Another good post from PAL blog about mean reversion strategies. I backtested all three in SPY and returns are high. So my question is: Why isn't everyone rich trading these simple systems? I guess this is a stupid question because not everyone can be rich but while tarders complain in America that markets are rough some very simple systems are making good money.
Why isn't everyone rich trading these systems?
- Thread starter lftrader
- Start date
-
- Tags
- rich trading trading systems
- Joined
- 25 February 2011
- Posts
- 5,685
- Reactions
- 1,218
Take a closer look at the comments on that blog regarding the decline in recent performance of the more popular of the three systems. I believe it provides some insight into one of the unavoidable vulnerabilities of trading systems.
- Joined
- 13 June 2007
- Posts
- 838
- Reactions
- 136
Greetings --
I researched many trading systems, written about them, and posted videos on YouTube.
All systems go through periods where profits add to account equity and losses subtract from it. Most traders agree that the limitation to continuing to trade a system is drawdown in excess of the trader's risk tolerance. The key to maintaining profitability over a long period of time is being able to recognize when the model portion of the system and the data it is processing are in sync and when they have fallen out of sync, then adjusting position size on a trade-by-trade basis in response to recent performance.
The videos will help clarify.
http://www.blueowlpress.com/video-presentations
Begin with "The Importance of Being Stationary", then "The Four Faces of Risk."
One of the metrics I have developed is "CAR25." It is the estimated compound annual rate of return for a trading system when the trade sizes are risk-normalized so that the drawdown does not exceed the trader's tolerance. CAR25 can be used as a metric to compare any uses of funds, including trading.
When various trading techniques are evaluated using risk-normalization, systems that trade frequently, trade accurately, hold a short period, and avoid large losing trades are superior. While all trades are trend-following for the period the trade is held, the entry technique that most often fits the criteria mentioned above is mean reversion, such as RSI2. (See my book "Mean Reversion Trading Systems" for a discussion of the RSI indicator and methods that relax the requirement that the lookback period be an integer.)
The key to staying profitable is continuing to monitor the recent trades, reducing position size when in a drawdown and increasing position size when the system is working well.
Thanks for listening,
Howard
I researched many trading systems, written about them, and posted videos on YouTube.
All systems go through periods where profits add to account equity and losses subtract from it. Most traders agree that the limitation to continuing to trade a system is drawdown in excess of the trader's risk tolerance. The key to maintaining profitability over a long period of time is being able to recognize when the model portion of the system and the data it is processing are in sync and when they have fallen out of sync, then adjusting position size on a trade-by-trade basis in response to recent performance.
The videos will help clarify.
http://www.blueowlpress.com/video-presentations
Begin with "The Importance of Being Stationary", then "The Four Faces of Risk."
One of the metrics I have developed is "CAR25." It is the estimated compound annual rate of return for a trading system when the trade sizes are risk-normalized so that the drawdown does not exceed the trader's tolerance. CAR25 can be used as a metric to compare any uses of funds, including trading.
When various trading techniques are evaluated using risk-normalization, systems that trade frequently, trade accurately, hold a short period, and avoid large losing trades are superior. While all trades are trend-following for the period the trade is held, the entry technique that most often fits the criteria mentioned above is mean reversion, such as RSI2. (See my book "Mean Reversion Trading Systems" for a discussion of the RSI indicator and methods that relax the requirement that the lookback period be an integer.)
The key to staying profitable is continuing to monitor the recent trades, reducing position size when in a drawdown and increasing position size when the system is working well.
Thanks for listening,
Howard
Wysiwyg
Everyone wants money
- Joined
- 8 August 2006
- Posts
- 8,428
- Reactions
- 284
Let me analyse this comment by the author ....
1. "performance of this system was basically flat in 2014" = red flag
2. "this could mean that its potential gains are being arbitraged out" = say anything but sound professional to fool the Neville No Ideas
3. "In comparison, the win rate (2) system returned 10.2% in 2014" = anything above a flat year performance is obviously better
4. "I expect both systems to become unprofitable soon if serial correlation returns to the markets" = for the apparent non serial correlated systems - sound professional but state the obvious. I expect it to rain if water falls from the sky
Ulterior motives could be to lure more unsuspecting dreamers into the market. Baiting the hooks for the real professionals to reel in profits from.
