![Question :?:](./images/smilies/icon_question.gif)
Dos and Donts of Drawdowns
Dos and Donts of Drawdowns
Risk and Reward will forever be linked to each other's hips in trading. Since a big part of trading success entails managing one's emotions through a drawdown, i was hoping we can learn from other members' experiences, rules of thumb and mental preparation to possibly form some dos and donts of managing drawdowns. Any takers ![Question :?:](./images/smilies/icon_question.gif)
![Question :?:](./images/smilies/icon_question.gif)
-
- Roundtable Fellow
- Posts: 58
- Joined: Wed Apr 30, 2003 4:35 pm
- Location: Netherlands
These suggestions are from the perspective of the long term trend follower trading in leveraged financial instruments (e.g., futures).
The most helpful advice is to understand your personal risk preference/ avoidance profile. This requires experience and a strong capacity for truthful self-examination. If you are inaccurate in your assesment of your capacity to withstand risk, your own best efforts at trading will be unable to prevent self-sabotage.
Next, I think it is useful to develop faith in your method. Testing can help this endeavor. Run simulations over numerous time frames and assess "holding period returns" (a bit of a misnomer, here). For instance, knowing that in any given twelve month period (i.e., from any particular start date to a date twelve months later) over the past thirty years of price data in your simulations you have shown a theoretical profit can ease one's mind and build faith that draw downs are a temporary, even necessary aspect of long term trend following. Looking at the data in 3, 6 and eighteen month horizons will provide you with more information which, in turn, may help you push through the draw downs.
Finally, I believe it is good to permit the mind to engage other interests. I like to write and sail. Reading and movies provide wonderful diversions. I do not advocate escapism. Instead, I think that success comes easiest to well rounded individuals. People with some other things going on are better able to process stressful circumstances. Perspective is gained. Draw downs are thus better understood and better dealt with.
Fair warning: if you are not sure of the first item above, the other two can be of little - or no - help to you.
Ken
The most helpful advice is to understand your personal risk preference/ avoidance profile. This requires experience and a strong capacity for truthful self-examination. If you are inaccurate in your assesment of your capacity to withstand risk, your own best efforts at trading will be unable to prevent self-sabotage.
Next, I think it is useful to develop faith in your method. Testing can help this endeavor. Run simulations over numerous time frames and assess "holding period returns" (a bit of a misnomer, here). For instance, knowing that in any given twelve month period (i.e., from any particular start date to a date twelve months later) over the past thirty years of price data in your simulations you have shown a theoretical profit can ease one's mind and build faith that draw downs are a temporary, even necessary aspect of long term trend following. Looking at the data in 3, 6 and eighteen month horizons will provide you with more information which, in turn, may help you push through the draw downs.
Finally, I believe it is good to permit the mind to engage other interests. I like to write and sail. Reading and movies provide wonderful diversions. I do not advocate escapism. Instead, I think that success comes easiest to well rounded individuals. People with some other things going on are better able to process stressful circumstances. Perspective is gained. Draw downs are thus better understood and better dealt with.
Fair warning: if you are not sure of the first item above, the other two can be of little - or no - help to you.
Ken
William-
I think you just received some very sound advice.
I would add the perspective that you tend to view your trading in, will make the task of trading easier or harder to bare and that tends to be a good indication of your trading being a profitable or losing venture.
Think of the mindset of a golfer. He is leading a tournament by a stroke and he has a 3 foot putt on the 18th for the win. As he stands over the putt, his mind can be thinking something along the lines of:
Trading is the same, once you have a real system and the environment setup to trade, your job is to simply give your systems the best chance to succeed. This comes down to management of oneself.
Once you know what you are dealing with, deal with it. I have found focusing on performance of trading in terms of equity might not be the "best" way, as your emotions might ride the equity curve. This might make the drawdowns an issue. Instead, focus on the performance of trading as a function of the management of the system. Did you place all the trades? Did you hesitate? If you are unhappy with your broker, did you replace him?
The beauty of this, is that thinking along these lines, IMHO, can/will help you in any endeavor life has in store for you.
Drawdowns are a cost of doing business, losses are ok and profits are ok. Acceptance of these are a good thing and all these issues should be tended to prior to trading.
