Bars and Candlestick formations

[Shakone]

Just want a few opinions.

Suppose you backtest various bar/candle formations, pin bars (even considering the previous bar), engulfing bars etc with a risk reward 1:1, and you find that the win probability is around 50-50 or slightly less than. Would you from that information then conclude that the bar formation offers no real information or edge and may as well just be a random entry? Or even more, conclude that you can't get a profitable system from using them? And if so, why?

 

[pboyles]

Just want a few opinions.

Suppose you backtest various bar/candle formations, pin bars (even considering the previous bar), engulfing bars etc with a risk reward 1:1, and you find that the win probability is around 50-50 or slightly less than. Would you from that information then conclude that the bar formation offers no real information or edge and may as well just be a random entry?

I did pretty much what you describe some time ago in tradestation and yes there wasn't really an edge unless you curve fitted the TP and SL level. I concluded that there's no way to profitably trade these patterns without some form of discretion/skill. Lets face it if it was possible someone would have made software to do it as its very simple to code.

 

[pboyles]

Just to be clear I dont think these patterns should be ignored, I think they do have value, just that you have to apply some skill to selecting which ones to take.

 

[the hare]

I wouldnt conclude they where random (although they are, but thats another story)
The biggest problem of course is the 1:1 constraint. Its completely arbitary. Every single academic study suffers from the same problem. They focus on entry, and chose a ridiculous and arbitary exit strategy.

I'll flip this around. If I was to enter on a pin bar, and show you a system that averaged 20% per month would you conclude that pin bars have an edge (or would you maybe consider that it might it just might be the exit, trade management and position sizing strategy thats responsible ?)

The component elements of systems tend to be synergistic (in a constructive or destructive way), measuring the effectiveness using such a blunt tool such as PnL is insanity actually.

 

[MajorDutch]

Just want a few opinions.

Suppose you backtest various bar/candle formations, pin bars (even considering the previous bar), engulfing bars etc with a risk reward 1:1, and you find that the win probability is around 50-50 or slightly less than. Would you from that information then conclude that the bar formation offers no real information or edge and may as well just be a random entry? Or even more, conclude that you can't get a profitable system from using them? And if so, why?

Hey Shakone

How are you coping post Wall Street 1928? exhausting wasnt it!

personally I use bars/formations as triggers only. Therefore if I am looking at a chart I am looking at areas that I am interested in first and not bars/formations. If price isn't in an area of interest (for example a previous price pivot zone) then I would not be looking to trade full stop whether a perfect bar/formation appeared or not.

I certainly wouldn't conclude that they are as good as a random entry but in isolation they may well be.

Re - random entries. I would like to understand more about this type of strategy and how some people deploy them. I don't think it is something I would ever trade however I am interested to know about this type of strategy just for completeness.

 

[Lord Flasheart]

50/50 is not a bad win ratio,it really depends how it is worked out and for how many pips. how often do we hear that a set up is 80% accurate. What does this mean,it has to be calculated to an exact pip level to be proper analysis. If a pattern was 50% to 10 pips then i would look at consecutive losing sequences. If it is 50% the the sequences would be different from say a roulette wheel because there is emotion involved.Then look at what happens after it reaches 10 pips(how far it goes) and then base a system on that. Jesus I'm getting carried away

 

[VielGeld]

I did one exercise in random entry a week or two ago. I flipped a coin and that decided whether I was short or long. There would be no stop, and profit is taken at 10 pips. All trades were winners, though one did have a drawdown of 70 pips and lasted 4 hours before being profitable. LOL!

Result is that entry matters, but it's really all about direction. I've been doing similar exercises with Robster in my trading journal thread, and I can say that direction is the single most important factor. Entry is just there to determine what kind of stop loss you want (tight or loose), and how much profit you can squeeze out of the trade. I believe it is entirely possible to get in at the bottom, and out at the top. It just takes skill.

The idea of entry is generally that it gives a clue to direction. Imo, your skill as a trader should be to determine that in advance just by watching price action. I have found candlestick formations to be pretty much useless (then again, I don't have 20 years of exp. watching those thing so I might be biased). The one true indicator that I've found reliable is support, resistance, and round numbers. These will always exist in the market. Thank God, otherwise it wouldn't be tradeable.

 

[MajorDutch]

I did one exercise in random entry a week or two ago. I flipped a coin and that decided whether I was short or long. There would be no stop, and profit is taken at 10 pips. All trades were winners, though one did have a drawdown of 70 pips and lasted 4 hours before being profitable. LOL!

Result is that entry matters, but it's really all about direction. I've been doing similar exercises with Robster in my trading journal thread, and I can say that direction is the single most important factor. Entry is just there to determine what kind of stop loss you want (tight or loose), and how much profit you can squeeze out of the trade. I believe it is entirely possible to get in at the bottom, and out at the top. It just takes skill.

