Statistical mean of the market [quant corner] | Page 6 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Explore Real (Statistical) Market Edges 21 replies

The Birth of a Quant Trader 107 replies

Quant trading & QuantLib 6 replies

Quant Corner 12 replies

questions from a future quant 39 replies

Attachments: Statistical mean of the market [quant corner]
Exit Attachments

Statistical mean of the market [quant corner]

  • #101
  • Oct 15, 2014 2:32pm Oct 15, 2014 2:32pm
Quoting Proximus
{quote} Just take a look at the MACD formula to understand where the problem comes from. MACD Line: (12-day EMA - 26-day EMA) Signal Line: 9-day EMA of MACD Line MACD Histogram: MACD Line - Signal Line So its an oscillator, but what oscillates ? The exponential MA. So you see all derivative indicators or oscillators, that is how i call them because they are derived from MA's, are using moving averages. The base of all oscillators are moving averages. So if the foundation shakes, the entire oscillator is worth nothing. So the problem is with the...
Yupp. I think i should add the the EMA i was talking about is something i ONLY use on E/U. I dont trade anything but the E/U.
My signature is: "Classified".
  • #102
  • Oct 15, 2014 2:35pm Oct 15, 2014 2:35pm
Quoting Proximus
{quote} What makes you think you can compete with their cumulative knowledge?
Im "cheating". Using a little thing called the "subconsious". Its way more advanced then any computer to date.
My signature is: "Classified".
  • #103
  • Oct 15, 2014 2:37pm Oct 15, 2014 2:37pm
Quoting Proximus
{quote} The Hedge Fund named Renaissance Technologies has atleast 91 PHD's working on that field.
A lot of cooks doesnt necessarily make a good soup.
My signature is: "Classified".
  • #104
  • Oct 15, 2014 2:53pm Oct 15, 2014 2:53pm
Quoting yaed
{quote} Yupp. I think i should add the the EMA i was talking about is something i ONLY use on E/U. I dont trade anything but the E/U.
The EMA is good, but it's far from the best. Other ARIMA models are already superior to it.

Quoting yaed
{quote} Im "cheating". Using a little thing called the "subconsious". Its way more advanced then any computer to date.
I dont think that will work, because human emotions will cloud your judgement, you really need an objective tool for it. I don't think trading based on gut feelings has ever made anybody a millionaire.

Quoting yaed
{quote} A lot of cooks doesnt necessarily make a good soup.
That may be true in other cases, but Renaissance is amongst the top 10 best hedgefunds in the world. James Simons the CEO of it is called the "Quant King", the king of all quantitative traders
"There's a sucker born every minute" - P.T. Barnum
  • #105
  • Oct 15, 2014 2:56pm Oct 15, 2014 2:56pm
Renaissance Technologies was sued recently for some tax illegalities. Some nice extracts from the docs about what they are doing.

A few things which go counter the usual wisdom on these forums. They manually intervene over the algos when market changes, and they CONSTANTLY update their models. Unlike most here, they do not believe in models that are profitable for 12 years, over all the markets. They optimize DAILY.

http://www.hsgac.senate.gov/subcommi...everage-limits

In some cases reviewed by the Subcommittee, the investment advisor used algorithms to engage in a high volume of trading, executing more than a 100,000 transactions per day. Many of those trading positions lasted minutes, and the overall composition of the securities basket changed on a second-to-second basis. One basket option account later reviewed by the Securities and Exchange Commission (SEC) was found to have experienced 129 million orders in a year

To execute the trades, Barclays gave RenTec direct market access through its proprietary trading software, enabling RenTec to place trade orders instantly into the market and receive immediate executions. The trade is done within milliseconds. In almost all instances, the trades were placed and executed by computer from RenTec's facilities, using its algorithms to initiate and execute the trades through the bank's proprietary trading software. The trades were typically executed with no human intervention on either side, except when RenTec personnel adjusted the algorithm parameters.