Gordon
I think you just received some very sound advice.
I would add the perspective that you tend to view your trading in, will make the task of trading easier or harder to bare and that tends to be a good indication of your trading being a profitable or losing venture.
Think of the mindset of a golfer. He is leading a tournament by a stroke and he has a 3 foot putt on the 18th for the win. As he stands over the putt, his mind can be thinking something along the lines of:
- Small break, smooth follow through
or
If I make this, I’m the Champ, What will I say as I get the trophy? I’ll have my agent do a deal with Nike.
Trading is the same, once you have a real system and the environment setup to trade, your job is to simply give your systems the best chance to succeed. This comes down to management of oneself.
Once you know what you are dealing with, deal with it. I have found focusing on performance of trading in terms of equity might not be the "best" way, as your emotions might ride the equity curve. This might make the drawdowns an issue. Instead, focus on the performance of trading as a function of the management of the system. Did you place all the trades? Did you hesitate? If you are unhappy with your broker, did you replace him?
The beauty of this, is that thinking along these lines, IMHO, can/will help you in any endeavor life has in store for you.
Drawdowns are a cost of doing business, losses are ok and profits are ok. Acceptance of these are a good thing and all these issues should be tended to prior to trading.
Gordon
-
- Roundtable Fellow
- Posts: 58
- Joined: Wed Apr 30, 2003 4:35 pm
- Location: Netherlands
my mental rehearsal
William,
An example: You made a appointment with your friend at 1:15 pm ( your worst drawdown during testing) When it's 1:17 pm ( your current dd is a bit worse ) do you know when he is coming( when will the dd end )? In my case I got a bad feeling. It's 1:30 pm ( your drawdown is bigger ) do you still think he is coming. Most of the time he is right on time ( your testing says that your drawdown is no more then 25% )....what happened....you ask yourself.......an accident ( is my system ok ?)...or won't come anymore ( ruin ) Shall I call him ( stop trading and researching, or second guessing )
It's 2:00 pm and you decide to call and when you are calling ( stop trading ) your friend is coming and tells what happened ( your tests give the worst drawdown till now ).
I am just trying to give an example how I feel when uncertainty is coming. What do I think etc...... If you trust your friend ( system ) completely you will never call and just wait till he comes. If you have enough time ( enough capital and low betsize ) you know that it is worth waiting......
This is my mental rehearsal in combination with testing. But I don't have a friend till now who I trust completely.....
Marc
An example: You made a appointment with your friend at 1:15 pm ( your worst drawdown during testing) When it's 1:17 pm ( your current dd is a bit worse ) do you know when he is coming( when will the dd end )? In my case I got a bad feeling. It's 1:30 pm ( your drawdown is bigger ) do you still think he is coming. Most of the time he is right on time ( your testing says that your drawdown is no more then 25% )....what happened....you ask yourself.......an accident ( is my system ok ?)...or won't come anymore ( ruin ) Shall I call him ( stop trading and researching, or second guessing )
It's 2:00 pm and you decide to call and when you are calling ( stop trading ) your friend is coming and tells what happened ( your tests give the worst drawdown till now ).
I am just trying to give an example how I feel when uncertainty is coming. What do I think etc...... If you trust your friend ( system ) completely you will never call and just wait till he comes. If you have enough time ( enough capital and low betsize ) you know that it is worth waiting......
This is my mental rehearsal in combination with testing. But I don't have a friend till now who I trust completely.....
Marc
-
- Full Member
- Posts: 11
- Joined: Mon May 12, 2003 10:15 pm
- Location: Chicago, IL
-
- Roundtable Knight
- Posts: 344
- Joined: Sun Apr 20, 2003 2:42 pm
- Location: Silicon Valley / San Jose, CA USA
- Contact:
To quote SirG
Couldn’t agree with you more. I would also add, thoroughly exploring and meditating on your "system concepts.â€
I have found focusing on performance of trading in terms of equity might not be the "best" way, as your emotions might ride the equity curve. This might make the drawdowns an issue. Instead, focus on the performance of trading as a function of the management of the system. Did you place all the trades? Did you hesitate? If you are unhappy with your broker, did you replace him?
