Do candles work?

Don't like your set-up... no stop?

Yes, I agree that I should have a stop if I were to build a real trading system. But that is not the goal. The goal of the experiment is to understand if the pattern has any predictive power. So my exit is based on timed on the close of the next day.

The results indicate that the win-ratio of around 52%. That means that this is not a reliable indicator of trend reversal
 
Yes, I agree that I should have a stop if I were to build a real trading system. But that is not the goal. The goal of the experiment is to understand if the pattern has any predictive power. So my exit is based on timed on the close of the next day.

The results indicate that the win-ratio of around 52%. That means that this is not a reliable indicator of trend reversal

Fair point. I'm not defending dojis. Slightly different definition here:
http://en.wikipedia.org/wiki/Doji

I wouldn't use your investigation of one candlestick shape to rule out the whole notion of candlesticks. I would be interested if you could investigate the daily pinbar at key levels.
 
Fair point. I'm not defending dojis. Slightly different definition here:
http://en.wikipedia.org/wiki/Doji

I wouldn't use your investigation of one candlestick shape to rule out the whole notion of candlesticks. I would be interested if you could investigate the daily pinbar at key levels.

I plan to test out a few more patterns over the next week. Candlestickgenius.com has a list of most reliable candles. I will post the results

I am not very familiar with the pin-bar. From what I have seen on the web, it is a doji with a open and close very close to each other (also known as the spinning-top doji/graveyard doji in candlestick parlance). I think I can look into it. Keep an eye for it
 
...The goal of the experiment is to understand if the pattern has any predictive power. So my exit is based on timed on the close of the next day.

The results indicate that the win-ratio of around 52%. That means that this is not a reliable indicator of trend reversal

Nothing has predictive power in trading, not even yourself. The only thing you can try to do is increase the probabilty of price going in your direction. A win ratio of around 52% is more than high enough to grow your account, given the right RR.
 
I plan to test out a few more patterns over the next week. Candlestickgenius.com has a list of most reliable candles. I will post the results

I am not very familiar with the pin-bar. From what I have seen on the web, it is a doji with a open and close very close to each other (also known as the spinning-top doji/graveyard doji in candlestick parlance). I think I can look into it. Keep an eye for it


I use PIN bars but with some discretion, in other words I dont take them all. No matter what criteria I came up with I never found a 'system' that could have been automated. If you look at collectiveC2 you'll see that hundreds of people have tried to create such systems but none of them will work over time. I think there is value in candlestick analysis but not on its own.
 
The results indicate that the win-ratio of around 52%. That means that this is not a reliable indicator of trend reversal
Hi kilotrader,
Welcome to T2W.
As RBurgess has already pointed out, a 52% win rate is more than sufficient to have an extremely profitable trading strategy, assuming all the other elements are in place. As is often mentioned here on T2W, some of the most successful traders ever have success ratios of less than 40%. If you don't understand how this could be, please read the Essentials Of First Steps Sticky.

On a general note, applying a mechanical based back test to candlestick patterns is never going to produce a high percentage result. To be honest, I'm surprised at how good the result is for the test you've done. Windowsill's comment about ignoring the candlesticks in the middle of the chart that don't work is both flippant and witty. It's also very true. Rather than doing endless tests to find the magic pattern that produces the success rate you want, I suggest your time would be far better spent looking for ways to eliminate the failed signals and to assess the market conditions that helped to ensure that the winning ones worked. That's the best way - and possibly the only way - to push up the 52% hit rate to something you'll find more acceptable.
Cheers,
Tim.
 
Kilotrader,
Testing of candlestick patterns has been conducted previously by a gentleman named Thomas Bulkowski. He tested a plethora of these patterns and has a book and website which help show which patterns actually work and at what percent of the time. That website is

http://thepatternsite.com

He does not charge anything to review his results on the web and also has a free download of software which can recognize some of the candlestick patterns. I have not tested this software so cannot comment on its effectiveness. Here is an excerpt from his site on the Bearish Doji Star,

"The bearish doji star candlestick is supposed to act as a bearish reversal pattern but doesn't. In fact, it is a wonderful continuation candle because that is what happens to price -- it continues rising 69% of the time, ranking 8th, where 1 is best. Looking at the candle pattern as it moves from a tall white candlestick, to a doji with a gap between, you would swear that momentum was slowing. It does, but the numbers suggests it picks up again and price moves higher."

Since I am only daytrading, I do not utilise this site and candlestick patterns for my trading, but it is an interesting read to see what has statisitically proven as effective pattern predictors as opposed to the urban legends that some of these patterns hold based on his analysis.

