Strongest Candlestick Formations

Hi,

Would like a discussion your top 3 candlestick formation pattern reversals at S&R zones in your view here are mine:-

1. Hammers/Pin Bar Reversals
2. Railway Tracks
3. Outside Bars

Also, has any study been done on percentage on how strong these are in a trending market? I heard 70-80% accurate is that true through experience?

Candles is an old and unprofitable method. It is like you sculpturing text messages on rocks and sending them via UPS when you can just use your mobile phone to do the same.

For me, the strongest formations are morning/evening stars and bearish/bullish engulfing patterns. I use them quite a lot and the success rate is around 80% plus believe it or not. When used in trading double tops/bottoms or head and shoulders I find they work the best.

I found a copy of Steve Nison's candlestick book over the weekend at the bottom of a box of junk, I must admit, its an entertaining read, I havnt laughed quite so much in ages.

Do you not like that book? I read it a while ago when I first started trading and I must admit - It taught me the basics I needed to know.

dlukic said:

Do you not like that book? I read it a while ago when I first started trading and I must admit - It taught me the basics I needed to know.

Click to expand...

If I'm being charitable, I think that Mr Nison is being fooled by randomness. Its quite interesting that even in his hand picked examples for the book, there are charts that show numerous failed examples of patterns covered elsewhere in the book.

Not surprisingly, there's quite a bit of academic research undertaken into candle patterns (its easy to do, as the patterns easily measured and identified). Nison has been openly critisised by a number of authors, and has to date been totally unable to refute that critisim other than to reply with platitudes regarding "applying the patterns in context", or "applying with discretion". Most of these academic studies completely ignore the bleedin obvious, and conclude that just because a particular entry signal isnt profitable, it doesnt work (ignoring the other 99% of factors that differentiate profitable from non profitable trades), however, they are right to point out that simple patterns are statistically non valid.

Of course particular patterns will occur at turning points, its a moments work to scan back through a chart and find a doji, or engulfing patern at a turning point, but that in itself is pretty meaningless. Its equally easy to find examples of patterns where long shadows identify area'a that later act as support or resistance. The problem is, exactly the same features apply on randomly generated time series, with equal statistical significance.

I'm not saying that candle patterns are not a reasonable entry trigger (and I have used them, and maybe even will again in the future), but the reality is, over a large enough sample size, they are no better than a random entry.

You see people combining and blending candles, you see people applying discretion, but thats no different to retrospectively modifying the parameters of an indicator to get into a trade, or to filter out a losing trade.

I'd probably argue that any book that focus's on entry signals isnt ever going to be a particularly great trading book. I accept that the book is a little more than that, but really, some of the nonsense he comes up with is beyond belief.

There's practically a whole chapter arguing why a candles which show open, high, low and close, are SUPERIOR to bars (which show precisely the same information!)

Maybe I'm being a bit unfair on Nison, but the least I'd expect out of any credible author is to back up his or her claims with hard statistics, rather than a few hand picked examples.

I tend to agree that candlestick patterns alone, comprising the usual suspects of 1, 2 or 3-day patterns of various sorts actually have very poor % win rates. Probably no better than a MA crossover system or looking for a 3-day pullback to enter a trend, and similar simple stuff.

But the thing is, any and all of these can be made profitable by the right money and position management, and sensible placing of stops and limits. Such patterns are just the chassis of the car - 4 wheels, on at each corner. After you have that, the really tricky (but rewarding) decisions come form the type of engine and body you select.

To be honest - I never really looked at it like that but you have some very good points. When I first started I used the book to mainly recognise specific candle formations and once I started trading and was taught properly I learnt about entry points etc. I never really looked back and said - I never used that book to teach me how to trade. In actual fact, that is true. I never use it as a reference as I see it as basic documentation of candle formations. In short, I think you have taught me something this morning that I subconsciously knew already but haven't thought about for years as I was distracted by real trading strategies taught by real professionals.

Albert Carey said:

Candles is an old and unprofitable method.

Click to expand...

That's quite a profound statement, and quite far reaching. I'd love to see the statistical basis on which this announcement is founded. It's a bit like saying "graphs are unprofitable" or "fundamental analysis is unprofitable".

You might not have mastered candlestick analysis yourself, but I think it's a bit of a stretch to write them off as generally unprofitable. They are like any form of technical analysis, they have their strengths and their limitations. They are particularly useful for showing potential changes in a trend, for instance dojis or engulfing bars. Long candles with small wicks either end show strong market momentum, i.e. a warning not to try to fight it.

I find them very useful anyway.

Good trading.

I'm afraid I would have to agree - Candlesticks provide me with a real-time visualisation on market conditions and behaviour. This kind of statement really does need to be backed up with factual evidence. I'm open to anything as long there is proof.

Albert Carey said:

Candles is an old and unprofitable method. It is like you sculpturing text messages on rocks and sending them via UPS when you can just use your mobile phone to do the same.

Click to expand...