If we wanted to be rich we would behave exactly like Buffett or Soros or Radge or Tech/A wouldn't we? There is much more to the game than that. It is called game time and as Howard eluded to you have to know when to trim the sails or go motor. Mechanically, I find that difficult because changes in range over time are not absolute in tradable issues.
Maybe I am an under resourced wanna be who hasn't found the magic yet.
1. "performance of this system was basically flat in 2014" = red flag
2. "this could mean that its potential gains are being arbitraged out" = say anything but sound professional to fool the Neville No Ideas
3. "In comparison, the win rate (2) system returned 10.2% in 2014" = anything above a flat year performance is obviously better
4. "I expect both systems to become unprofitable soon if serial correlation returns to the markets" = for the apparent non serial correlated systems - sound professional but state the obvious. I expect it to rain if water falls from the sky
Ulterior motives could be to lure more unsuspecting dreamers into the market. Baiting the hooks for the real professionals to reel in profits from.
If we wanted to be rich we would behave exactly like Buffett or Soros or Radge or Tech/A wouldn't we? There is much more to the game than that. It is called game time and as Howard eluded to you have to know when to trim the sails or go motor. Mechanically, I find that difficult because changes in range over time are not absolute in tradable issues.
Maybe I am an under resourced wanna be who hasn't found the magic yet.
- Joined
- 19 October 2013
- Posts
- 50
- Reactions
- 0
I don't think anyone can answer your question (unless they have a crystal ball) but here are a few observations:
- The learning curve for understanding and developing trading systems is high for the average investor. Average investors would probably be more inclined towards buying a black box product or following other people's advice (i.e. stock picks) than putting in the hard yards to learn.
- Those who are getting rich would most likely not share their work/ideas for fear of losing their edge. This will give everyone a perception that perhaps these systems do not work hence cannot get rich from trading them.
I have also had a few interesting experiences in trying to teach fellow investors about systems trading etc. Initially they are very interested and amazed at what it can achieve - particularly if they have a basic understanding of statistics and probability. However, I find that they all end up reverting back to whatever method they had developed themselves, i.e. stock picking on hot copper, fundamental analysis etc. I think this has more to do with psychology of people gravitating towards things they feel more comfortable with than anything else.
- Joined
- 13 June 2007
- Posts
- 838
- Reactions
- 136
Greetings again --
The issue really is stationarity -- and dynamic position sizing in response to changes in stationarity.
The report quoted -- along with almost all reports of trading system tests -- assume that:
1. The result posted is "the" result. It is not.
The purpose of a trading system's model is to recognize signals that precede profitable trades in the data it is processing. It is important that the distribution of signal-profit pairs remain stationary over the period of test and following period of trading. Lost of stationarity results in loss of profits. Given any distribution of trades, the Best we can hope for is that the distribution is stationary. If it is, then we can expect that the future trades will resemble the trades discovered during development. We Cannot assume that the same trades will occur in the same sequence in the future. Consequently, we cannot rely on metrics, such as maximum system drawdown, that are sequence-dependent.
2. The results are typical. They are not.
We do not stop adjusting systems until the results are good. So good that they are well above the mean of the distribution of results.
3. The distribution is stationary. It is not.
Determining the period of stationarity is one of the important tasks of the system developer. It determines the combined length of the in-sample and out-of-sample periods. If a system is stationary for, say, four years, then using walk-forward IS and OOS periods of four and two years just about guarantees that the OOS period does not resemble the IS period and the results will not be similar.
4. That genuine learning took place during model development. It may not have.
The model will always fit the in-sample data. We guarantee that by expanding our search for something profitable until it does. Whether the model learned to recognize a real and persistent profit-anticipating signal or was just fit to the random noise in the data can only be determined by a separate validation process. That is, testing on data that is both more recent than the data used to set the rules and determine the parameters, and that has not been used or tested before.