When RenTec took control of the option accounts at Deutsche Bank and Barclays, it employed a proprietary investment strategy that utilized a computer-based set of algorithms which was managed by over 90 professionals and required continual adjustments. The trading strategy used hundreds of thousands of short-term trades per day and produced an average of 26 to 39 million trades per year

Still another document, prepared by Barclays, indicated that RenTec conducted about 500,000 trades per day, describing it as a “very high frequency trader"

RenTec explained to the Subcommittee that those employees constantly worked on and modified the algorithm, carefully modeling new factors and introducing them into the algorithm over time to alter how it decided what trading recommendations to make

The models don’t see the government intervention but we do and we are nervous that something could happen. So we have actually intervened and we do that from time to time when things like this happen.

The end result was that RenTec’s trading strategy changed on a regular and ongoing basis.

RenTec placed approximately 129 million trade orders, involving nearly 10,000 different financial instruments across 18 countries

In order to be able to constantly adjust its trading model and implement its strategy, RenTec employs a large number of employees with doctorates and training in advanced mathematics, physics, and computer science. 162 According to RenTec, over 90 members of RenTec’s staff hold doctorates in mathematics and other scientific disciplines.
  • #106
  • Oct 15, 2014 3:07pm Oct 15, 2014 3:07pm
Quoting Adal
Unlike most here, they do not believe in models that are profitable for 12 years, over all the markets. They optimize DAILY.
They optimize daily because they trade at the milisecond level. So if I trade at the weekly level (which I said in my first posts) why would i optimize so frequently?

The average trade duration with my current model is about 6-7 days long, so it could work for a much longer time.

The model itself could work for 12 years, but the parameters have to be changed to adjust to new data, but the model itself, given that only economic crisises are distorting the markets sharply, could work until the next economic crisis.

My model was tested on EUR/USD at the moment, and encompasses both the trending periods of 2002-2008 as the ranging periods between 2008-2014.
If it works on both with a decent adaptability what other market shape can destroy my model?

So in conclusion: the 12 year survivability is ok, but the parameters themselves have to be optimized
"There's a sucker born every minute" - P.T. Barnum
  • #107
  • Oct 15, 2014 9:21pm Oct 15, 2014 9:21pm
My program was getting too slow as i added new features to it to calculate my stuff better, so it really got slow. So I had to check my code and make it more efficient, therefore i packed more stuff into variables and functions rather than just sitting there, and i`ve succesfully increased my calculation speed by +41.45%.

Which is a tremenduously big advancement, now i can calculate stuff much faster, so my research speed will go up.

Attached Image (click to enlarge)
Click to Enlarge Name: x.png Size: 82 KB


This is how the processing speed varies over 400 iterations, it really a big difference, and I wasted much time due to this calculation ineficiiency, but no more.

From an average 149 ms / iteration we got an average of 86 ms/iteration, almost halved.

So yes my research will be almost twice as fast.
"There's a sucker born every minute" - P.T. Barnum
  • #108
  • Edited 1:36am Oct 16, 2014 12:50am | Edited 1:36am
Quoting Proximus
{quote} I dont think that will work, because human emotions will cloud your judgement, you really need an objective tool for it. I don't think trading based on gut feelings has ever made anybody a millionaire.
You assume human emotions will cloud somebodies judgment. You assume gut feelings are involved. Maybe thats true, maybe its not. Im just saying the subconsious handles massive amounts of data. I have been training to get rid of this pesky thing called "feelings". By that i dont mean the good "feelings", but the other stuff.

Let me just give you an example: Sometimes i open up a chart of a pair and i place a trade within 10 seconds or so. Then i open up another pair, place another trade and so on until i have a bunch of open positions. I guess this is sort of a grand mean average sort of an idea. Most of the time i do pretty good within a few hours and i close here and there if its in nice profit, if its not maybe i reenter. This is not what i usually do, i just trade E/U.

