Higher Edge within a Single Candlestick | Page 90 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

An Edge for an Edge (System Development) 116 replies

Beating the Odds - Edge Within a Single Candlestick (Explained) 71 replies

Looking for higher timeframe candlestick overlay indicator 10 replies

top 2 single candlestick pattern 7 replies

There is No Edge in a Single Candlestick 64 replies

Attachments: Higher Edge within a Single Candlestick
Exit Attachments

Higher Edge within a Single Candlestick

  • #1,781
  • Nov 12, 2017 11:34pm Nov 12, 2017 11:34pm
Quoting despacito7
I am agree that for long run it is difficult to get the same good result without any good statistical proved data (which ever any idea/system we use). If i understand it correctly, this 9/10 result above can be applied to Forex world too, right ?? If yes, which TF will give the best result (i just predict that it is below H4, but which one is the best among H1 to M1 )
As the textbooks say, 8 wins out of 10 is luck, but 800 wins out of 1000 is evidence of an edge. It's all about the law of large numbers delivering statistical validity.


I see both similarities and differences between roulette and trading:

 In both games there are underlying mechanics (fundamentals) that generate outcomes (technical analysis, if you like). In roulette, the underlying mechanics involve the physics provided by the croupier, wheel, ball, etc, and the outcome is the number spun. In trading, I have outlined some of the underlying mechanics here, and price is the outcome. All TA-based studies are ultimately derived from price, time and (occasionally) volume.
i.e. fundamental drivers --> price --> TA

 Both are negative expectancy games by default. In roulette the payout amounts always favour the house by 37 to 36, while with trading the trader must contend with broker costs, that likewise affect every transaction.

 In roulette, unless the physics provide an exploitable bias, the game is completely random, and no playing or betting system can reduce the house edge. Historical outcomes have no statistical significance, and hence analysing past trends of numbers is futile. Whereas with trading, the price movement is not a random walk, making it possible to gain an edge by studying past data using statistical or technical analysis. This is due to a number of repetitive factors like prevalent belief systems that create 'self fulfilling prophecies', the psychological behaviour of traders, and -- to whatever extent that it is possible -- manipulation by heavyweight players. I listed some symptoms of non-randomness in myth #22 here.

 In both roulette and trading, an analysis of the underlying mechanics (fundamentals) gets the analyst one step nearer the 'real action'. Some institutional traders have expert knowledge and access to 'inside' information and techniques which give them a significant advantage over the average retail trader.


To answer your final question, in my opinion timeframe is largely irrelevant in trading. Although markets are not totally fractal in nature (see myth #13 in the above link), trader psychology/behaviour is such that price tends to follow similar accumulation --> breakout --> markup/down patterns in all time frames. Personally I prefer the longer timeframes for at least three reasons (alignment with macroeconomic factors; lower involvement requirement; relatively lower costs; ...), but there are successful intraday traders also. Intraday trading facilitates the big advantage of higher frequency, allowing gains to be compounded more rapidly.
  • #1,782
  • Nov 13, 2017 5:19am Nov 13, 2017 5:19am
Quoting hanover
{quote}As the textbooks say, 8 wins out of 10 is luck, but 800 wins out of 1000 is evidence of an edge. It's all about the law of large numbers delivering statistical validity. I see both similarities and differences between roulette and trading:  In both games there are underlying mechanics (fundamentals) that generate outcomes (technical analysis, if you like). In roulette, the underlying mechanics...
When it comes to the market, this statement is only partially true!
I gave my argument here. But let me state it again.

Unlike the roulette, the market provides environment which allows the luck factor to persist for a very long time!
You can have perfect track record even after 1000++ trades and then out of nowhere you can crash.
And the worst part is - there is no way to know if your results are caused by persistent luck or not! According to probability theory all profitable traders may be just lucky winners. (I will let the math wizards in the forum to calculate the chances for this).

