Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Why you shouldn't show how much you have made in FX 35 replies

Incorporating Slippage Assumption in backtesting 4 replies

Market Predictability - Technical Signals 25 replies

So I accepted a Prop Trading job in South Beach Miami 43 replies

Contesting the Accepted View of Bernanke 0 replies

Attachments: why have you accepted the price-predictability assumption?
Exit Attachments

why have you accepted the price-predictability assumption?

I think one of the biggest differences between profitable traders and unprofitable traders is that a profitable trader should pick a genuine price chart from a bunch of random ones, as nubs has apparently done. If you cannot identify nonrandom chart behavior you have no hope of exploiting it. Demonstrating nonrandomness in one chart though doesn't indicate nonrandomness in other charts - especially across different instruments. Finding patterns on a gold chart doesn't mean gbp/chf is exploitable.

As you say, what nubcake called "swagger" is probably shifting volatility, which is a big indication of nonrandomness. I trade 1H charts with a 10 period ATR, which I use to time entries. A randomly generated price chart will show variable volatility and will exhibit patterns suggesting nonrandomness, but those patterns won't be repeated regularly. The way volatility rises and falls predictably at the same time every day is something a random generator cannot reproduce - even if its entries are increased and decreased throughout the day.

The fundamental picture also testifies against EMH. Pull up a daily USD/JPY chart, scroll back to reveal over a year of price action, and try to pinpoint the moment Shinzo Abe announced his intentions to pound the yen. It isn't hard to spot. Or look at the last six years on EUR/USD and correlate price action with various stages of the global financial crisis. These sweeping moves on long term charts create tremendous directional bias on short term charts.

Finally, EMH implies that it is impossible to consistently and sustainably profit from price movements. Creative MM such as martingale can delay drawdowns to appear profitable in the short term, but if there is just one guy in the world steadily profiting from forex without doing this, that means EMH is wrong. And there are several MyFXbook accounts showing performances that cannot possibly be achieved without exploiting clear repeating, nonrandom chart behavior.
si hoc legere scis nimium eruditionis habes
why?

because everything else in life is in waves and a trend is a wave. I learned to not go against it. The patterns (like breakout pattern) just give me opportunity to risk less because they show me the moment and location were 1 side is going to win.
It's a game.
Quoting Fundaments
Hello Redlion, When you analize a security with a time horizon bigger than a couple of months you are making a prediction of the future. The market efficiently gathers all predictions and shows us the average vision. After that, each individual will have its vision based on the same information and will therefore obtain different results.
where is that thread with random charts?
.
I'm sure there are traders who studied charts and prices very deeply and discoverd some innefficiencies.

Well, even I found one in the EUR/USD after "going old school".
But it's very hard I must confess.
The bigger the market the more efficient it is.
Inefficiencies are only temporary.
Quoting GEfx
{quote} Hi Red, the source of my evidence is -- me. My trading is based on my experience as a trader, my educational background in economics and finance, etc. As for market structure, there are two parts of market structure to consider. The first is the story the charts are telling you every day; the second involves who is in control of the market, and you identified them, with heavy emphasis on central banks (CB), tier 1 banks, and hedge funds. Forex is a heavily manipulated market, and those three groups do much of the manipulation. Often, the...
Hi Red, I've been bothered by my use of the word predictable in this post. I think an accurate way to state it is that you can never predict what the market will do, but you must be able to anticipate the market in order to plan your trades. That might be splitting hairs, but I was thinking about my trading process and at no time do I predict. The process is interpret market structure, anticipate PA, plan trades, and execute the plan. Predicting means that you know what the market is going to do; no one knows what the market is going to do (outside the trading rooms of the CB's, maybe). Case in point: I made no trades on Monday or Tuesday this week, not because I have some great predictive powers, but because PA did not match my analysis of market structure, and did not fit my assumptions/plans...until today. Today, PA finally lined up with my assumptions on the AUD and NZD, and all that was left was to execute the trading plan on the EUR/AUD, GBP/NZD, etc. None of that involves predicting. It's just trading.

