# Is candlestick charting an effective trading tool in timing the markets?

In the world of technical analysis, is candlestick charting an effective trading tool in timing the markets?

If used correctly, how accurate can they be in picking turning points in the market?

• See my updated answer; I provided you with some useful links about TA. Since TA forms the basis for the patterns candlestick charts seek to represent, any evaluation of TA is more than appropriate for evaluating visual tools like candlesticks. I finally found my copy of the Aronson book, and he makes the point that TA patterns like head-and-shoulders are actually worse than random. You'd be better off making a random trade whenever you would have used a h&s pattern. – John Bensin May 26 '13 at 13:46

I am strongly skeptical of this.

In fact, after reading your question, I did the following: I wrote a little program in python that "simulates" a stock by flipping a coin. Each time the coin comes up heads, the stock's value grows by 1. Each time the coin comes up tails, the stock's value drops by 1.

I then group, say, 50 of these steps into a "day", and for each day I look at opening, closing, maximum and minimum. This is then graphed in a candlestick chart.

Funny enough, those things look exactly like the charts analysts look at. Here are a few examples:

If you want to be a troll, show these to a technical analyst and ask them which of these stocks you should sell short and which of them you should buy.

You can try this at home, I posted the code here and it only needs Python with a few extra packages (Numpy and Pylab, should both be in the SciPy package).

In reply to a comment from JoeTaxpayer, let me add some more theory to this.

My code actually performs a one-dimensional random walk. Now Joe in the comments says that an infinite number of flips should approach the zero line, but that is not exactly correct. In fact, there is a high chance to end up far from the zero line, because the expected distance from the start for a random walk with N steps is sqrt(N).

What does indeed approach the zero line is if you took a bunch of these random walks and then performed the average over those.

There is, however, one important aspect in which this random walk differs from the stock market: The random walk can go down as far as it likes, whereas a stock has a bottom below which it cannot fall. Reaching this bottom means the company is bankrupt and gets removed from the market. This means that the total stock market, which we might interpret as a sum of random walks, does indeed have a bias towards upwards movement, since I'm only averaging over those random walks that don't go below a certain threshold. But you can really only benefit from this effect by being broadly diversified.

• Another big difference are the support and resistance levels and trends, the random walk does not have them, do you think something like the Amazon graph can be created with a random walk? and some level prices have a psychological effect and react more than the others. Anyway that does not mean we can predict the next price direction. – Enrique Dec 10 '18 at 2:53

• In the world of technical analysis, is candlestick charting an effective trading tool in timing the markets?

It depends on how you define effective. But as a standalone and systematic strategy, it tends not to be profitable. See for example Market Timing with Candlestick Technical Analysis:

Using robust statistical techniques, we find that candlestick trading rules are not profitable when applied to DJIA component stocks over 1/1/1992 – 31/12/2002 period. Neither bullish or bearish candlestick single lines or patterns provide market timing signals that are any better than what would be expected by chance. Basing ones trading decisions solely on these techniques does not seem sensible but we cannot rule out the possibility that they compliment some other market timing techniques.

There are many other papers that come to the same conclusion.

• If used correctly, how accurate can they be in picking turning points in the market?

Technical analysts generally fall into two camps: (i) those that argue that TA can't be fully automated and that interpretation is part of the game; (ii) those that use TA as part of a systematic investment model (automatically executed by a machine) but generally use a combination of indicators to build a working model. Both groups would argue (for different reasons) that the conclusions of the paper I quoted above should be disregarded and that TA can be applied profitably with the proper framework.

Psychological biases

It is very easy to get impressed by technical analysis because we all suffer from "confirmation bias" whereby we tend to acknowledge things that confirm our beliefs more than those that contradict them. When looking at a chart, it is very easy to see all the occurences when a certain pattern worked and "miss" the occurences when it did not work (and not missing those is much harder than it sounds).

Conclusions

• the only ways to know if you are "using it correctly" are to either paper trade it honestly (no "I missed that signal yesterday": only include trades as you would have done them in real time) or backtest your strategy
• a good technical analysis system without a proper money / risk management framework will generally lose over the long term
• markets are extremely competitive: it is highly unlikely that a simplistic strategy would oupterform over the long term
• markets are extremely competitive (did I say that already?): it is very difficult to find profitable strategies, even for sophisticated participants
• I thought I remembered reading a paper that discussed this but I didn't have a chance to find it. +1 for the point about confirmation biases too; humans are excellent at picking out patterns. Sometimes a little too excellent. – John Bensin Apr 16 '13 at 16:46
• The whole point in TA is to use complimentary techniques to predict market movements. These can be confirmation of trends continuing or confirmation of them ending & reversing. Most competent TA do not suffer from "confirmation bias" as they understand that they are not in control of the market & that the market is unbiased. That is why TA do use money/risk management to protect their capital & profits. They are not after 100% success rate, most would be happy with a 50% success rate. What they do is make profits by letting their winning trades run & keeping their loosing trades to a minimum. – Victor Apr 16 '13 at 23:14
• @Victor Did you read the study assylias linked to? Why ask this question if you don't plan on examining the evidence? Although the study isn't conclusive, numerous studies have been done, and the evidence isn't exactly 100% in the favor of TA. – John Bensin Apr 17 '13 at 2:39
• @JohnBensin, I have skimmed through it, they view on a short term use of it - up to 10 days. My question is who is funding the research? I have ready many different so called studies on the effectiveness of TA and market timing, and the assumptions they use are not in anyway related to how TAs actually use their techniques. – Victor Apr 17 '13 at 4:38
• @Victor I did not conclude that you can't find a profitable TA strategy. I just highlighted that simplistic approaches generally won't work. Also don't overestimate the number of technical analysts who actually can make money. – assylias Apr 17 '13 at 6:37