Here is another simulation with 100 equity curves (1000 trades each), average win rate is 90%. Expectation per trade is below zero. (meaning that ALL curves will drop to zero if the sample is large enough. For example 100 000 trades.

Attached Image (click to enlarge)
Click to Enlarge Name: LuckFactor.png Size: 162 KB


But we can clearly see that a few of the curves are going pretty well. (5% winning traders??? )
So, what's the conclusion?
The conclusion is that not only we need a sample with large number of trades, but we also need a large number of samples!

This again confirms that it's impossible to eliminate risk and uncertainty from the game of trading.
Even the best trader in the world could lose everything at some point. (Because the persistent luck works both ways!!!).
  • #1,783
  • Edited 8:27am Nov 13, 2017 7:47am | Edited 8:27am
Quoting alphaomega
{quote} When it comes to the market, this statement is only partially true! I gave my argument here. But let me state it again. Unlike the roulette, the market provides environment which allows the luck factor to persist for a very long time! You can have perfect track record even after 1000++ trades and then out of nowhere you can crash. And the worst part is - there is no way to know if your results are caused by persistent luck or not! According to probability theory...
Underlying your example of a random distribution is an assumption of small negative expectancy per trade with frictional costs of trading....which under a gaussian distribution via the Central Limit Theorem will tend towards a normal distribution and always go to zero......but given the markets penchant for fat tails, we may just find that a better descriptor of the market is defined by a variety of stable distribution. If that is the case then you may just find that it is the abnormal sigma moves that fundamentally are responsible for an edge over mere luck inter dispersed with periods of random price behaviour.

For example complex systems such as the financial markets may have short sharp periods of significant disruption driven by economic disparities or booms and busts from crowd behaviour and like other complex physical systems exhibit periods of 'punctuated equilibrium' inter-dispersed with periods of random price action and mean reverting behaviour. That at least is probably a more apt description of a financial market than one which can be defined by an idealized mathematical distribution.

While random chance and the associated wide fluctuation in possible return distributions can mean that even a trader with a significant edge can be unlucky and come undone, it is more likely that a trader with a definitive edge will experience a positive return distribution than a negative one.....but yes....nothing is certain in this game....hence the caveats applied. An edge is never a certainty, just a probability..... provided we are dealing with a sample of outcomes and not the entire population of outcomes.
  • #1,784
  • Nov 13, 2017 9:29am Nov 13, 2017 9:29am
Quoting alphaomega
{quote} So, what's the conclusion? The conclusion is that not only we need a sample with large number of trades, but we also need a large number of samples! This again confirms that it's impossible to eliminate risk and uncertainty from the game of trading. Even the best trader in the world could lose everything at some point. (Because the persistent luck works both ways!!!).
Hi Alpha,
As you said, it is impossible to eliminate risk and uncertainty from the game of trading ... can we say that in every position of our order, we must always use a stop loss (or at least a hedging position) ?? From your image i see that the risk is 5% , i assume that this is the stop loss of every trade (we risk 5% of our equity for that trade, defined by stop loss), using a 90% win probability system with R:R (risk:reward) of around 10:1 (please correct me if i am wrong on this). Could you please give a clue on this below :
Lets say we have a setup with 80-90% win probability, and we put 2% risk on each trade. With R:R (risk:reward) of 2:3 , i think it will be a good enough to go. But how about using bad R:R (lets say 3:1 instead of 2:3) setup with very small TP target ? As i am not a good statistician, i am not sure how dangerous it will be if my take profit is only around 1-4 pips (so my stop loss is around 3-12 pips) ....
If i have a setup with TP for 20 pips, and SL of 60 pips (having same R:R of 3:1) , i think it will be safer than a setup with small TP (1-4 pips) and small SL (3-12 pips) ... i am afraid that price will hit my SL first before touching my TP on this small SL ...
I don't think that we can trade without SL and hedging, but just close our order after X hours or X bars, no matter what happen, even if we are in a (big) loss position, although we have a setup with >95% probability for that trade ... What must we do in this case ?
Gracias .
I am failed,but i found 1,000 ways that won't work ...
  • #1,785
  • Nov 13, 2017 11:27am Nov 13, 2017 11:27am
Quoting VEEFX
{quote} Allow me Despacito as you are my wife's favorite song... she along with our daughter just love to piss me off by putting this song on infinite loop when we go on long road trips. They both have a bug for spanish music... oh the Havana song just got added to their spotify list. anyways. here's my 2 cents on what I do SL and TP should be part of strategy development and have no place in building/seeking an edge. Edge lies at the signal level along with all the relevant stats. Once an edge is found (good luck finding one in this life time......
VEEFX,