Couldn’t agree with you more. I would also add, thoroughly exploring and meditating on your "system concepts.â€
My $.02 worth is that as part of your system development, you should establish a hard floor on the drawdown you will tolerate. Your backtesting should give you a good idea how wide this ultimate stop should be. You SHOULD NOT ride an Enron all the way down to zero, or let an open short position go multiple hundreds of percent against you.
Will you wind up bailing on a position right before it turns back in your direction? Sure, trade long enough and it's bound to happen.
So have a re-entry strategy, with a very tight stop - or just forget about that position and let your energies focus more usefully elsewhere.
Will you wind up bailing on a position right before it turns back in your direction? Sure, trade long enough and it's bound to happen.
So have a re-entry strategy, with a very tight stop - or just forget about that position and let your energies focus more usefully elsewhere.
I will add my $0.02 as well. I have found that the advice of people who say "maybe you might want to investigate (X), perhaps in a computer test" has always been vastly superior to the advice of others who use the S word: "You SHOULD do (Y) and you SHOULD never do (Z)."
Generally the suggesters have a better track record than the dictators. That's my suggestion based upon observation
.
Generally the suggesters have a better track record than the dictators. That's my suggestion based upon observation
![Very Happy :D](./images/smilies/icon_biggrin.gif)
The problems of causality and of imperfect induction in the social sciences are beyond the scope of this forum but in general, I steadfastly believe risk cannot be quantified with precision in the social sciences as it is routinely done in the physical sciences. I feel we need to bestow some attention on the fundamental problems of philosophy to understand most system and risk issues. Mathematics and simulations by themselves are devoid of reason; they are simply tools for computation not reasoning. Reasoning is best done between your ears. Attempting to come up with an exact measure is futile. Following trends, letting profits run and cutting losses are the extent of our understanding; I find them plenty enough.
Permanent vs Temporary Change..
Identifying Permanent Change vs. Temporary
It seems as though you have to know the difference between a terrible drawdown and when your system is no longer working. Dunn and Campbell mention that they have changed or tweaked their systems over time to survive. How does someone know when to change their system? I assumed that trend following succeeds because markets ultimately trend and therefore pratically any tf system should do well enough and therefore requires no change
Any help? How do we determine permament change vs temporary change??
TIA
In reading a book by Elder he stated that systematic traders ultimately fail because their survival depends on a very discretionary element - knowing when your system is broken. c.f. has stated that the reason many of the turtles performance has slipped is because they haven't kept up with the times. To a fault, i gather, they may have stayed too loyal to their system that brought them. However, i have also read on this forum that you should rarely if ever change your system - sticking with a system is the most important element...Q: In the 20-plus years that you have been trading, what has been your greatest challenge?
Campbell: Maintaining our discipline. It is one thing to develop winning strategies, but quite another to overcome your doubts during the tough times and to stick to your guns. Another major challenge is finding the proper balance between maintaining one’s discipline and recognizing that adjustments must be made in response to subtle market changes. The difficult part is distinguishing between permanent changes, which require adjustments, and temporary changes for which adjustments should probably be avoided.
Campbell: It is axiomatic that research and development is critical to the success of any CTA. No advisor can last long on the basis of his or her warm personality or nice marketing material. If all the hoopla is not supported by real-time profits the world will beat a path to someone else’s door. That’s why we have intensified our research efforts to develop new and different approaches to trading, as well as to improve existing strategies. We currently have six full-time professionals, holding three PhDs spending 100% of their working days to find better ways to make money. That doesn’t mean we are abandoning the trading strategies that have worked for us all these years, but it does indicate that we recognize the world is constantly changing and that we need to stay on the cutting edge if we are to remain at the forefront of the industry.
Dunn: Research and development has been essential. Without it the likelihood of our survival would have been considerably less. Adapting to a changing environment requires an open mind and the willingness to expand resources to explore new ideas and technologies.
It seems as though you have to know the difference between a terrible drawdown and when your system is no longer working. Dunn and Campbell mention that they have changed or tweaked their systems over time to survive. How does someone know when to change their system? I assumed that trend following succeeds because markets ultimately trend and therefore pratically any tf system should do well enough and therefore requires no change
Any help? How do we determine permament change vs temporary change??
