portfolios were anything but randomly chosen.