"SL is NOT at all risk management as SLs are not guaranteed by most FX retail brokers."

What according to you is an ideal risk management? Could you please delve more into this...

Thanks
  • #1,786
  • Nov 13, 2017 11:38am Nov 13, 2017 11:38am
Quoting VEEFX
{quote} Sorry man. Unfortunately, I don't save my deleted posts.... I had written a 2-page long post in the "5yrs and no strategy" thread on how I view RM. Any strategy that incorporates Risk Management (via SL) at the individual trade level is a failed strategy. It won't last for more than 3 yrs... even if you are starring at your trade station every second. EDIT: and moving SL to BE type position mgmt does not make it a risk-free trade lol
Thanks VEEFX...
  • #1,787
  • Nov 13, 2017 12:36pm Nov 13, 2017 12:36pm
Quoting VEEFX
Any strategy that incorporates Risk Management (via SL) at the individual trade level is a failed strategy. It won't last for more than 3 yrs...
With respect, that's a sweeping generalization that would be difficult to prove. And doesn't CrucialPoint use SL-based RM with his scalping?

Agree that SL is not guaranteed to be filled, if there's insufficient liquidity. But that's true for any order type.
  • #1,788
  • Nov 13, 2017 1:23pm Nov 13, 2017 1:23pm
Quoting hanover
{quote}As the textbooks say, 8 wins out of 10 is luck, but 800 wins out of 1000 is evidence of an edge. It's all about the law of large numbers delivering statistical validity. .
and what about 126 wins out of 126 trades,
Attached Image (click to enlarge)
Click to Enlarge Name: 66666.png Size: 27 KB
don't predict direction, but dance with flow of water
  • #1,789
  • Nov 13, 2017 1:27pm Nov 13, 2017 1:27pm
Quoting priceaction1
{quote} and what about 126 wins out of 126 trades, {image}
that is just great
As Above, So Below
  • #1,790
  • Nov 13, 2017 3:24pm Nov 13, 2017 3:24pm
"Stop Loss is Loss Management, not Money Management"
dancingphil
Maadik Hugiis. IQ 69.
  • #1,791
  • Nov 13, 2017 4:18pm Nov 13, 2017 4:18pm
Quoting MoneyZilla
"Stop Loss is Loss Management, not Money Management" dancingphil
Speaking of Management, I have the following "Manager Agents" embedded as part of my automation all working in a cohesive and comprehensive manner for 99% hands-free operation... structured approach to system development.

 

  1. Configuration (input parameters) Manager
  2. Strategy Manager
  3. Account Manager
  4. Position(Trade/Portfolio) Manager
  5. Entry Manager
  6. Exit Manager
  7. Growth (Profit) Manager
  8. Risk/Loss Manager
  9. Margin Manager
  10. Session (Date of week, Time of Day etc) Manager
  11. Environment (Trend/Range conditions) Manager
  12. Trend (trending signal) Manager
  13. Range (mean reverting signals) Manager
  14. Goals (Money/LotSize) Manager
  15. Display (On Screen Monitoring) Manager
  16. FIFO/Hedging Manager
  17. Filtration Manager (Is New Positions Allowed?, IsModifyPositions Allowed? Is this the right time of day/week, Is this the right pair for the current session etc)
  18. Alert Manager
  19. Compliance (audit/logging/monitoring) Manager
  20. Report (stats/dashboard) Manager
  21. and recently added DrawDown Manager