5. The position size is fixed or is determined within the trading system model.
It should not be fixed. It should vary as the performance varies.
It should not be determined within the trading system model. It should be determined within the trading system Management model.
This gives the trader an opportunity to tune position size as performance changes, and to recognize the onset of system failure in time to reduce position size to protect account.
-------------------
To beat the apparently not dead horse one more time, I recommend watching the two video presentations:
The Importance of Being Stationary
https://www.youtube.com/watch?v=iBhrZKErJ6A&feature=youtu.be
The Four Faces of Risk
https://www.youtube.com/watch?v=Vw7mseQ_Tmc&feature=youtu.be
---------------------
My view of individuals managing their accounts by mimicking the investing / trading techniques of Buffett and others who hold long periods is that that is way too risky. At least for me. Buffett had a drawdown of over 50% in 2009. Government bailouts, which have yet to be unwound, gave Buffett his account back in a few years. If 2009 had turned into 1929, it would have taken 30 years to get back to even. If the issue held was Enron, there never was a recovery.
Unless the person managing the funds is sitting inside the boardroom of the company being traded, there is no information that is not present in the list of trades and their prices. There is no way to distinguish a small correction from the beginning of a wealth-destroying crash. Given that, the rational management procedure is to reduce exposure in response to losses. I want to be flat, sitting on the sidelines, with most of my account intact, analyzing when to return to trading through monitoring of recent shadow trade performance -- rather than in a steep drawdown hoping for a recovery or bailout before I need to sell something to liquidate funds for some other purpose.
While my systems work focuses on short holding periods, it is primarily because the results of applying those methods can be demonstrated. That said, I do have a macro view of the financial world ahead, and it is not optimistic. The Western world experienced an exceptional advantage over the period following World War II up to about 2000. That has given buy-and-hold and long-holding-period techniques a favorable tailwind. I doubt that will continue. Whether I am right or wrong, I expect systems that trade frequently, trade accurately, hold a short period, and avoid serious losing trades to out perform systems that hold longer.
That said, if some technique other than I one I recommend works for you, then ignore me and continue on.
-----------------
Thanks for listening,
Howard
The issue really is stationarity -- and dynamic position sizing in response to changes in stationarity.
The report quoted -- along with almost all reports of trading system tests -- assume that:
1. The result posted is "the" result. It is not.
The purpose of a trading system's model is to recognize signals that precede profitable trades in the data it is processing. It is important that the distribution of signal-profit pairs remain stationary over the period of test and following period of trading. Lost of stationarity results in loss of profits. Given any distribution of trades, the Best we can hope for is that the distribution is stationary. If it is, then we can expect that the future trades will resemble the trades discovered during development. We Cannot assume that the same trades will occur in the same sequence in the future. Consequently, we cannot rely on metrics, such as maximum system drawdown, that are sequence-dependent.
2. The results are typical. They are not.
We do not stop adjusting systems until the results are good. So good that they are well above the mean of the distribution of results.
3. The distribution is stationary. It is not.
Determining the period of stationarity is one of the important tasks of the system developer. It determines the combined length of the in-sample and out-of-sample periods. If a system is stationary for, say, four years, then using walk-forward IS and OOS periods of four and two years just about guarantees that the OOS period does not resemble the IS period and the results will not be similar.
4. That genuine learning took place during model development. It may not have.
The model will always fit the in-sample data. We guarantee that by expanding our search for something profitable until it does. Whether the model learned to recognize a real and persistent profit-anticipating signal or was just fit to the random noise in the data can only be determined by a separate validation process. That is, testing on data that is both more recent than the data used to set the rules and determine the parameters, and that has not been used or tested before.
5. The position size is fixed or is determined within the trading system model.
It should not be fixed. It should vary as the performance varies.
It should not be determined within the trading system model. It should be determined within the trading system Management model.
This gives the trader an opportunity to tune position size as performance changes, and to recognize the onset of system failure in time to reduce position size to protect account.