TIA
In this very question, I see the ugly head of FUD at work, undermining all your actionsWilliam wrote:How do we determine permament change vs temporary change??
![Sad :(](./images/smilies/icon_sad.gif)
My 2p perspective is this:
You always are on the look out for new systems, and
You constantly back test them on the same set of data, over the same period of time, and
You constantly phase out the worst performers and phase in the best one.
In that way, no fear, uncertainty or doubt anymore: you always do the same thing, all the time, which is keeping changing! But if your test time period is "long enough" I wouldn't expect you to switch systems every other day. On the contrary, only whan the tests do show that new system X turns out better than old system Y over an extended period of time, should you consider the switch. And I probably would make it piecemal, slowly reducing exposition through X and slowly increasing it through Y.
How about looking at what the Pro’s do? Does anyone know how some of the bigger trend guys decide when to change their systems? I know we could learn a lot from how trend followers like Dunn, Henry and Campbell decide to change their systems. Granted, they are probably not eager to give away those details. Anyone seen any interviews those trend followers that gives some insight into this subject?
JT
JT
-
- Roundtable Knight
- Posts: 344
- Joined: Sun Apr 20, 2003 2:42 pm
- Location: Silicon Valley / San Jose, CA USA
- Contact:
here is a previous discussion on when to change system rules:
viewtopic.php?t=374
this post also talked briefly about Dunn's experience with drawdowns and risk:
viewtopic.php?t=407&highlight=changing+systems
i agree with Verec: i am more interested in finding better systems and implementing them, rather than waiting for an unexpectedly large drawdown and THEN thinking about possibly changing my systems.
one very basic reason for changing/modifying/adding systems is sample size. suppose you do a system A test over a 10-year period that returns 20%/year with the largest drawdown being 15%. if you implement that system over the next 10 years, what is the chance that you will see a higher drawdown? i think it might be a lot higher than people think, only because there is a larger sample size of trades: as you get a larger sample, you experience more outliers.
now for comparison, suppose you do a system B test over a 5-year period that returns 20%/year with the largest drawdown being 15% (same as system A). if you implement that system over the next 15 years, what is the chance that you will see a higher drawdown? it will have a high probability of experiencing a larger drawdown than system A (because your test period is smaller, system B is quadrupling the sample size, versus system A only doubling the sample size.) both systems have a 20-year period, but you can see how a larger test period can have better performance (or less crappy performance) going forward.
knowing "when" to change systems is like predicting the future, which is something you cannot do, and it is something that good trend-followers don't do. it's more like you find a better system than your current one, and you go with the better one.
Edward
viewtopic.php?t=374
this post also talked briefly about Dunn's experience with drawdowns and risk:
viewtopic.php?t=407&highlight=changing+systems
i agree with Verec: i am more interested in finding better systems and implementing them, rather than waiting for an unexpectedly large drawdown and THEN thinking about possibly changing my systems.
one very basic reason for changing/modifying/adding systems is sample size. suppose you do a system A test over a 10-year period that returns 20%/year with the largest drawdown being 15%. if you implement that system over the next 10 years, what is the chance that you will see a higher drawdown? i think it might be a lot higher than people think, only because there is a larger sample size of trades: as you get a larger sample, you experience more outliers.
now for comparison, suppose you do a system B test over a 5-year period that returns 20%/year with the largest drawdown being 15% (same as system A). if you implement that system over the next 15 years, what is the chance that you will see a higher drawdown? it will have a high probability of experiencing a larger drawdown than system A (because your test period is smaller, system B is quadrupling the sample size, versus system A only doubling the sample size.) both systems have a 20-year period, but you can see how a larger test period can have better performance (or less crappy performance) going forward.
knowing "when" to change systems is like predicting the future, which is something you cannot do, and it is something that good trend-followers don't do. it's more like you find a better system than your current one, and you go with the better one.
Edward
A month or so ago I implemented my plan for dealing with a DD that is deeper than a certain threshold. One pillar of the plan requires a pause in some of my live trading with paper trading as a substitute. I have been a little disappointed with my emotions in this period of semi-cessation.