Staying in my lane...
  • #1,792
  • Nov 13, 2017 6:30pm Nov 13, 2017 6:30pm
Quoting VEEFX
{quote} Speaking of Management, I have the following "Manager Agents" embedded as part of my automation all working in a cohesive and comprehensive manner for 99% hands-free operation... structured approach to system development. Configuration (input parameters) Manager Strategy Manager Account Manager Position(Trade/Portfolio) Manager Entry Manager Exit Manager Growth (Profit) Manager Risk/Loss Manager Margin Manager Session (Date of week, Time of Day etc) Manager Environment (Trend/Range conditions) Manager Trend (trending signal) Manager Range...
OMG you seem to be missing a crucial member.....an HR Manager :-)
  • #1,793
  • Edited Nov 14, 2017 1:57am Nov 13, 2017 7:35pm | Edited Nov 14, 2017 1:57am
Quoting Copernicus
{quote} OMG you seem to be missing a crucial member.....an HR Manager :-)
Nope, My 2 employee Global PLC will be outsourced for $10 per paycheck as part of my Compliance Manager function LOL
Staying in my lane...
  • #1,794
  • Nov 18, 2017 11:45am Nov 18, 2017 11:45am
Quoting priceaction1
and what about 126 wins out of 126 trades,
It sure sounds nice, but it doesn't have any meaning if you don't define the trade conditions first. For example: No scaling in or out, TP 1pip + Spread, SL NA -> 126 winners in a row isn't a problem, but ultimately it's still a fool's game. Same goes for all the grid, martingale, whatever "systems".
  • #1,795
  • Nov 18, 2017 1:57pm Nov 18, 2017 1:57pm
Back on topic: the higher edge within a single candlestick. (BTW, why "higher"? Higher than what exactly?)

I've been secretely stalking these single candlestick threads from the start, without really knowing why. I mean, a candlestick is just 4 arbitrary data samples combined, two fixed in time and two fixed in price, right? I'm still kind of hooked on this shit, it's probably the old "I know something and you don't, nanananananaaa"

A few weeks ago I had one of these lightbulb moments you get sometimes when you torture your brain day in, day out with the same topic over and over again... see my first post in this thread, which basically states that "candles" constructed from random data exhibit the same OH/OL/CH/CL ratios that are found in candles constructed from actual market data.

How is that useful? Isn't that the usual whining about how the markets are all random anyway and the b0rkers are out to get you and nobody makes any money and it's all just a scam yada yada yada....?

Let me try to explain. I'm 100% with hanover when it comes to trading/gambling random data, it's not possible to gain an edge and you'll never be able to develop a model to overcome the costs of the game, no matter if it's casino games or whatever. That does *not* mean you can't walk away from the roulette wheel with a fat grin and a few K in your pocket now and then, it just means that - whatever "system" you employ - you can never be sure that variance won't kill you today or tomorrow. You can't create a game with positive expectancy on random data.

Working with that idea in mind, I suddenly found myself with a lot less work to do - no need for testing everything from SMAs to rocket science stuff a la John Ehlers and weird candle patterns like the famous "Three Horny Virgins on a Rolling Hilltop" anymore. I could focus instead on one thing only: where does market data deviate from random data? Is it possible to define price points that show a better-than-random propensity to become turning points?

Like always, when you know what you're looking for, the search becomes so much easier. Easy enough in fact that even I could come up with a workable idea that could be destilled down to a less than perfect, but still useful algo that shows price points where random more or less goes out the window and the likelyhood of a reversal increases considerably over the likelyhood of continuation (something that is *not* possible with random data, see also "Gambler's Fallacy").