Finally, while over the last 20 years I have read and studied a ton of material on market theory, I have never once found any of it relevant to trading the forex.
see the reason for this thread is because, we can come up with answers that make intuitive sense, or what we think/postulate the market is ......yet when objective studies are implemented such as studies that in other areas are used for random variable filters such as Serial correlation correlograms show that the market returns are neither totaly normally distributed and show the same behavior as random noise.


here is a qqplot showing the distribution of euro log returns against a normal distribution
Attached Image (click to enlarge)
Click to Enlarge Name: qqplot.jpg Size: 38 KB


here is a histogram
Attached Image (click to enlarge)
Click to Enlarge Name: hist.jpg Size: 35 KB


here is an auto correlation (correlogram plot) showing the behavior with lag of 5 acting as random
Attached Image (click to enlarge)
Click to Enlarge Name: acf5.jpg Size: 36 KB


lag 21
Attached Image (click to enlarge)
Click to Enlarge Name: acf21.jpg Size: 37 KB


lag 126
Attached Image (click to enlarge)
Click to Enlarge Name: acf126.jpg Size: 42 KB



here is what a positive autocorrelation looks like (where there is persistance or in other words one observation is more likely to carry momentum as a result of a previous observation)

at lag 5
Attached Image (click to enlarge)
Click to Enlarge Name: acfpositive5.jpg Size: 37 KB

at lag
Attached Image (click to enlarge)
Click to Enlarge Name: acfpositive21.jpg Size: 41 KB


at lag
Attached Image (click to enlarge)
Click to Enlarge Name: acfpositive126.jpg Size: 44 KB
AVT INVENIAM VIAM AVT FACIAM
Quoting GEfx
{quote} Hi Red, I've been bothered by my use of the word predictable in this post. I think an accurate way to state it is that you can never predict what the market will do, but you must be able to anticipate the market in order to plan your trades. That might be splitting hairs, but I was thinking about my trading process and at no time do I predict. The process is interpret market structure, anticipate PA, plan trades, and execute the plan. Predicting means that you know what the market is going to do; no one knows what the market is going to...


I understand what you are saying, and I think I should clarify, I am using predictable in the sense of non-random, that in fact one can extrapolate possible future movement from past observation or current economic factors. I also understand that there are profitable traders, I have had long profitable runs, but I want to question what is at the core of our endevour and from the objective data it seems that the market is more efficient than what we believe it to be. Perhaps in this thread as you challenge my Ideas, I can improve further as a trader. I thank you for your participation and continued discussion.

I am not convinced that is skill what lead me to profit......my null hypothesis is that is just random luck.

and if infact there is something to be learned about trading skill...that would be very interesting.
AVT INVENIAM VIAM AVT FACIAM
  • Post #29
  • Quote
  • Edited at 2:14am Dec 12, 2013 1:45am | Edited at 2:14am
In 2013, the Nobel prizes in Economics went to three individuals. Two of them based on their analysis of asset pricing. Out of the two one argued for efficient markets and the other for inefficient markets; both won Nobel prizes.

There is no cut and dry here. The markets are neither always efficient nor inefficient. The word "predict" does not belong in speculative trading. You are guessing plain and simple. Sometimes you are going to be correct and sometimes you wont.

In my opinion, there are two ways to trade. One is such as GEfx described where he does his due diligence and trades FA on daily charts. Yes, I can see how doing this type of trading can give the trader a sense of confidence in their ability to "predict" price. You are basing your trades and assumptions on real empirical data. When you zoom out this far, macroeconomic events will make a difference on the chart.

Then you have the 2:1 RR chartists who trade the 15M and have 30 pip stop loss. These my friends are the poster boys for random walk. To all those who are saying that whenever humans drive price action it will always have a degree of predictability I ask you this: Are you privy to the behavioral patterns of the total collection of these humans at every given point? Because if you are not it might as well be random. For us retail traders, price may not be a true identical and independently distributed process but it comes very close.

Just look at some of the theory described in the first two pages; at some of the charts Red Lion posted. All that analysis still doesn't tell us anything relevant for the retail level. How much time goes into doing these studies just for someone to tell us that the market *MIGHT* behave like this to win a Nobel prize.