I do want to state unequivocally that I am not in any way affiliated with this site or Thomas Bulkowski. I am not even a self proclaimed expert like some. Good luck with your trading and hopefully this site will help with your analysis and trading.
 
Hi KT, welcome to T2W :).

In answer to the question, no they don't. Nothing "works" in the sense that if you blindly follow every signal generated you will make money and enjoy a high strike rate. The other problem is how you define "works". Depending on your trade management, a set up could work beautifully or fail miserably.

Candles are a tool just like anything else, and just like all the other tools we use, they don't actually exist. You call them into being by choosing to display price in a certain way, but If you choose to display price in a different way, your candle will disappear. Price is the only real thing on your chart.

Take the beloved daily pin bar for example. Change to an hourly timeframe, and your pin vanishes to be replaced by a series of candles moving in one direction then reversing to finish near where the day started. You do not even have to change timeframes. Say your broker is IG - your daily pin will be the record of price starting at midnight and finishing at midnight. However, if you use a New York-based broker, you might well find that instead of seeing a candle that ends at midnight GMT, you will have a candle that closes at 5 pm Eastern (although given how the markets work, in this particular case the look will tend to be very similar).

Candles are just a visual representation of how price behaved over a given period, but this can of course be useful. Going back to the pin bar, assume a bullish pin - that is to say, open and close near the top of the bar, with a long tail or lower shadow. This tells us that price opened high, moved down strongly, and then reversed sharply. It's important to consider orderflow and so on, which is a bit much to go into here, but obviously you can see that there was strong selling pressure which was subsequently overwhelmed by buying pressure. This could suggest a reversal. It could also suggest a good, logical place to put your stop - after all, if the bottom of the pin is broken, you could reasonably say that the latest support level has been invalidated.

Now this might or might not be enough to trade on. Remember though, the market is not saying "There's a bullish pin - time to buy!". So what you could do is look at supporting factors.

For example, does the pin occur at an area of significance, such as support / resistance / pivot zone? If so, this could be a good factor, as (using our example above) support indicates an area where buying overwhelms selling. Applying this in a practical sense, we can perhaps say the following:

1. Price is approaching an area where we know buying has been strong in the past.
2. Price has tested this area, and we have a visual confirmation that the area has held (the pin).

So in this case it could be the area that has the real significance, and the pin bar is our confirmation or "trigger" for entering the trade.

Maybe you want to adopt a more conservative approach. So you look for further supporting factors. For example, you might want to trade in the direction of the prevailing trend. So you take the pin, but only if it is a pullback in an overall uptrend.

Or you might decide, following your research, that round numbers have significance. So if the pin occurs at a round number, testing the area below it, you could view this as additional confirmation.

You could refine things still further - for example, you could require that the pin closes above the round number. You could require that the pin is large relative to recent bars, reasoning that a small pin might not actually be giving you much information.

So now you might have a good looking set up based on a significant area and a good trigger candle. Nexy you might decide to look around and see if you can see reasons why the trade might "fail".

Assume the candle closes below the round number instead of above, and you believe that the round number is an area of significance. Surely you want to be trading away from it and not into it? So no trade - unless possibly you just want a quick trade up to the round number, taking full profit when it gets there.

Or assume that there is a resistance area close above your pin. Would you take the trade knowing that you are going to potentially run into trouble almost immediately?

You can also chuck in the endless variety of indicators, fibs, MAs and so on if you want. I don't use anything of that nature so can't comment on them, but anyway hopefully you get the idea.

So now you've got your set up and entry. Congratulations! You've got maybe 3% - 5% (if that) of what it takes to trade successfully.

How will you manage your trade? What do you do when price hits the first touble area? Do you exit, move to breakeven, let it ride? Does your decision depend upon the strength of your set up, the strength of the resistance area, what you had for breakfast?

If you work that out, maybe you've got another 5%.

Now you need to work on money management, discipline, patience and the whole psychological side of trading. That's the missing 90%.

You're approaching things in a good way - learning, investigating and doing your own testing. That way you can have some idea of expectancy before you put your capital at risk. But a successful set up is likely to be a combination of factors.

So, do candles work? It depends on what you mean by "work". A daily pin bar, with a little QC, has about a 90% chance of getting to the first trouble area as I define it. Someone else might hit only 50% of the time, but secure a much bigger profit when they do. It all depends on your approach.