I doubt very much that this is what CrucialPoint et al have in mind with their single candlestick. Or maybe this is what they had in mind - place some kind of Zen puzzle here to force people to think. Either way, thanks a lot you guys. I was kind of stuck in what I was doing without making any progress, and you kicked me out of it.

Here's what it looks like:

Attached Image (click to enlarge)
Click to Enlarge Name: eurusd-e-m1-roboforex-cy-ltd.png Size: 73 KB


Obviously there's still a ton of work to do, this is just a really crude model I hacked together over the last few weeks. MQL is probably the worst thing I ever saw since VB, and who the fuck designed the code editor, the Spanish Inquisition?

Like I said, less than perfect. In directional moves, it might trigger in the opposite direction occasionally, and it'll take someone smarter than me to work around that. Traders discretion is advised. Sometimes, it's also just plain wrong, but that's really quite rare. And if price goes into one of these occasional ~5 pip ranges, like before news, the logic breaks down completely and spews all over the place

But most of the times, it's "about approximately right", and there are quite a few cases where it's so fucking spot on you can feel like the King of Forex
  • #1,796
  • Nov 18, 2017 2:29pm Nov 18, 2017 2:29pm
@Broke: Congratulations. Would be great if you'd share some hints in the future. Was working on something similar but without much success.
  • #1,797
  • Nov 18, 2017 4:15pm Nov 18, 2017 4:15pm
Quoting AntiCre
@Broke: Congratulations. Would be great if you'd share some hints in the future. Was working on something similar but without much success.
I hope I didn't give the impression that was a "buy at the blue dots, sell at the orange dots" kind of deal. I'm pretty sure that wouldn't be fun at all. The dots just mark places where I think random gets weak and something else might take over. It's still just a market - if Deutsche or UBS or some CB decides to start selling or buying just right here, all theoretical wanking dies a quick sudden death

What I wanted to hint at is that you can stop looking at indicators, candle patterns, Fibonacchi, Pivot points, Murray Math, Idiot Wave etc., because if you look at market data as a continuous stream, it will exhibit the same characteristics as data from a random source, which renders all approaches based on linear math useless. Maybe that's a naive and unsophisticated point of view, I don't know. I'm not a quant math wiz with a PhD in $insert_some_insane_stuff_here. It's just what I can glean from what I can understand and code. Decide for yourself.

For me, the dots and candles and stuff need to be seen in the context of what actually moves price, which is macro, sentiment, news, and bigger fish pushing price wherever the fuck they want. This is the really important stuff, and understanding it should be the first priority for everyone who doesn't want to be just a liquidity provider in this game.

But seen in this context, I think the concept of looking where current price deviates from what you would expect in a random distribution could be really valuable for us retail muppets. It's a blunt instrument, but it's still a lot better than all the TA mumbo jumbo that is out there.
  • #1,798
  • Nov 18, 2017 4:44pm Nov 18, 2017 4:44pm
Quoting Broke
{quote} It sure sounds nice, but it doesn't have any meaning if you don't define the trade conditions first. For example: No scaling in or out, TP 1pip + Spread, SL NA -> 126 winners in a row isn't a problem, but ultimately it's still a fool's game. Same goes for all the grid, martingale, whatever "systems".
if you remember the high of cable.

https://www.forexfactory.com/showthr...89#post7581289

https://www.forexfactory.com/search....182565&page=37
Attached Image (click to enlarge)
Click to Enlarge Name: led.png Size: 87 KB
don't predict direction, but dance with flow of water
  • #1,799
  • Nov 18, 2017 4:47pm Nov 18, 2017 4:47pm
Only two are needed.

Attached Image
  • #1,800
  • Nov 18, 2017 5:20pm Nov 18, 2017 5:20pm
Quoting venividivici
Only two are needed. {image}
Really? I guess that you think that both imply strenghtened up force. However, I can show you many cases where these won't work. For instance, the exhaustion of a trend is often shown in a very long marubuzo and prize falls/rises afterwards.
1 trader viewing now