A good event would be for all my paper trades to start making money. The system can be deemed ‘working’ and confidence is restored (not that it was ever diminished
). At a predefined point of ‘draw up’ I stop paper trading and start risking money again. This would be an ideal situation. I should be pleased to see it eventuate. Interestingly, as a paper trade makes money, I feel frustrated by the fact that I have no money on it and am missing out on the one trade that I needed to take to get out of my hole. No use in it being a paper profit, right? I do not like this emotional gut reaction. On the flip side, I feel an irksome sense of triumph when a paper trade, a trade that I was ‘clever enough to have passed over’ turns into a loser. This warm feeling is wrong. I should be neutral if my paper trades are losing money and concerned if they continue to do so as it indicates that the system is indeed not ‘turning around’.
I have identified a number of possible equity curve eventualities that could evolve from my present situation and already have conceded to myself that some of them may well be emotionally ‘interesting’.
There is a lot more I could say on the topic but I am still discovering myself in the process so I will speak more later. Suffice to say that there is far more to a deep draw down that simply being in a stage where you are not making money.
I do not know what the pros do, but I should say that I am at this stage more interested in what the pros used to do whilst they were aspiring to become pros. That is, what road did they travel to reach their present point and did they get pebbles in their shoes along the way![Wink ;)](./images/smilies/icon_wink.gif)
damian
A good event would be for all my paper trades to start making money. The system can be deemed ‘working’ and confidence is restored (not that it was ever diminished
![Wink ;)](./images/smilies/icon_wink.gif)
I have identified a number of possible equity curve eventualities that could evolve from my present situation and already have conceded to myself that some of them may well be emotionally ‘interesting’.
There is a lot more I could say on the topic but I am still discovering myself in the process so I will speak more later. Suffice to say that there is far more to a deep draw down that simply being in a stage where you are not making money.
I do not know what the pros do, but I should say that I am at this stage more interested in what the pros used to do whilst they were aspiring to become pros. That is, what road did they travel to reach their present point and did they get pebbles in their shoes along the way
![Wink ;)](./images/smilies/icon_wink.gif)
damian
Eck,
Thats a great article, thanks for refreshing my memory about its existence.
Thats a great article, thanks for refreshing my memory about its existence.
I think these quotes really help me refocus on the right road. Thanks fellas for the input.Dunn: We only have two systems. The first system is the one I started with in 1974. The other system, we developed and launched in 1989. The major strategic elements of these two models—how and when to trade, how much to buy and sell—have never changed in almost 30 years. The markets that are available to trade and the amount of liquidity have changed completely. We have vastly different markets today, dominated by financial futures that didn’t even exist when we started. To the extent that physical commodities are available, their volume is quite modest, except for crude oil, and consequently systematic traders like us are constrained by capacity limitations. Although we started completely in physical commodities, and continue to trade them, now the money we have under management is over 90 percent in financial futures. We also have made a number of improvements in structuring the portfolio among the futures that we trade. The commitments or weights we give to different futures contracts have gone through a number of evolutions. We expect change. None of the things that have happened in the development of new markets over the past 30 years strike us as making the marketplace different in any essential way. The markets are just the markets. I know that is unusual. I know in the past five years a lot of competitors have purposefully lowered the risk on their models i.e. they are de-leveraging them or trying to mix them with other things to reduce the volatility. Of course, they have also reduced their returns.
Cleland: I would argue strenuously that what makes markets trend—the basic idea of the free market hypothesis—remains largely intact. Free markets are still driven by the fact that price is allowed to equilibrate shifts in supply and demand, and the human emotions of fear and greed. The most pronounced change is the diversification opportunities at the portfolio level. Our earliest portfolios in 1972 were probably composed of only eight or 10 markets. Today our portfolios have about 60 markets and some managers have 100. That’s probably the biggest change. In some cases, these additional markets are not even new markets. Foreign exchange markets existed in 1972, but they just weren’t accessible to what was then a fledgling CTA community. Second, we are building more models. We began with one model 30 years ago, and for 20 years, we had two models. We currently have eight models and in another 12 months, we’re likely to have 12 or 14. Improved technology enables us to do a better job in both risk management and in the markets. That has in turn attracted more assets, which has improved our revenue stream, which has enabled us to either buy more and better technology, or hire higher caliber staff. In that sense, the business has changed a lot.