I'll finish with this. We all like to think we are sophisticated. A lot of us have degrees in economics or financial engineering and we sometimes like to think that in some way we are better traders because of it. But are we really? Does being able to run multiple regression and posting charts really do anything for us? A "simpleton" can create an EA that gets him into a trade based on his specific method, have no stop loss, let it run until it hits profit and keep doing this over and over again based on the principal of mean reversion and collect profit. May take him years but he will undoubtedly collect profit. Because of this he is labeled as a "newb" or "gambler" or flat out "loser" compared to the guy who has a "sophisticated" method of trading that requires him to spend many, many hours to collect a profit that's not significant when you actually break it down by dollar earned per hour spent. Even if the EA trader collects $500 in a year, that's $500 in which he did not have to even spend 1 minute on trade analysis while the guy who made $5000 spent a 1000 hours. The EA trader made $500/hr if we assume it took an hour to setup the EA and the "sophisticated" trader made $5/hr. At the end of the day it doesn't matter how you trade; the only thing that matters is being able to extract the most amount of money from the market while expending the least amount of effort. In my opinion, that is the best type of retail trader.

If you claim to be able to predict the markets long term then this is an assumption you made based on short term returns and ego because I promise if you were able to do this consistently and return a significant profit year after year, you would not only not be here but better than 99.9% of actual multimillionaire traders who do this for a living.

Just wanted to provide some food for thought. Forget about efficiency or inefficiency. Embrace randomness and figure out a technique for trading that can profit from it. Anytime you only have one leg in a trade you are just plain speculating.
Remember, Red. Hope is a good thing, maybe the best of things...
Quoting ForexFury
At the end of the day it doesn't matter how you trade; the only thing that matters is being able to extract the most amount of money from the market while expending the least amount of effort. In my opinion, that is the best type of retail trader.
^ This

Quoting ForexFury
Forget about efficiency or inefficiency. Embrace randomness and figure out a technique for trading that can profit from it. Anytime you only have one leg in a trade you are just plain speculating.

^ This too. My trading took off as soon as I realized that the market is random in the short term and the only thing anyone can do is to go with the flow.
Fight of the Century: Keynes vs Hayek, Round 2
  • Post #31
  • Quote
  • Edited at 8:21am Dec 12, 2013 2:51am | Edited at 8:21am
The other day I've taken the time to read the article you have posted: http://www.technicalanalysis.org.uk/trend/Alex64.pdf (thanks for sharing btw). From it I was shocked to learn that the bulk of economists studying the subject as of year 2002 were not doing science. Either (a) economists do not want to infirm their hypothesis, (b) they don't know the scientific method in order to quantify behavior, or (c) they are not thinking. What economists have done is basically take arbitrary points in time (eg, Friday's close/weekly close), and compute how these autocorrelate. Not surprisingly, what they have found using such 'method' is that they were right all along in defending the proposition that the markets are a random walk. Should, scientists from other fields conduct their work the same way, they would likely have concluded everything is a random walk, and we would still be living in the dark ages.

Alexander, the author of the article, by simply doing one thing right, was able to find predictable market behavior (trends). What did he do right? He objectively defined a plausible relation between a variable (predictor) and a behavior (prediction), and measured it. Alexander used brains + the good old scientific method, and not surprisingly the result was sound science.

Trading strategies are much like what Alexander did (usually more complex due to more predictors being used), and nothing like what most economists do, or were doing until 2002.

PS: Taleb is MORE right about the economists, than I ever thought possible!


EDIT

Btw, does this look like random walk? (The account shown is not mine. Serves merely to illustrate the point.)
Attached Image (click to enlarge)
Click to Enlarge Name: technical.jpg Size: 98 KB
"To hold, you must first open your hand. Let go." - Lao Tzu
Quoting vox dei
Btw, do these look like random walks?
I don't know if it LOOKS LIKE but it IS!
Despite the ego of people everybody's winrate isn't fixed but oscillates around an average winrate. Same thing for the R:R. Even if you use fixed SL and TP the slippage creates variation. Therefore your expectancy is a random variable. Consequently your equity curve is a random walk.