Also let me give you another example: On the 30th of september i said to my woman "I think its going to turn." In other words that the E/U drop would stop or something like that. Look at the charts. However, i didnt take my hunch seriously enough. Had a drawdown of 20% for a bit, then i closed it out at a loss of 12% a few minutes later. Im still on the good side, but anyway, i really got to thinking what the hell i was doing. I dont think it was a "gut" feeling. It was more of dream. That was true. Maybe i saw that in the charts. Maybe i had some help from my formulas. Maybe i reached that conclusion from all my many hours of analysis in 5-15 different ways. I honestly dont know, but it did not seem to come from my gut. I know its a figure of speech.
My signature is: "Classified".
  • #109
  • Oct 16, 2014 12:54am Oct 16, 2014 12:54am
Quoting Adal
Renaissance Technologies was sued recently for some tax illegalities. Some nice extracts from the docs about what they are doing. A few things which go counter the usual wisdom on these forums. They manually intervene over the algos when market changes, and they CONSTANTLY update their models. Unlike most here, they do not believe in models that are profitable for 12 years, over all the markets. They optimize DAILY. http://www.hsgac.senate.gov/subcommi...everage-limits...
Thats interesting. To us poor people (not 50.000.000 $ lying around) that sort of thing is out of the question, the speed and amount of orders i mean. But i think a neural network can be a nice thing. Those things REALLY knows numbers, and fast. But of course the hedgefunds has the hardware and all that. So if people like us are going to do something like that its a matter of REALLY thinking outside of the box. Thats OK with me, i never even WHERE in the box. I did see a shrink and he OKed me so im not fucked in the head at least
My signature is: "Classified".
  • #110
  • Edited 3:44am Oct 16, 2014 2:18am | Edited 3:44am
Want to have something to chew on for a bit?
Put a envelope period 21, MA method exponential, 0.1% deviations on a E/U M1 chart and have a look a few days back. Come back and tell me it doesnt do anything with the price. My idea is that there are a few hedgefunds that knows about this. How they use it exactly i dont know. Probably a bunch of "IF X then Y" in there.

Maybe you can do some calculations on that or something i dont know.

If you find that interesting i have some other stuff you might want to play around with. If you dont find that interesting, i will stop pestering you.
My signature is: "Classified".
  • #111
  • Oct 16, 2014 12:28pm Oct 16, 2014 12:28pm
Quoting yaed
My idea is that there are a few hedgefunds that knows about this.
Well then now all of them know about this because you just posted it on a public forum.

Those indicators are not worth a dime, no way a bunch of MT4 indicators to be the holy grail, they are not good, i am working with very advanced ones. Retail indicators are just plain rubbish, so i`m not even going to look at them.If it would be a holy grail do you think they would put it in the MT4 for free? I dont think so.
"There's a sucker born every minute" - P.T. Barnum
  • #112
  • Oct 16, 2014 1:30pm Oct 16, 2014 1:30pm
Quoting Proximus
{quote} Well then now all of them know about this because you just posted it on a public forum. Those indicators are not worth a dime, no way a bunch of MT4 indicators to be the holy grail, they are not good, i am working with very advanced ones. Retail indicators are just plain rubbish, so i`m not even going to look at them.If it would be a holy grail do you think they would put it in the MT4 for free? I dont think so.
Slow down cowboy! I didnt say anything about a holy grail did i? You go back to your very important calculations and i will stop bothering you. I thought you where the type of person actually interested in trying things.
My signature is: "Classified".
  • #113
  • Edited 2:02pm Oct 16, 2014 1:34pm | Edited 2:02pm
Quoting yaed
{quote} Slow down cowboy! I didnt say anything about a holy grail did i? You go back to your very important calculations and i will stop bothering you. I thought you where the type of person actually interested in trying things.
I am, but i already tried all those indicators, here is the thread:

http://www.forexfactory.com/showthread.php?t=487974

And they are not good, so I can safely say that these types of strategies wont work. The holy grail is not in the retail indicators, there are many more indicators which are not that public, try searching for those instead

EDIT: I am not saying that none of them are profiatable, because some of them could be, but just as a long term reliable strategy they are not, i bet that any system you find on this forum is not going to work for a very long time. As I see there are tops 3-4 systems on this forums which are 4-5 years old, the rest of them are rubbish, and will not work in long term, so trading after that is not a very reliable method.
"There's a sucker born every minute" - P.T. Barnum
  • #114
  • Edited 8:16am Oct 17, 2014 2:25am | Edited 8:16am
Quoting Proximus
{quote} I am, but i already tried all those indicators, here is the thread: http://www.forexfactory.com/showthread.php?t=487974 And they are not good, so I can safely say that these types of strategies wont work. The holy grail is not in the retail indicators, there are many more indicators which are not that public, try searching for those instead EDIT: I am not saying that none of them are profiatable, because some of them could be, but just as a long term reliable strategy they are not, i bet that any system you find on this forum...
Look, i didnt say anything about a "system", i didnt say anything about some juvenile idea about a "holy grail", i didnt say anything about long term profit. And no i havent found one good "system" here. Its like you think i have been talking about something i havent said a word about. You are talking to me now, not some 12 year old who just found out how to open up an account with some shitty broker and put a bunch of MACDs and stochastics and shit on a chart. Just read what i write and dont read what i didnt write.