I think the above statement would be better framed in the context of using candlesticks as a sole and stand-alone trading approach, in that context there is a degree of truth. I believe any indicator/tool has merit, but the merit is extremely limited if not utterly useless if not combined with other analysis techniques. I would call it a layering approach - maybe you combine the candlestick analysis with how other asset classes will impact the instrument you are looking to trade etc etc

I just do not believe that a one-dimensional trading approach can be successful over a sustained period.

A question for the daily candlestick traders. Suppose you have a swing low and a pin bar forms on daily. The James16 people will tell you to enter long a few pips above the pin, stop a few pips below the pin. What do you think about the fact that your entry is worse than the WORST buying price from yesterday, and if it goes down to your stop, you'll be selling at the worst price for selling from yesterday?

Of course I realise that is yesterday. But it does seem less than ideal to me.

I think the only way one can do reasonably ok with candles is to make sure:

1) there is no strong trend present.
2) wait for confluence.

However I tend to agree on Steve Nison. Although I have learnt the above from him (by buying his DVD some 4- 5 years ago) these days I am repeatedly receiving sales e-mails from him. Its kind of worrying; if his methods are so profitable why doesnt he just get his head down, trade and make money as opposed to going on overdrive trying to sell his product!

where people slip up is they get hung up on the patterns and of course the candle entry is just one factor in a set up. far more important is what has happened to price in the past and whether the price is meaningful to other participants.

It is more helpful to think of say pin bars and engulfing patterns in terms of price travel during that day (in respect of the daily pin) so for a bearish engulfing bar price tests higher then rejects, keeps rejecting and closes at a lower price than the lowest price on the previous day. This is a pretty good indication that price didn't like that new high level and moreover price didn't close in yesterday range which could signal price going range bound (which it does most of the time). With a pin bar we are seeing a price reject and a close within the range of previous days trading. These bars are just price representations nothing more. It's when people start going all voodoo it really gets silly, imo Nison goes a bit too far.

Far more important are price pivot zones, round numbers imo. just my 2c

DDI said:

I think the only way one can do reasonably ok with candles is to make sure:

1) there is no strong trend present.
2) wait for confluence.

However I tend to agree on Steve Nison. Although I have learnt the above from him (by buying his DVD some 4- 5 years ago) these days I am repeatedly receiving sales e-mails from him. Its kind of worrying; if his methods are so profitable why doesnt he just get his head down, trade and make money as opposed to going on overdrive trying to sell his product!

Click to expand...

just shorted crude 9640 stop 9700.

will post chart to show why as candlesticks played a role.

lets see how it developes

Ok.

1) first There was a bearish engulfing (BE1) and follow through thus thats resistance (which coincided with S1 daily pivot)

2) Later we see a failure (but no candle pattern)

3) Price rallies back to resistance and a shooting star (SS)

4) and now a bearish engulfing (BE2) and importantly for me the ADX/ADMi shows there is no strong upward trade

5) So i entered short given such confluence.

Lets see how it goes. will probably come out around 9580

Am still in (with a bit of cheating) - I got stopped at 9700 but immediately shorted at 9695 when it failed to bread,

Anyway now we have a Dark cloud cover (DC) and another bearish doji or maybe even a shooting star (SS@) - so 4 consecutive bearish patterns

am out at 9640 for beakeven didnt want to lose any money on the trade...was up about 40 pips at one stage

Also another factor in current exit was given the time of the day and tomorrow closed,

Shakone said:

A question for the daily candlestick traders. Suppose you have a swing low and a pin bar forms on daily. The James16 people will tell you to enter long a few pips above the pin, stop a few pips below the pin. What do you think about the fact that your entry is worse than the WORST buying price from yesterday, and if it goes down to your stop, you'll be selling at the worst price for selling from yesterday?

Of course I realise that is yesterday. But it does seem less than ideal to me.

Click to expand...

Yes what you say is true, and I think most people find this kind of pin bar pattern has a pretty low win rate. It is relevant that the entry is higher than the previous intraday high and ther stop is lower than the previous intraday low because - IF - the pattern is a valid signal, these are significant s/r levels and your position management should respect them. There are two problems - the distance from entry to stop is a full day's range plus several points, making position size control critical, but even worse, the pattern is just not highly reliable.

But most swing patterns are like this, more a matter of position and exit planning than clever TA.

Hi DDI, Many thanks for the detailed posts with charts and reasoning. I had one doubt, What made you choose 9700 as the stop point?
thanks!

DDI said:

Ok.

1) first There was a bearish engulfing (BE1) and follow through thus thats resistance (which coincided with S1 daily pivot)

2) Later we see a failure (but no candle pattern)

3) Price rallies back to resistance and a shooting star (SS)

4) and now a bearish engulfing (BE2) and importantly for me the ADX/ADMi shows there is no strong upward trade

5) So i entered short given such confluence.

Lets see how it goes. will probably come out around 9580

Click to expand...