The idea of entry is generally that it gives a clue to direction. Imo, your skill as a trader should be to determine that in advance just by watching price action. I have found candlestick formations to be pretty much useless (then again, I don't have 20 years of exp. watching those thing so I might be biased). The one true indicator that I've found reliable is support, resistance, and round numbers. These will always exist in the market. Thank God, otherwise it wouldn't be tradeable.

can you expand on what you mean by 'it's really all about direction'. If I use a 200ma to give me the direction is this all you need?

 

[VielGeld]

No, not really. I just meant direction in the time frame you're using.

Let's say the market is selling off. It hit "$100", and it's now dropping like a rock. You have (imo) two things you can do here:

1) Go with it.

2) Wait for buyers to come in, and catch the bounce.

Which do you choose?

Well, the way I see it is that you'll look for support. If major support is determined to be "$50", and the market is at "$75", then you have potential for $25 profit. What about the bounce? This is more nebulous. I wouldn't try it in stocks, but I think it's fine in currencies given the momentum shifts or else it's just going to plough right through you like when EUR/USD went from 1.38 to 1.35 a week or two ago (when you'd think every 100 pips is major support).

It's like Netflix when the bad news hit. It was at 300 at its top, and dropped down to 100 before catching its breath. You could have gone short anywhere in between and made money, there wasn't any specific entry to look for in there.

Of course, I go with the 30s, 5m, 15m, and 1h time frames to determine this stuff, and I don't have any exp at all. But it seems to work, so I do it.

 

[MajorDutch]

No, not really. I just meant direction in the time frame you're using.

Let's say the market is selling off. It hit "$100", and it's now dropping like a rock. You have (imo) two things you can do here:

1) Go with it.

2) Wait for buyers to come in, and catch the bounce.

Which do you choose?

Well, the way I see it is that you'll look for support. If major support is determined to be "$50", and the market is at "$75", then you have potential for $25 profit. What about the bounce? This is more nebulous. I wouldn't try it in stocks, but I think it's fine in currencies given the momentum shifts or else it's just going to plough right through you like when EUR/USD went from 1.38 to 1.35 a week or two ago (when you'd think every 100 pips is major support).

It's like Netflix when the bad news hit. It was at 300 at its top, and dropped down to 100 before catching its breath. You could have gone short anywhere in between and made money, there wasn't any specific entry to look for in there.

Of course, I go with the 30s, 5m, 15m, and 1h time frames to determine this stuff, and I don't have any exp at all. But it seems to work, so I do it.

what you are describing is what I call 'gap fill' where a major S/R is broken and price looks to fill the gap to the next decision point at next major S/R.

 

[wallstreetwarrior87]

I did pretty much what you describe some time ago in tradestation and yes there wasn't really an edge unless you curve fitted the TP and SL level. I concluded that there's no way to profitably trade these patterns without some form of discretion/skill. Lets face it if it was possible someone would have made software to do it as its very simple to code.

:clap::clap::clap:

There is no way to ever reach the highest level in trading that you possibly can without discretion (after taking luck out of the equation).

Then again, whats average, and whats luck , and how do we determine this?

Discretion is the only thing that I have (as do you, as does your next door neighbor) that is different from all other participants in the market. Im thinking in terms of high IQ employeses at banks and insitutions here, whom have all the technology at their disposal. If parameters can be used then you can bet they have tried everything possible from a systematic point of view. Basically if a retail trader tries out some parameters in tradstation etc, then you can bet that the most able of minds have done so as well.

Hence any parameters that create an edge will fade over time (the edge that is).
This is because there is no grey area with systematic trading, rules are black and white. Thus easy to copy and replicate, then the more that use these parameters, the less effective they will be.

One day someone is going to create a thread that will discuss the possibility that discretion is key that opens the golden fortress! The difference between good results and great results ultimately will come down to discretion.

Of course discretion is the final cherry on the cake, all other ingredients must be in place before discretion is of any use or value.

If not, a hammer will always be just a hammer, and a doji just a doji and so on, no matter what parameters are used

 

[VielGeld]

what you are describing is what I call 'gap fill' where a major S/R is broken and price looks to fill the gap to the next decision point at next major S/R.

Sure, on that scale it's a "gap fill", I guess? I look to smaller ranges like 10-20, maybe 50 pips, though.

Trading is all about getting in at A, and going to B. To do this, I just find points where buyers will want to buy, and sellers will want to sell, and with what kind of strength.

And the place where this is most apparent is around S/R. If buyers are getting weak close to xx00, then chances are you can safely bet on the reversal. The reverse is true too. But if either is strong at that point (there are many attempts to break through xx00), then better bet on S/R being broken.

So which S/R is significant depends on time frame. Scalpers will look for maybe 5-10 pips max. Position traders want to build a case for 1000 pip moves which needs fundamentals. It's all really relative.