I interned for about six months at a firm that employed a few technical analysts, so I'll try to provide what little information I can. Since the bulk of the intra-day trading was decided algorithmically, technical analysts had two main functions:

1. Working with the researchers to translate potentially profitable chart patterns into algorithms for the trading machines to run.
2. Watching for patterns in securities that weren't already monitored by the firm's trading algorithms. In this case, the firm relied on them to either make the trades or instruct the computers to watch that security, based on the algorithms developed through point 1.

This basically boils down to my answer to your question. There are still enough people, trading firms, etc. who believe in candlestick charting and other visually subjective patterns that if you notice a trend, pattern, etc. before the majority of traders observing, you may be able to time the market successfully and profit. This is becoming increasingly dangerous, however, because of the steps I outlined above. Over time, the charting patterns that have been proven effective (often in many firms individually since the algorithms are all proprietary) are incorporated into computer algorithms, so the "traders" you're competing with to see the pattern are increasingly low-latency computer clusters less than a few blocks from the exchange.

Summary:

Candlestick charting, along with other forms of subjective technical analysis, has its believers, and assuming enough of these believers trade the standard strategies based on the standard patterns, one could conceivably time the market with enough skill to anticipate these traders acting on the pattern and therefore profit. However, the marginal benefits of doing so are decreasing rapidly as computers take over more trading responsibility.

Caveats:

1. I know you're in Australia, where the market penetration of HF/algo traders isn't as high as in the US, so it might be a few more years before the marginal benefits cease to be profitable; that being said, if various forms of technical analysis proved wildly profitable in Australia, above and beyond profits available in other markets, rest assured that large American or British trading firms would already have moved in.

2. My experience is limited to one trading firm, so I certainly can't speak for the industry as a whole. I know I didn't address candlestick charts specifically, but since they're only one piece of visual technical analysis, I tried to address the issue as a whole.

3. This somewhat ties into the debate between fundamental or technical analysis, which I won't get into. Investopedia has a short article on the subject. As I said, I won't get into this because while it's a nice debate for small traders, at large trading firms, they don't care; they want to make profit, and any strategy that can be vetted, whether it's fundamental, technical, or astrological, will be vetted.

### Update

I want to add more information to my answer to clear up some of the misconceptions in the comments, including those talking about biased studies and a lack of evidence for or against technical analysis (and candlestick charts; I'll explore this relationship further down).

It's important to keep in mind that charting methods, including candlestick charts, are visually subjective ways of representing data, and that any interpretations drawn from such charts should, ideally, represent objective technical indicators. A charting method is only as good as the indicators it's used to represent. Therefore, an analysis of the underlying indicators provides a suitable analysis for the visual medium in which they're presented.

One important study that evaluates several of these indicators is Foundations of Technical Analysis: Computational Algorithms, Statistical Inference, and Empirical Implementation by Lo, Mamaysky, and Wang. Lest anyone accuse its authors of bias, I should point out that not only is it published by the National Bureau of Economic Research (a highly reputable organization within economics and finance), but also that the majority of its authors come from MIT's Sloan school, which holds a reputation second to none.

This study finds that several technical indicators, e.g. head-and-shoulder, double-bottom, and various rectangle techniques, do provide marginal value. They also find that although

human judgment is still superior to most computational algorithms in the area of visual pattern recognition, ... technical analysis can be improved by using automated algorithms

Since this paper was published in 2000, computing power and statistical analysis have gained significant ground against human ability to identify and exploit for visual pattern detection like candlestick charts.

Second, I suggest you look into David Aaronson's book, Evidence-Based Technical Analysis: Applying the Scientific Method and Statistical Inference to Trading Signals. He finds similar results to the Lo, et. al. paper, in that some technical indicators do add value to the investment process, but those that do are those that can be represented mathematically and thus programmed directly into trading algorithms (thus bypassing visual tools like candlestick charts).

He describes how studies, including Lo, et al., have found that head and shoulders patterns are worse than random, i.e. you would earn higher returns if you simply traded at random. That point is worth than repeating. If a day-trader is using a candlestick chart and using head-and-shoulders patterns as part of their toolkit, he's rolling the dice when he uses that pattern and returns that come from its application come from chance.