I have made "systems" in the hundreds. Ive learned other peoples "systems" in the hundreds. "Holy grails" and "systems" are for kindergarden 12 year olds.

If you have a bunch of "advanced" indicators, lets have a look and maybe i can do something for you with my thinking. Im trying to do something constructive here. If you just want to talk about how "advanced" your indicators are, even though you know there are thousands of PHDs working hard to do the same kind of thing, which would mean you should understand your "advanced" stuff is kindergarden math to them. Im trying to think outside the box here. Im trying things. Im open to new ideas.

Quoting Proximus
I've build in the last 4 months a complex Artificial Inteligence which can trade upon indicators profitably, also it can check and compare the each indicator's validity.
Well then, either let me have a look or make that thing super secret and "advanced" if you want to, but then i guess its sort of hard to work together to make something happen. Ive decided to release pretty much all my info and its located here:

http://www.forexfactory.com/showthre...493882&page=10

I do this in the spirit of sharing. Maybe its usefull to someone, maybe its not. But the hedgefunds and banks knows all about that stuff already. But maybe there is something they have overlooked in the combination of discrete S/R and simple statistical models like the envelopes and such. Im not trying to create an EA right now, there are far to many variables. I use my mind to make decisions, i do not mechanically do the same thing over and over again because im CERTAIN that does not work. In other words i am CERTAIN it is not possible to make a low tech EA that is profitable in the long term under the circumstances given. I might be wrong, but this is the hypothesis im working with until proven otherwise.

This dude

http://www.forexfactory.com/hayseed

Seem to be doing something with EAs thats working, but i cant really confirm that is the case given the data i have. I CAN confirm what im doing is working, sort of in the long run. But what im doing is based in intelligent decisions, and sometimes stupid decisions, not mechanical repetition. 99% of the people on this forum are set in their thinking that a simple mechanical repeating process is possible to earn them infinite amounts of money in a relatively short term. The ASSUME this hypothesis on the basis of their own personal WISHES. I do not assume that hypothesis. I have not data to support such a hypothesis. The "systems" people talk about are such mechanical repeating processess. The EAs people talk about are such mechanical repeating processess. There is no difference between a "system" and a simple "EA", they are the same, only the first is operated by a boring person and the later is operated by a boring program. Lets take for instance the case of a "system" or "EA" that does not take S/R into account. It gives you a signal that aims the TP south of 1.25 in E/U. I would say that "System" or "EA" is worth less then a pile of horse shit.

What im saying is that "holy grail", "systems", simple "EAs" are all the same thing: repeating mechanical processess, and i have no reason to believe they "work". I do what i have reason to believe "work".

I hope with this post i have made myself absolutely clear. If you feel like it you can contribute to my thread or we can continue talking here, or just keep on going with your "advanced" things. Either way im happy and need not convince anybody of anything.

If you want to talk "statistics" i can show you my statement of about 60 trades in a row that was profits. Thats not even statistically possible. The chance of that happening ever is smaller then me being an alien from the andromeda galaxy with five dicks on my head and a pussy bigger then Kansas city. Thats because statistics alone wont tell you much of anything. The big picture is what matters

Like for instance:

Attached Image (click to enlarge)
Click to Enlarge Name: eu big picture.jpg Size: 299 KB


Attached Image (click to enlarge)
Click to Enlarge Name: september and october.jpg Size: 169 KB


Attached Image (click to enlarge)
Click to Enlarge Name: october.jpg Size: 180 KB


I mean that in statistics you might assume there is a 50/50 chance of being right/wrong, and "skill" is what offsets that chance. And maybe you want to determine "skill" by how much the trader is able to change this 50/50 chance of being right. Thats a VERY, VERY crude, not to say stupid statistical hypothesis. If you do that i have the skill of a god. But thats just stupid. What is important here is this: Either i am right about the big picture or i am not. All my trades is taken in the direction of the big picture. Maybe im wrong or maybe im right. That depends. But every individual trade should NOT, i repeat, should NOT be viewed as a separate statistical entry to compute ANYTHING with.