-------------------
To beat the apparently not dead horse one more time, I recommend watching the two video presentations:
The Importance of Being Stationary
https://www.youtube.com/watch?v=iBhrZKErJ6A&feature=youtu.be
The Four Faces of Risk
https://www.youtube.com/watch?v=Vw7mseQ_Tmc&feature=youtu.be
---------------------
My view of individuals managing their accounts by mimicking the investing / trading techniques of Buffett and others who hold long periods is that that is way too risky. At least for me. Buffett had a drawdown of over 50% in 2009. Government bailouts, which have yet to be unwound, gave Buffett his account back in a few years. If 2009 had turned into 1929, it would have taken 30 years to get back to even. If the issue held was Enron, there never was a recovery.
Unless the person managing the funds is sitting inside the boardroom of the company being traded, there is no information that is not present in the list of trades and their prices. There is no way to distinguish a small correction from the beginning of a wealth-destroying crash. Given that, the rational management procedure is to reduce exposure in response to losses. I want to be flat, sitting on the sidelines, with most of my account intact, analyzing when to return to trading through monitoring of recent shadow trade performance -- rather than in a steep drawdown hoping for a recovery or bailout before I need to sell something to liquidate funds for some other purpose.
While my systems work focuses on short holding periods, it is primarily because the results of applying those methods can be demonstrated. That said, I do have a macro view of the financial world ahead, and it is not optimistic. The Western world experienced an exceptional advantage over the period following World War II up to about 2000. That has given buy-and-hold and long-holding-period techniques a favorable tailwind. I doubt that will continue. Whether I am right or wrong, I expect systems that trade frequently, trade accurately, hold a short period, and avoid serious losing trades to out perform systems that hold longer.
That said, if some technique other than I one I recommend works for you, then ignore me and continue on.
-----------------
Thanks for listening,
Howard
CanOz
Home runs feel good, but base hits pay bills!
- Joined
- 11 July 2006
- Posts
- 11,543
- Reactions
- 519
Thanks for stopping by Howard, great post.
- Joined
- 25 February 2011
- Posts
- 5,685
- Reactions
- 1,218
Yes thanks again, Howard. As usual, I cannot help but be impressed by the depth and thoroughness of your analyses of trading system model development.
- Joined
- 25 February 2011
- Posts
- 5,685
- Reactions
- 1,218
Howard Bandy's offerings are hard to surpass. When watching his presentations and reading his materials, one quickly comes to the realisation that there is a lot more to performance measurement than the arbitrary examination of a few sample periods.
Wysiwyg
Everyone wants money
- Joined
- 8 August 2006
- Posts
- 8,428
- Reactions
- 284
Could you supply a link to real trading account results where have they done so well?
tech/a
No Ordinary Duck
- Joined
- 14 October 2004
- Posts
- 20,403
- Reactions
- 6,326
Radge has 4 or 5 going and you'll need to be a member to view.
There are a few more I know of and will post when I look them up at home.
I think there are a few things holding back Joe public/
(1) Knowledge--We aren't Howard Bandy
(2) Understanding what we should be looking for.
(2a)How we do that
(2b)What to do when we find it.
(2c)How to evaluate that correctly.
(2d)How to implement and monitor.
Howard has completed volume of work I've chewed through most.
They are fantastic books.
While I'm sure all the answers are there I need to have a 12 mth
Howard Bandy course to gain a competency in what most is advanced
systems testing.
You can only do a certain amount from a book/s---with the knowledge you have.
Howards or anyone else's!
There are a few who are educated enough in enough fields to really benefit,
But for the majority---
Having said that I know you don't have to be a Howard Bandy to succeed.
Sure would be good though!
- Joined
- 13 September 2013
- Posts
- 988
- Reactions
- 530
Complexity brings with it different issues i believe, how often have you heard the KISS principle attached to a number of things in life? Why should trading be different?
People like to make systems trading sound harder than it needs to be, mind you i am not a multi millionaire. A lot of the top systems traders talk about simple being best
If a system has a large numbr of inputs or indicators there is a greater chance of one of these breaking down than a simple ROBUST system with fewer inputs.
CanOz
Home runs feel good, but base hits pay bills!