This reminds me of that old story about a company that sends out pamphlets predicting the results of sports games, complete with "strategies" and "data" to back up the predictions. The company sends out several versions of the pamphlet every game, each predicting a different winner. Given a large enough sample size, by the end of the season, there are a few people who have received a pamphlet that accurately predicted the winner for every game and they're convinced the system is perfect. The others weren't so lucky, however. Relying on candlestick charts and TA patterns that are relics from the pre-computerized era is reassuring to some traders and gives them a sense of control and "beating the market," but how long will chance remain on your side?

This is why I maintain that visual tools like candlestick charts are a slowly dying medium. They certainly still add value to some trading firms, which is why Bloomberg terminals still ship with this functionality built in, but as more and more research shows, automated algorithms and statistical indicators can provide more value. It's also important to think about whether the majority of the value added by visual tools like candlestick charts comes in the form of profit or a sense of security to traders who learned the field using them over the past few decades.

Finally, it's extremely important to realize that the actions of retail investors in the equities market cannot begin to represent the behaviors of the market as a whole. In the equities markets alone, trading firms and institutional investors dwarf retail investors, and the difference in scale is even more vastly pronounced in derivatives and currency markets. The fact that some retail investors use candlestick charts and the technical indicators they (hope) underlie them provides nothing but minor anecdotal evidence as to their effectiveness.

• @EugeneS I did refer to HFT in some cases, but keep in mind that algorithmic trading != HFT in all cases. Over longer periods, TA might be more effective, but a) I don't know enough about that issue specifically to comment one way or another, and b) trading on longer periods relies on seeing the pattern and acting on the pattern quickly enough; computers can do both far better than humans. – John Bensin Apr 16 '13 at 14:53
• @EugeneS Just because something is hard for you or I to automate doesn't mean it's hard for a large financial firm that has the funds to hire numerous mathematicians, software developers, financial engineers, etc. This is why I always stress that the resources you and I have available pale in comparison to those available to major firms, which places us at a major disadvantage. – John Bensin Apr 16 '13 at 15:35
• @JohnBensin, HFT can actually assist someone using TA in the medium to long term. If a particular pattern is seen by many traders and they all act on it in the same way, then the predicted outcome has a larger chance of coming true. You talk about the private investor/trader having to act quickly. In longer term trading you don't have to act as quick as with shorter term trading. If you are looking at a profit target of 20-30% would you not take the trade because it has already gone up 1-2%. Most technical traders would actually see this as confirmation to take the trade. – Victor Apr 16 '13 at 22:59
• @Victor I gave you a response based on my experience, since in my experience, candlestick charts and subjective visual TA is a dying medium. – John Bensin Apr 17 '13 at 1:03
• @Victor Capital speaks for itself. Considering that institutional investors are increasingly moving towards algorithmic trading (which does incorporate some aspects of TA/FA, obviously), the habits of individual investors aren't a major factor. Perhaps they are in smaller markets like Australian equity, but not in the major markets, e.g. derivatives, which dwarf equity markets by a sizable factor. – John Bensin Apr 17 '13 at 2:34

Candlesticks and TA are a relic of pre-computer trading, period. Market makers use sophisticated algorithms not for trading, but manipulations.

From my 15 years of experience, no technical indicator actually ever works. Those teaching technical indicators are either mostly brokers or broker promoted so called technical analysts. And what you really lose in disciplined trading over longer period is the taxes and brokerages. That is why you will see that teachers involved in this field are mostly technical analysts because they can never make money in real markets and believe that they did not adhere to rules or it was an exception case and they are not ready to accept facts. The graph given above for coin flip is really very interesting and proves that every trade you enter has 50% probability of win and lose. Now when you remove the brokerage and taxes from win side of your game, you will always lose.

That is why the Warren Buffets of the world are never technical analysts. In fact, they buy when all technical analysts fails. Holding a stock may give pain over longer period but still that is only way to really earn. Diversification is a good friend of all bulls. Another friend of bull is the fact that you can lose 100% but gain any much as 1000%. So if one can work in his limits and keep investing, he can surely make money.

So, if you have to invest 100 grand in 10 stocks, but 10 grand in each and then one of the stocks will multiply 10 times in long term to take out cost and others will give profit too... 1-2 stocks will fail totally, 2-3 will remain there where they were, 2-3 will double and 2-3 will multiply 3-4 times.

Investor can get approx 15% CAGR earning from stock markets... Cheers !!!

• Buffet did actually try technical analysis very early in his career, but gave it up rather quickly. The quote from him on it is "I realized that technical analysis didn't work when I turned the chart upside down and didn't get a different answer." – zeta-band Sep 7 '17 at 16:10
• If you lose 100% you cannot make any more money because you have lost everything. Much of your statements and logic do not make any sense. – Victor Dec 10 '18 at 6:41

Buffet's answer quoted above is just so significant. TA is a dynamic beast nowadays and it only works if you know (and are ahead of) the current algorithms being played by the large market manipulators, oops meant participants.

It's truly amazing how this gigantic scheme called the stock market is a gaming machine to drain money out of people, and how the government allows it to do so in the name of "free markets" that are supposed to reflect the value of things.

Keep TA very simple, with main pivot levels and main trendlines and ma's that apply to that instrument. For example, what worked in the recent past is traders taking out people's stops below support levels, then reverting the stock.