Advertisement
Humans vs. Robots
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 12, 2013 at 05:53
It's the golden ratio, sir.
Dec 12, 2013 at 21:54
ahuruglica posted:
I would like to have healthy discussion when manual and automatic traders share their experience.
Thanks
Ardit H.
manual is like steam engine. worked well in the 1800's - 1900's. could go across the country fast.
automated system is like. plasma rocket. you can go across country fast, or go to Mars.
manual can't go to Mars.
Dec 12, 2013 at 21:55
The Gleason report, was a study (ad hock) revealing a mechanical moving average strategy which tested successful from 1963 to 1983, was less successful or not at all, from 1983 up to 2003.
This report though not scientific by any stretch, demonstrates the evolution of price behavior over time, and the likely effect technology has in that evolution. The ability of automated systems to read the market and therefore trade it, change the market. The more systems there are, the more change occurs.
Quantum mechanics famous "double slit" test, where merely observing an event, changed the results of the event, are another example of this principle. Automated trading systems do much more than observe the market, and so all the more change it.
This cause/effect of automated systems on the market, prove that for a system to remain successful over time, it must be adaptive, to capitalize on price behavior evolution, which would occur on all time scales.
This report though not scientific by any stretch, demonstrates the evolution of price behavior over time, and the likely effect technology has in that evolution. The ability of automated systems to read the market and therefore trade it, change the market. The more systems there are, the more change occurs.
Quantum mechanics famous "double slit" test, where merely observing an event, changed the results of the event, are another example of this principle. Automated trading systems do much more than observe the market, and so all the more change it.
This cause/effect of automated systems on the market, prove that for a system to remain successful over time, it must be adaptive, to capitalize on price behavior evolution, which would occur on all time scales.
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 12, 2013 at 21:56
I remember reading the book The Intelligent Investor by Benjamin Graham. On the last chapter he describes the market as a real person. He also names him as Mr. Market. That chapter always stood on my mind. He describes Mr. Market as a real person having real human behavior. Sometimes he is emotional, moody, euphoric and often irrational. He states that Mr. Market is to serve us, not guide us.
We humans have feeling, and sometimes feeling get in our way. But as humans we also have friends and family and we can understand their feelings too. That is why we can adapt faster to market changes.
In my opinion good EAs are not designed to replace human minds. They are just a tool to speed up the process and perform mechanized tasks like any machine or computing equipment does.
Of course, this is just my humble opinion.
We humans have feeling, and sometimes feeling get in our way. But as humans we also have friends and family and we can understand their feelings too. That is why we can adapt faster to market changes.
In my opinion good EAs are not designed to replace human minds. They are just a tool to speed up the process and perform mechanized tasks like any machine or computing equipment does.
Of course, this is just my humble opinion.
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 13, 2013 at 16:17
Imagine Switzerland fixing the market.
forex_trader_55569
Dec 04, 2011からメンバー
30 投稿
Dec 13, 2013 at 21:48
dgaf posted:
Robots follow rules & trade with no emotions.
Yes, that is our main point as wel. Our system is an EA with manual oversight (same rules) - http://www.myfxbook.com/members/pipinvestments/pip-investor-20-1-year/605364
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 13, 2013 at 21:54
@pipinvestments
Sorry pal that is considered spam.
Anyway, the account is demo, make te same result on real account and don't use custom start date. Then you are free to come back.
Sorry pal that is considered spam.
Anyway, the account is demo, make te same result on real account and don't use custom start date. Then you are free to come back.
Dec 14, 2013 at 11:32
Manual traders and automated systems most often Both use indicators. Indicators are a form of automation, and typically part of automated trading model logic, and manual traders technical trading plan.
The most common indicator could be the Moving Average. Moving average =Moving Mean.
It is widely acknowledged by statisticians that Mean can be distorted by outliers in the sample. Whereas Median gives a more robust measure of price than Mean.
http://www.bvmarketdata.com/pdf/Median-Mean.pdf
Most charting platforms are filled with Moving Average choices, but not often Moving Medians.
The most sought after quality in an automated trading strategy is ROBUSTNESS. It increases the odds for good OOS results, and helps guard against time erosion. Manual traders also pursue a robust styles of trading.
My question is this: If Robustness is the goal of a successful trading plan (automated or manual), then why isn't the use of the more Robust indicator (Moving Median) more prevalent than Moving Average (Moving Mean)?
The most common indicator could be the Moving Average. Moving average =Moving Mean.
It is widely acknowledged by statisticians that Mean can be distorted by outliers in the sample. Whereas Median gives a more robust measure of price than Mean.
http://www.bvmarketdata.com/pdf/Median-Mean.pdf
Most charting platforms are filled with Moving Average choices, but not often Moving Medians.
The most sought after quality in an automated trading strategy is ROBUSTNESS. It increases the odds for good OOS results, and helps guard against time erosion. Manual traders also pursue a robust styles of trading.
My question is this: If Robustness is the goal of a successful trading plan (automated or manual), then why isn't the use of the more Robust indicator (Moving Median) more prevalent than Moving Average (Moving Mean)?
May 04, 2012からメンバー
1534 投稿
Dec 14, 2013 at 12:00
VPS: $29.99/month
Robot: $500
Cornering the market in your sleep.... Priceless.