Each time I read a thread where the term random appear I keep repeating myself "do they know want random means?". It happens that in English the word random can have up to 6 different translations in my language (in French: aléatoire, hasard, fortuit, incohérant, quelconque, arbitraire)! I suppose it is the same in some other languages. That doesn't help. Worse of all I looked in the Oxford dictionary for the definition and IT IS WRONG!

http://www.oxforddictionaries.com/de...nglish/random:
Quote
Statistics governed by or involving equal chances for each item
Random doesn't at all imply equidistribution! There are a lot of "let's toss a coin" examples in FF. Today "let's roll two dice". The sum of the dice is a random variable. 12 can be made with a double 6. But 7 can be realized with 1+6, 2+5, 3+4, 4+3, 5+2 and 6+1. 7 has six times more chance to come than 12. This is not at all "involving equal chances for each item". You cannot predict the outcome but you can systematically bet 7 which is the most likely.

Can you guess the next value of this sequence? 54756, 25625, 15844, 41056, 11025, 10404, 1600, 25600, 13600, 29600
No you can't? Yet it is absolutely not random at all. As long as you don't know the formula to generate the next number it is random FOR YOU. Should the market be 100% deterministic you would still not know the formula to generate the next candle.

Oh you think you are so smart you can?! Trends are easy to spot? Have a look at the picture below. DJIA daily. From the first higher low (13 Jul 2009) to the latest highest high (29 Nov 2013) it went up 8108 points. 8108 over 1130 sessions that's 7.18 points per trading day. Look at the bottom. The ATR is between 70 and 450. The trend is less than 10% of the price movement and can be as low as 1.5%. This is a random walk with drift (http://people.duke.edu/~rnau/411rand.htm). The trend is burried in ~95% of not-the-trend.
Attached Image (click to enlarge)
Click to Enlarge Name: DJIA.png Size: 62 KB
No greed. No fear. Just maths.
Quoting the redlion
here is a qqplot showing the distribution of euro log returns against a normal distribution
May you generate a qq-plot against a Cauchy distribution?
No greed. No fear. Just maths.
Quoting PipMeUp
{quote} I don't know if it LOOKS LIKE but it IS! Despite the ego of people everybody's winrate isn't fixed but oscillates around an average winrate. Same thing for the R:R. Even if you use fixed SL and TP the slippage creates variation. Therefore your expectancy is a random variable. Consequently your equity curve is a random walk. Each time I read a thread where the term random appear I keep repeating myself "do they know want random means?". It happens that in English the word random can have up to 6 different translations in my language (in French:...
Hi PipMeUp,

20/20, this is another reminder of how much I suck at explaining things.

My point/belief, call it whatever you like: The number of variables at play in the market at any given moment are unknown, their exact weight unknown. Hence, the market taken as whole in fact should behave like a random walk (=unpredictable), and if one was to test for this hypothesis using only a dependent variable (eg, taking arbitrary points in time) and doing autocorrelation like the economists studies seem to focus on according to Alexander, then imho confirmation of one's own hypothesis would be the likely result. No usable/tradeable knowledge would be produced.

If however one varies an independent variable that is supposed to affect a dependent variable, and measures the independent variable impact indirectly through the dependent variable like Alexander did (and like it is commonly done in scientific experiences), then results other than the 'the market is random walk' begin to emerge, small glimpses of predictability within the huge unpredictable market picture begin to appear. Do this using 2+ independent variables and the magnitude of the glimpses should increase a bit... Usable/tradeable knowledge would be produced.

On a sidenote, I've just observed that you have a wrong impression about me. The screenshots serve to exemplify the point, not as auto-promotion. One of the accounts even belongs to some stranger. I do not have any pretension/illusion of being smarter than anyone else here. In fact I consider most individuals that have an interest in the markets above average intelligence individuals. Some such as yourself, Hanover, theredlion, etc etc are way smarter than me.

Quoting vox dei
Financial markets are overflowing with above average IQ individuals. The very same individuals that lose on a regular basis. Clearly, IQ is overrated.
It's all good btw.