Rzepczynski: There has been surprisingly little change. Models we developed 20 years ago are still in place today. Obviously, we trade a different mix of markets. We’ve also added new programs over the last 20 years, but relative to many of our peers, we have not made significant adjustments in our trading models. We believe that markets are always changing and adjusting, and the information that’s important to investors will also change. In the 1980s, everyone was interested in the money supply figures… everyone would wait by their phones until that number came out. In the 1990s, the information du jour was unemployment numbers. But people’s reactions to the markets are fairly stable. Uncertainty creates trends and that’s what we’re trying to exploit. Even if you have better and faster dissemination of information, the one thing we haven’t really improved is people’s ability to process information. We’re trying to exploit people’s reaction, which is embedded in prices and leads to trends. These reactions are fairly stable and may not require major adjustments of models.
I just had a look at InnerWorth (that was mentionned in some other thread here), and one thing strikes me as relevant to this thread (how do you cope with changing market conditions, how do you know when to follow the rules and when to break them).
Q: How did we justify the existence of i that number whose square root is -1?
A: By adding one level of abstraction or indirection, and moving out of the line into the plane.
As someone expressed "the only constant in life, is change", and taken to the letter, this probably provides the clue to which hang on to; at some level you need some stability: not everything can change from under your feet at once.
If you go to that "next level" where stability exists, then you can view "change" as part of the process, in the same way as issuing a single trade does change your portfolio equity, but doesn't otherwise change your other endehavors.
The question, then, becomes:
Can you find a stable (ie: not changing) process by which you can cope with changing market conditions?
And to that, I would definitely answer a big YES!
Q: How did we justify the existence of i that number whose square root is -1?
A: By adding one level of abstraction or indirection, and moving out of the line into the plane.
As someone expressed "the only constant in life, is change", and taken to the letter, this probably provides the clue to which hang on to; at some level you need some stability: not everything can change from under your feet at once.
If you go to that "next level" where stability exists, then you can view "change" as part of the process, in the same way as issuing a single trade does change your portfolio equity, but doesn't otherwise change your other endehavors.
The question, then, becomes:
Can you find a stable (ie: not changing) process by which you can cope with changing market conditions?
And to that, I would definitely answer a big YES!
JK, good to see you are keeping me honest ![Wink ;)](./images/smilies/icon_wink.gif)
I trade a blend of systems that in sum are ‘the system’. The component (sub-system) that I refer to as being in deep drawdown is a more generic trend follower, like Aberration. The system component that you mention in your post is currently producing a flat yield curve, making 10-15% one month and losing it the next. Thus the net performance of my overall ‘strategy’ is deep drawdown.
I do not know how the big commercial systems are doing at the moment, I tend to only check in on them once a year or so, which is coming up pretty soon. I concede I may have been a little brash in my ‘better than any commercial system I have seen’ comments. Certainly the backtest performance metrics are the best I have seen, but how can I make a meaningful comparison unless I did a side by side test of the two system on the same portfolio and risk settings. I have only run the side-by-side with two other systems.
Floor testing procedure: probably. In fact yes. Anything with the word ‘historical’ is already floored. Did I do the best I could do under the circumstances? I think so. I could however have easily produced better or worse test results by changing the test portfolio.
cheers
damian
![Wink ;)](./images/smilies/icon_wink.gif)
I trade a blend of systems that in sum are ‘the system’. The component (sub-system) that I refer to as being in deep drawdown is a more generic trend follower, like Aberration. The system component that you mention in your post is currently producing a flat yield curve, making 10-15% one month and losing it the next. Thus the net performance of my overall ‘strategy’ is deep drawdown.
I do not know how the big commercial systems are doing at the moment, I tend to only check in on them once a year or so, which is coming up pretty soon. I concede I may have been a little brash in my ‘better than any commercial system I have seen’ comments. Certainly the backtest performance metrics are the best I have seen, but how can I make a meaningful comparison unless I did a side by side test of the two system on the same portfolio and risk settings. I have only run the side-by-side with two other systems.
Floor testing procedure: probably. In fact yes. Anything with the word ‘historical’ is already floored. Did I do the best I could do under the circumstances? I think so. I could however have easily produced better or worse test results by changing the test portfolio.
cheers
damian