But for most people most of the time the answer is no, they don't work - on their own. Think about a car - you want it to get you from A to B. If it does, the car "works".

So you take just the wheels and off you go. Except you don't, you get cross, and you say "Wheels are f****** useless". So you stop using wheels and get an engine instead.

Again you go nowhere. "F*** engines too!"

Now you try rear-view mirrors and a cup holder. Eventually you decide that the whole concept of cars is useless and you abandon the idea of getting from A to B.

You need to assemble the components first of all. If you still get nowhere, then maybe you need to learn how to drive.

Even then, you have to use your tools in the appropriate way. Is a Rolls crap because it's useless off-road? Is a Defender crap because it's not very comfortable on the motorway? No, they're among the best, but a bad driver will still total them or get them stuck in a field.

I use candles all the time - I'm actually in a trade as I write this, based on a daily pin (AUD/USD, about halfway to my target). But you have to remember that they're just something to consider when you're deciding whether or not to trade.

Don't forget - nothing on your chart is real apart from price. Candles, fibs, MACD, the trend, support and resistance - these are all fictions, or at best highly subjective. S / R and pivot zones are the basis of how I trade, but they only exist (for the most part anyway) because I believe that they do, and trade accordingly. I could believe in a support area, trade and make money. You could equally believe it doesn't exist, trade and make even more money, because your completely different interpretation could be as good as or better than mine.

Hope some of this ramble is helpful, if you want to ask anything or shoot down what I've written, please feel free :).
 
Last edited:
Hi there - I am more of a trader on fundementals myself - I am not one of these guys who place no value on technicals though! - My favourite candle patterns are -Bearish/Engulfing when at high price area - Bullish/Engulfing when at low price area - Shooting Stars when at new highs - Hammers when at new lows - As a minimum I like the graphic display of the battles between Bulls/Bears and it is more than 52% reliable as an indicator of a change in trend. - good book to get is by Steve Nison.
 
Candlesticks patterns works like anything else, if you use right patterns you'll get higher probabilty for win trades.
 
If a strategy simply involved finding a doji after a strong up/down trend to indicate a possible reversal, then this highly "mechanical" and simple system would be taken advantage off.

The 52% statistic you state is exactly as I expected it to be. There is no edge in simply finding a doji and taking a trade on it. You need an edge in the markets to be successful. However, an edge can be developed by combining doji patterns with other techniques.

Amit

I am new to the world of candlesticks and recently started to read about them. I completed a back-test on Doji as a trend-reversal signal. The results are located at:

http://blog.kilotrader.com/2010/05/do-candlestick-patterns-work.html

And they are NOT impressive

What did I do wrong here? Do candlestick signals work?
 
Every pattern must be combined with TA, candlestick patterns are just one of the hundreds indicators.

This is my favorite indicator for detecting standard patterns:
http://www.forexmetatraderindicator...cognize-candle-sticks-patterns-mt4-indicator/

It could recognize over 30 patterns, but I use this indicator only for 3 patterns. You can find several standard candlestick pattern indicators online if you use manual strategy like me. If you searching for advanced pattern I suggest you UCRP it's not a miracle, but this is the most accurate pattern ever. It really works, if someone have better one please post a link, it's really hard to find patterns like UCRP, pros are hiding their patterns.
 
[
What did I do wrong here? Do candlestick signals work?[/QUOTE]

Consider this: every conversation needs a context otherwise you will find the subject difficult to follow. Candlesticks need context. If a market is trending and has been trending for a while, such as Spot Gold, seeng a doji suddenly appear would be more like a signal for a pause rather than a signal for a change in direction.

Finding other technical aspects to confiem the candlestick is very important. For this reason alone researching just candlesticks and reading books on candlesticks will not help you become a better trader. It will only take up your time. You are better to check back on one particular candlestick type/pattern at certain price levels where other technical aspects can also be found. Tweezers for example, this pattern is always a good signal to make money on any time frame!!

Check out

www.chart-workshop.de
 
I am new to the world of candlesticks and recently started to read about them. I completed a back-test on Doji as a trend-reversal signal. The results are located at:

http://blog.kilotrader.com/2010/05/do-candlestick-patterns-work.html

And they are NOT impressive

What did I do wrong here? Do candlestick signals work?

Where you enter/exit a trade and how you manage a trade (moving stop to BE for example) also used to (hopefully) increase win rates/expectation, looks like you just entered the next day's open and exited the following day's close no matter what which I think is too rigid.

I keep meaning to do something like this with pinbars...
 
Top