- Joined
- 11 July 2006
- Posts
- 11,543
- Reactions
- 519
I think the other thing that allot fail to realize is that:
1.) it takes a portfolio of systems to usually make consistent returns, without consistency the comfort level disappears and the pain sets in, stopping the trader from continuing to execute the strategies as they were designed....
2.) allot of people don't have the capital to apply to a group of trading systems.
3.) allot of developers sell or lease the systems to further augment cash flow. Some developers are selling optimized crap that can't stand up to any test of time, the sales or leases are there only source of real income.
This guy here certainly gives a trader something to think about before trading an algorithm, Dave Walton.
Better System Trader
1.) it takes a portfolio of systems to usually make consistent returns, without consistency the comfort level disappears and the pain sets in, stopping the trader from continuing to execute the strategies as they were designed....
2.) allot of people don't have the capital to apply to a group of trading systems.
3.) allot of developers sell or lease the systems to further augment cash flow. Some developers are selling optimized crap that can't stand up to any test of time, the sales or leases are there only source of real income.
This guy here certainly gives a trader something to think about before trading an algorithm, Dave Walton.
Better System Trader
probably because of this (referring to the IBS indicator referenced in the link)...
"The results show that the IBS technical indicator is a strong and consistent predictor of closeto-
close returns for equity ETFs trading in the U.S. Equity ETFs tend to mean revert on daily
timescales; when the closing price is in the top of the day’s range, close-to-close returns are
lower, and vice versa. These results are generally not present in overseas markets, however."
(source: http://tinyurl.com/j48nqqp)
not all systems work in all market at all times. A system that works for someone does not necessarily transfer easily elsewhere. the system at some point in time might start to fail. and that is why people tend to develop new systems to handle the market they trade in. interesting to note also the paper the quote is from also raises other issues like "catching a falling knife" (pg 7).
"The results show that the IBS technical indicator is a strong and consistent predictor of closeto-
close returns for equity ETFs trading in the U.S. Equity ETFs tend to mean revert on daily
timescales; when the closing price is in the top of the day’s range, close-to-close returns are
lower, and vice versa. These results are generally not present in overseas markets, however."
(source: http://tinyurl.com/j48nqqp)
not all systems work in all market at all times. A system that works for someone does not necessarily transfer easily elsewhere. the system at some point in time might start to fail. and that is why people tend to develop new systems to handle the market they trade in. interesting to note also the paper the quote is from also raises other issues like "catching a falling knife" (pg 7).
- Joined
- 30 June 2007
- Posts
- 7,200
- Reactions
- 1,225
OP, this is trading.....we're talking about free money. It's obvious with such a situation that you'd have to rank above the 99th percentile to live off your proceeds, pay all your expenses, mortgage, children, holidays, superannuation. I know only one person who does that here, skc. There's probably others, but this should give you an idea what you're up against.
OTOH, making a part time living is a bit easier. A few here would be in that basket.
OTOH, making a part time living is a bit easier. A few here would be in that basket.
- Joined
- 13 September 2013
- Posts
- 988
- Reactions
- 530
Just because a system was flat one year doesn't mean it's broken or will never work. I would say a fair chunk of pro traders with 20+ years record would have had losing years but still traded th same and stuck with their systems. A lot probably lost cash in 08 or in this sideways market over the last few years doesn't mean their edge is gone
- Joined
- 30 June 2007
- Posts
- 7,200
- Reactions
- 1,225
Pro traders can't have losing years. To be professional means that even losing months must be rare. Who or what is going to pay for living expenses?
- Joined
- 13 September 2013
- Posts
- 988
- Reactions
- 530
People like Bill Dunn, Jerry Parker have had multiple losing years. They are pro traders.
Im sure a retail trader that trades for a living doesn't spend every cent they earn in the good times, they have to be prepared for periods when their system or style isn't making money over longer periods and put money aside
CanOz
Home runs feel good, but base hits pay bills!
- Joined
- 11 July 2006
- Posts
- 11,543
- Reactions
- 519
There is a difference between 'fund manager' and 'trader'....it usually means a fund manager gets a salary and a trader gets a P/L....
Similar threads