Cheers
vox

EDIT: I've edited previous post and let only the account that belongs to complete stranger as example.
"To hold, you must first open your hand. Let go." - Lao Tzu
Quoting vox dei
On a sidenote, I've just observed that you have a wrong impression about me. The screenshots serve to exemplify the point, not as auto-promotion. One of the accounts even belongs to some stranger. I do not have any pretension/illusion of being smarter than anyone else here. In fact I consider most individuals that have an interest in the markets above average intelligence individuals. Some such as yourself, Hanover, theredlion, etc etc are way smarter than me.
Don't take it personal. There is nothing at all personal in my answer. It isn't important to whom the account belongs. Being up or down going isn't important either. I was just answering your question "Btw, does this look like random walk?". Per definition it is a random walk, irregardless of what it might look like.
No greed. No fear. Just maths.
Quoting the redlion
{quote} one can extrapolate possible future movement from past observation or current economic factors.
He Red, we look at this from different perspectives. Maybe you are seeking statistical proof that trading is profitable. I have a simplistic view of the currency markets. The forex charts represent the activities that are being carried out by CB, tier 1-3 banks, and hedge funds (they are the big players). These underlying activities are documented in the business processes that these institutions use to implement policies and conduct business. For example: CB's act to arrest a sudden and unwelcome move in a nations currency (usually in conjunction with that nation's Treasury); a bank acts ahead of an option expire to protect a large option; a bank begins to transfer $100M to a customer's bank in Germany; a hedge fund transfers $50m in US 10-yrs to 10 year bunds. There are no chat rooms where these guys can coordinate their activities. They use the charts to talk to one another. My little view of the whole thing is that they are extrapolating past observations into the future, using current economic assumptions, and using a daily work plan that must be completed.

An important factor for our trading is the understanding that, the fastest way to become an ex-bank trader is to buy high and sell low. Collectively they all have the same objective, and they are all doing essentially that same thing: move the market, wait for a small pull back, then move it again. When all of that work is completed, reverse it all to return to where the day started (keep in mind that while they are moving money out of the bank, they have held off the work to move money into the bank).

Of course, all of this changes at FA time, except their objectives. So, do some bank traders lose money? Yes. Do banks sometimes lose their clients' money on dumb predictions? Yes. Do banks lose money trading (and I'm talking about their balance sheets)? Rarely. Therefore, it's best to ignore trader talk and simply trade the charts, because doing so puts you in line with the real underlying purpose of the forex. Doing this has very much put me in a statistically advantageous position. However, I have no idea how to put the math together to prove it.
Quoting PipMeUp
{quote} Don't take it personal. There is nothing at all personal in my answer. It isn't important to whom the account belongs. Being up or down going isn't important either. I was just answering your question "Btw, does this look like random walk?". Per definition it is a random walk, irregardless of what it might look like.
All right. I've fell into a common human mistake, auto-reference. Thanks for clearing that up.
"To hold, you must first open your hand. Let go." - Lao Tzu
Quoting PipMeUp
{quote} I don't know if it LOOKS LIKE but it IS! Despite the ego of people everybody's winrate isn't fixed but oscillates around an average winrate. Same thing for the R:R. Even if you use fixed SL and TP the slippage creates variation. Therefore your expectancy is a random variable. Consequently your equity curve is a random walk. Each time I read a thread where the term random appear I keep repeating myself "do they know want random means?". It happens that in English the word random can have up to 6 different translations in my language (in French:...
Go study bro, the random walk theory is outdated by 20 years. You obviously haven't been paying attention to the research papers in finance, and you're falling in the fallacy that because something looks like a random walk, then it must be a random process, but turns out it is not. There are papers who test if liquid markets are stochastic processes and they reject the hypothesis with a high margin on the p-value. Search for 'chaos and financial markets', Lyupanov exponent, Hurst exponent, and go study basic statistics and chaos dynamics theory because you don't get it.
  • Post #39
  • Quote
  • Edited at 10:06pm Dec 12, 2013 9:52pm | Edited at 10:06pm
Quoting the redlion
I am curious on what justification do any of us have to have accepted the assumption of price predictability and market innefficiency? Is it faith or hope that in fact innefficiencies exist in the market and that we are actually able to profit? or do we actually have evidence and sufficient reason to support the asumption of an inneficient forex market? In economics and finance it is often said that profit comes from information, a well informed investor is able to profit from temporary market inneficiencies or mispricing of certain assets, if in...
Well first proof it that, if it would be totally random without the capability of prediction then nobody on earth could make money with it.And just looking at Goldman Sachs's or Senvest Partners LTD Hedge Fund's profit, that shows exactly the opposite.