Please click "Vouch" if you liked my post. If not, just put me on your Blocked list. :o)
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 14, 2013 at 14:11
@FxMasterGuru
Corner the market yourself. I am blocking you here cause you SPAM.
This is a discussion page not you billboard.
Corner the market yourself. I am blocking you here cause you SPAM.
This is a discussion page not you billboard.
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 14, 2013 at 22:08
Do you mean RSI indicator?
Dec 15, 2013 at 09:29
FxMasterGuru posted:
V
Robot: $500
Cornering the.. Price...
I hope your coding skills are better than your marketing skills.. Cause they sort of suck.
Turning an potential stranger into a prospect, and then into a paying client involves a lot more than spam,
In fact it doesn't involve spam at all.
Dec 15, 2013 at 09:46
Actually forget my post above. I am still learning. @MoneyCode has helped me realize something new.
Sleep is for the weak
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 15, 2013 at 10:18
Feb 04, 2012からメンバー
80 投稿
Dec 15, 2013 at 10:38
FxMasterGuru posted:
VPS: $29.99/month
Robot: $500
Cornering the market in your sleep.... Priceless.
It is me again FxMasterGuru, and I was just making a JOKE based on MasterCard's famous advertisement campaign: http://www.mastercard.com/mel/personal/en/promotions/mastercard_tv_commercials.html
Come on, guys! IT WAS A JOKE! Where is your sense of humor...?? Forex trading does not have to be always so serious. At least not on a weekend when the markets are closed. Relax!!
forex_trader_136673
Jun 28, 2013からメンバー
842 投稿
Dec 15, 2013 at 11:44
Do you like to watch commercials during movies? I bet not.
Dec 15, 2013 at 22:07
Actually I have done quite a bit of testing using stochastic oscillator. Testing with RSI in comparison with Stochastic- I found stochastic more cyclical than RSI, so I never invested much time testing RSI. Though I did read a report on a successful system that used a 200 MA, in combination with a short term RSI pull back for entries.
With stochastics, I did such things as:
Establish threshold levels, and count the number of bars contained in that level, this helped provide a quantifiable exhaustion factor. For instance, how many bars has the stochastic 50 period been riding above 70 level. I would combine different periods of Stochastics on one chart to reflect the most commonly used chart intervals, such as 60 minute chart 50 period, 15 minute chart 50 period, 5 minute chart 50 period, and convert each one to a one minute chart stochastic indicator. I've also created averages of Stochastics and averages of the threshold counters. I've tested many trading rules with stochastic based signals as primary indicators.
It taught me how the different time frame stochastic readings influence each other. It revealed the surprising exhaustion readings and helped establish for me how useful they could or couldn't be. It established averages, of which are the most common readings to occur and also the outer limits of the readings. I've also read empirical studies done with stochastic which I could confirm from my own experience were true.
They are useful and can be profitable. The problem as with any system is that to have a robust trading plan, the "degrees of freedom" must be considered. With each indicator you add or rule you add, the less statistically significant your results become in relation to your data set size (unless you enlarge your data set to compensate for each time you add more rules). For this reason I try to limit my code, parameters and indicators to be a small as possible. Otherwise all my work would be wasted, (and that is a not a good feeling, at all..) because the over fit would certainly deteriorate the OOS results and it would likely fail in "live" trading. So I've cut back on multiple indicators in the interest of simplicity.
I would rather put my money into a safe system, that may not shoot directly to the moon. Rather than put my money in a rocket ship that has a chance to blow up on the launch pad. Though I believe the 2007 or 2008 trading championship was one by a system that used 19 parameters, if I'm not mistaken.
With stochastics, I did such things as:
Establish threshold levels, and count the number of bars contained in that level, this helped provide a quantifiable exhaustion factor. For instance, how many bars has the stochastic 50 period been riding above 70 level. I would combine different periods of Stochastics on one chart to reflect the most commonly used chart intervals, such as 60 minute chart 50 period, 15 minute chart 50 period, 5 minute chart 50 period, and convert each one to a one minute chart stochastic indicator. I've also created averages of Stochastics and averages of the threshold counters. I've tested many trading rules with stochastic based signals as primary indicators.
It taught me how the different time frame stochastic readings influence each other. It revealed the surprising exhaustion readings and helped establish for me how useful they could or couldn't be. It established averages, of which are the most common readings to occur and also the outer limits of the readings. I've also read empirical studies done with stochastic which I could confirm from my own experience were true.
They are useful and can be profitable. The problem as with any system is that to have a robust trading plan, the "degrees of freedom" must be considered. With each indicator you add or rule you add, the less statistically significant your results become in relation to your data set size (unless you enlarge your data set to compensate for each time you add more rules). For this reason I try to limit my code, parameters and indicators to be a small as possible. Otherwise all my work would be wasted, (and that is a not a good feeling, at all..) because the over fit would certainly deteriorate the OOS results and it would likely fail in "live" trading. So I've cut back on multiple indicators in the interest of simplicity.
I would rather put my money into a safe system, that may not shoot directly to the moon. Rather than put my money in a rocket ship that has a chance to blow up on the launch pad. Though I believe the 2007 or 2008 trading championship was one by a system that used 19 parameters, if I'm not mistaken.