In other words you say: "Im trading D1, and either im selling or buying." From 25th of July until the 3rd of October this was VERY easy as you can see on these charts. And then on the 3rd of october we hit 1.25 and it havent been there ever since. Maybe it wont ever come back there, who knows. Im not placing any TPs south of that at least.

Now, all thats cleared away, i want to tell you why im talking to you Proximus. Because you seem to be something other then... well you know... an average person of average intellect. I might be wrong, this is just a hunch i get. I dont know you so well. Lets see where we go from here. The future is ours. We make it.
My signature is: "Classified".
  • #115
  • Oct 17, 2014 10:57am Oct 17, 2014 10:57am
Here is the first daily chart posted by yaed but seen by my filter. The filter computes bar by bar so it couldn't recognize the range in the second fourth of the period. Instead it looks more like a double top. Despite this detail the two charts look alike, aren't they?

Note: the indicator isn't smooth because it is computed on a 40 pip range bar chart and drawn on the daily. So it looks distorted.

Attached Image (click to enlarge)
Click to Enlarge Name: EURUSD-D1.png Size: 20 KB

Colors legend:
grey: neutral, sideway
orange: downward move
red: strong downward move
yelow: very strong downward move
green: upward move
light green: strong upward move
blue: very strong upward move (not present on this screenshot)

@yaed: very happy to see someone who makes the difference between timeframe and time scale.
No greed. No fear. Just maths.
  • #116
  • Oct 17, 2014 1:19pm Oct 17, 2014 1:19pm
Quoting PipMeUp
Here is the first daily chart posted by yaed but seen by my filter. The filter computes bar by bar so it couldn't recognize the range in the second fourth of the period. Instead it looks more like a double top. Despite this detail the two charts look alike, aren't they? Note: the indicator isn't smooth because it is computed on a 40 pip range bar chart and drawn on the daily. So it looks distorted. {image} Colors legend: grey: neutral, sideway orange: downward move red: strong downward move yelow: very strong downward move green: upward move light...
Well interesting. Makes things a bit easier. Way i see it, there is no timeframes, time is a continous flow. Its all a matter of perspective. I mean even if you trade tick scale, W1 still matters, even if you trade W1, tick scale is what CREATES W1
My signature is: "Classified".
  • #117
  • Oct 17, 2014 8:51pm Oct 17, 2014 8:51pm
Quoting yaed
{quote} Look, i didnt say anything about a "system", i didnt say anything about some juvenile idea about a "holy grail", i didnt say anything about long term profit. And no i havent found one good "system" here. Its like you think i have been talking about something i havent said a word about. You are talking to me now, not some 12 year old who just found out how to open up an account with some shitty broker and put a bunch of MACDs and stochastics and shit on a chart. Just read what i write and dont read what i didnt write. I have made "systems"...
No i`m scientifically analyzing my systems, that is the difference. Most people who make systems just test it 2-3 times and consider it profitable by granted, and scratch their heads why it is not working 2 weeks later.
Yes my systems may be kindengarten for somebody else, but atleast i`m trying, without trying you will never find it.
Also if you consider my advanced filter basic, then how do you consider your idea of using envelopes?

Quoting yaed
Well then, either let me have a look or make that thing super secret and "advanced" if you want to, but then i guess its sort of hard to work together to make something happen. Ive decided to release pretty much all my info and its located here: http://www.forexfactory.com/showthre...493882&page=10 I do this in the spirit of sharing. Maybe its usefull to someone, maybe its not. But the hedgefunds and banks knows all about that stuff already. But maybe there is something they have overlooked in the combination of discrete S/R and...
I`m happy if you think like that, but my research is my research, i will decide how much will i share with the public and how much i will keep it to myself, you just cant expect everyone to make their findings public, because that would destroy the competition, and FX is pretty much a competition game.
I am happy to share with people who also reciprocally help me, but i bet that there are 1000s of lazy guys wandering through the posts and reading them, without doing any work themselves or helping my work. Why should I share my work with those guys who contribute nothing to my success?