So they are mostly market makers in that sense, but still they know when to put in the market a huge money to cash with it.While the Hedge Funds are only professional traders and still can speculate pretty well.


Or look at Warren Bufett, a pro stock trader, since the stocks have a tendency to go up, he just bought a few stocks and waited until they reverse and then sell it.This simple strategy bought him 70 BILLION DOLLARS!!


Forex is a dual game on the other hand since you have 2 markets colliding.But the method is the same, check if the EURO is stronger than the DOLLAR, either of them is stronger, the price goes in that way.It is predictable if you look at the fundamental point of view.


But you guys are discussing weather it is predictable in the technical point of view.Well this might not be so simpe, but lets take a look:

In mathematical sense i view the market as the "universe" of trading strategies.So it contains all types and combinations of strategies that can ever exits.So the total winrate of the market is 1 or 100%, because it contains all types of strategies and all combinations of it, weather it is based on: price action,divergence, indicators, chart patterns, eliott wave, etc.


So lets look it closer:

-My first argument is that, all combinations and types of strategies are in the market, and so by looking the mere fact that if you take strategy X and strategy Y they will be inequal, there is no 2 strategy on earth that can give the same results either on winrate, RR, or expectancy of profit point of view.


-So i can confirm that all strategies are unique, but if we assume this that means that you can order the strategies by their expectancy.

So strategy X has an expectancy of -0.1 ,strategy Y has of 0.1, strategy Z has one of 0.3

So you can assume that Z is better than Y and Y is better than X, so you can order them ascendingly by X,Y,Z


-So there will be always a system that will be better than the other, and the better system you get the closer you are to the absolute best one a Holy Grail


-But on the other hand i was talking about the total WR of the market being 100%, that means that there must be a system there with 100% winrate too, because if you look, 99% winrate might not have a sufficiently good RR to be profitable, like a system where your TP is 1 pip and your SL is 1000 pips.Of course your winrate will be very close to 100% but 1 wrong trade and all your profit flies away...


But a winrate of 100% will always win.So tehnically there is no system which can predict the market by 100%, because prediction is a bad word, the better word is speculation.For example the system where you dont have a SL is a 100% winrate (if you have enough capital to never get a margin call)


So that is why we have to weight the systems not by WR and not by RR but by the combination of them, the expectancy.


So to answer your question, my view is that, there are system which have a 100% winrate, but to predict the market with 100% is impossible, because the market will always have random elements to it even if its not totally random, so you cant even predict it with 80% accuracy let alone 100%.But of course a 51% accuracy would be enough (if its not tied and doesnt modify the RR with it, as the 1 pip TP and 1000 pip SL does) , but to answer that question i dont know.


Since if you tie the RR and the WR together than they have influence on eachother, but to predict the winrate indifferently , and not influencing the RR, its really a mistery.


That is what Neural Networks are trying to do

"There's a sucker born every minute" - P.T. Barnum
  • Post #40
  • Quote
  • Edited at 12:50am Dec 13, 2013 12:33am | Edited at 12:50am
Quoting Jasus
{quote} Go study bro, the random walk theory is outdated by 20 years. You obviously haven't been paying attention to the research papers in finance, and you're falling in the fallacy that because something looks like a random walk, then it must be a random process, but turns out it is not. There are papers who test if liquid markets are stochastic processes and they reject the hypothesis with a high margin on the p-value. Search for 'chaos and financial markets', Lyupanov exponent, Hurst exponent, and go study basic statistics and chaos dynamics...
Jasus,

You must be be a multimillionaire since you can predict market movement. I'd love to hear more about how Forex is not a random walk from you.

Empirical data people. F*ck theory. Show us results!!
Remember, Red. Hope is a good thing, maybe the best of things...
0 traders viewing now