Quoting yaed
This dude http://www.forexfactory.com/hayseed Seem to be doing something with EAs thats working, but i cant really confirm that is the case given the data i have. I CAN confirm what im doing is working, sort of in the long run. But what im doing is based in intelligent decisions, and sometimes stupid decisions, not mechanical repetition. 99% of the people on this forum are set in their thinking that a simple mechanical repeating process is possible to earn them infinite amounts of money in a relatively short term. ...
I dont understand where you are going with this? Statistics are a good tool and useful when dealing with numbers.
I`m also aiming at a mechanized model, however the difference here is that I have an adaptive model not a rigid static model.
"There's a sucker born every minute" - P.T. Barnum
  • #118
  • Oct 18, 2014 1:18am Oct 18, 2014 1:18am
Quoting Proximus
{quote} No i`m scientifically analyzing my systems, that is the difference. Most people who make systems just test it 2-3 times and consider it profitable by granted, and scratch their heads why it is not working 2 weeks later. Yes my systems may be kindengarten for somebody else, but atleast i`m trying, without trying you will never find it. Also if you consider my advanced filter basic, then how do you consider your idea of using envelopes? {quote} I`m happy if you think...
Im not saying anything about your filter, i dont know what it is or what it does.

I agree that we should probably not share everything we find on a public forum since there are people who dont deserve getting it. I agree hwolheartedly with that. I change my mind every now and then and share or not share everything, it all depends on my mood. For an example, yesterday i thought of something that i dont want to share. Not right now at least. Im not sure either where i was going with that stuff i was saying about statistics, i was simply explaining my view on it applied in trading.
My signature is: "Classified".
  • #119
  • Oct 18, 2014 12:25pm Oct 18, 2014 12:25pm
Quoting yaed
{quote} Im not saying anything about your filter, i dont know what it is or what it does. I agree that we should probably not share everything we find on a public forum since there are people who dont deserve getting it. I agree hwolheartedly with that. I change my mind every now and then and share or not share everything, it all depends on my mood. For an example, yesterday i thought of something that i dont want to share. Not right now at least. Im not sure either where i was going with that stuff i was saying about statistics, i was simply explaining...
Ok then we have similar viewpoints.

My filter, I explained it in earlier posts, its like a moving average used to generate BUY & SELL signals. If the price is above my filter, then its a BUY, if it's below it's sell. I dont use SL and TP, yet, until i dont determine the maximum DD. After the Max DD is determined then the SL = Max DD, and the LOT size will be set accordingly to a 3-4% account risk. Until this the BUY/SELL signals's accuracy will be measured, and I`m looking for very small DD.

The overall metric that i use is called PROXI SCORE, which I invented it, you can find the description here:
http://www.forexfactory.com/showthre...87#post7797887

The highest the PROXI SCORE, the more accurate is the filter.

But here is the catch, my filter is not like basic moving averages, my filter is totally adaptive, and adapts to the price overall, my goal is to adapt it to 13 years of historical data, and by this I expect it to work for another 10 years atleast , or until the next economic crisis. The market's shape is not really changing, there is only a big trending phase 2001-2008, and a ranging phase 2008-2014, so if I already adapt my filter to both trending and ranging phases, i expect another trending phase in the near future if the ECB is messing up the EURO, so my filter can adapt to this with no problem.
"There's a sucker born every minute" - P.T. Barnum
  • #120
  • Oct 18, 2014 12:36pm Oct 18, 2014 12:36pm
My Goals in terms of PROXI SCORE is about a PROXI SCORE over 50,000. That would qualify a HOLY GRAIL.

Currently I got one of 655. So my progress is 655/50000 = 1.31%

My research is at 1.31% of completion. Pretty slow research, but even a 25% completion would already give me a tradeable system So no need to hunt for the 50.000 result.
"There's a sucker born every minute" - P.T. Barnum
0 traders viewing now