This is why trendfollowing is hard
This is why trendfollowing is hard
I have been usingveritrader/tradingblox since 2004 - I can honestly say this is the hardest time ive ever faced - why ? because this is when the realities of the emotional pull really kick in - up about 35% year to date - up about 100 % since August last year - every bone in my body desperately wants to take the money and run like hell (thanks floyd) - surely if i get out here i'll get back in on a nice draw down and everything will work out - NO !!! making money is so much harder than losing money from an emotional tug point of view - im trying to analyse what Im scared of right here - if i give back some profits - so what ?? it really is a learning experience even after you have been doing this job for 20 years !!!
Take some money of the table!
Why not take some money of the table and play with a smaller trading capital.
Bahamas sounds nice!
Bahamas sounds nice!
-
- Site Admin
- Posts: 9015
- Joined: Tue Apr 06, 2004 1:41 pm
- Location: Boston, MA
- Contact:
IF that was part of your system design and you backtested this scenario and you are comfortable with the results, then this is fine. But if you did not backtest taking money off the table, then unfortunately that option is not on the table. Trade what you test, and test what you trade.Why not take some money of the table and play with a smaller trading capital.
Recent market
Well put Tim.
That being said - I had a similar experience as Chris in recent weeks. I found that having a big equity run up in such a short period of time was much more stressful than I could have imagined. I think the hardest part was a few weeks back when you could just feel a top coming. I told someone that I work with that I felt like I was standing in the middle of the road and could see a gigantic truck coming at me, but still refuse to get out of the way.
I suppose none of this is really a suprise - I know from my testing and experience that this has happened before, and will happen again.
That being said - I had a similar experience as Chris in recent weeks. I found that having a big equity run up in such a short period of time was much more stressful than I could have imagined. I think the hardest part was a few weeks back when you could just feel a top coming. I told someone that I work with that I felt like I was standing in the middle of the road and could see a gigantic truck coming at me, but still refuse to get out of the way.
I suppose none of this is really a suprise - I know from my testing and experience that this has happened before, and will happen again.
Re: Recent market
Sounds like an episode from modern day Battlestar Galactica. hahaphlub wrote:Well put Tim.
That being said - I had a similar experience as Chris in recent weeks. I found that having a big equity run up in such a short period of time was much more stressful than I could have imagined. I think the hardest part was a few weeks back when you could just feel a top coming. I told someone that I work with that I felt like I was standing in the middle of the road and could see a gigantic truck coming at me, but still refuse to get out of the way.
I suppose none of this is really a suprise - I know from my testing and experience that this has happened before, and will happen again.
You're going through what everyone else goes through. I found that I'm just now able to emotionally remove myself from the ups and downs because of one thing: serious diversity of trading systems and the markets I trade. It requires more capital, but you will be less worried about your trading if you're spreading risk from equity markets to Forex to commodities, and from short timeframes to long. Going one step further, if you can automate your trading so you don't have to make any real decisions (other than turning on the software), you're going to experience less emotion and fewer gray hairs.
D
Many years back, this type of "heat" feeling is what drove Tom Basso to add some portfolio level risk management tools that he wrote about in a booklet distributed by the CME and gladly shared with anyone. I think it was the silver market that had his trend following portfolio changing by large percentages every day. When we have moves like those in wheat, beans, gold, silver, etc, like recently, what were once small risks to the portfolio (i.e. 1.5% of equity at entry), can turn into large positions.
A few things you can do include using individual market risk limits and total portfolio heat limits. For example, if any one position's open risk exceeds a given threshold, you scale back that position size until it's risk falls below that threshold. In huge parabolic runs, risk scaling-out algorithms often work very nicely in my experience. Also, you can limit total portfolio risk to a given level and scale out of positions as that risk limit is exceeded. Also, you can limit volatility portfolio risk to a given level, and when equity vol exceeds that limit, you scale back all positions. Also, you can limit an individual market's volatility as a percentage of equity (i.e. if Gold position has an equity vol greater than some % of portfolio equity, you scale it back to get it within bounds).
These are just a few of the risk management tools that help create a smoother reward to drawdown (marked to market daily DD) ratio over the long haul and help you sleep better at night too.
A few things you can do include using individual market risk limits and total portfolio heat limits. For example, if any one position's open risk exceeds a given threshold, you scale back that position size until it's risk falls below that threshold. In huge parabolic runs, risk scaling-out algorithms often work very nicely in my experience. Also, you can limit total portfolio risk to a given level and scale out of positions as that risk limit is exceeded. Also, you can limit volatility portfolio risk to a given level, and when equity vol exceeds that limit, you scale back all positions. Also, you can limit an individual market's volatility as a percentage of equity (i.e. if Gold position has an equity vol greater than some % of portfolio equity, you scale it back to get it within bounds).
These are just a few of the risk management tools that help create a smoother reward to drawdown (marked to market daily DD) ratio over the long haul and help you sleep better at night too.
If you have stong feelings about taking money off the table, or lightening up, these positively intended feelings (which are intended to get you focused on your risk control) will find a way to express themselves, probably through some unplanned drama. I think you have two options: work through your feelings, preferrably with a support group, so that you can learn to live with them; or, incorporate these feelings into your trading system through a tested set of objective rules that will allow these positively intended feelings to express themselves in a structured, profitable, and undramatic manner.
-
- Roundtable Knight
- Posts: 122
- Joined: Thu Apr 17, 2003 9:49 am
There's no big problem here. Just make "taking money off the table after a huge run-up" a part of your fully mechanical system. Try modifying your system as follows:
Calculate Bollinger Bands on the system's equity curve. When you are in a period of enormously-bigger-than-normal profitability (i.e. above the upper Bollinger Band), reduce positionsize of all trades (existing and new) by X percent. Take money off the table by exiting a portion of each trade ("cashing out"). If and when the equity curve crashes back below the upper band, your reduced exposure means your %drawdown will be smaller.
Your Bollinger Band calculation probably needs to include the fact that the equity curve has geometric growth and is plotted on a log scale.
Now you've got a new, fully mechanical system, that you can backtest. If you like it then you can follow Tim's excellent advice to trade what you test, test what you trade.
Your system code probably need to keep track of two equity curves, the No-Cashing-Out equity curve (upon which you calculate BBands), and the With-Cashing-Out equity curve (your system's actual results).
Calculate Bollinger Bands on the system's equity curve. When you are in a period of enormously-bigger-than-normal profitability (i.e. above the upper Bollinger Band), reduce positionsize of all trades (existing and new) by X percent. Take money off the table by exiting a portion of each trade ("cashing out"). If and when the equity curve crashes back below the upper band, your reduced exposure means your %drawdown will be smaller.
Your Bollinger Band calculation probably needs to include the fact that the equity curve has geometric growth and is plotted on a log scale.
Now you've got a new, fully mechanical system, that you can backtest. If you like it then you can follow Tim's excellent advice to trade what you test, test what you trade.
Your system code probably need to keep track of two equity curves, the No-Cashing-Out equity curve (upon which you calculate BBands), and the With-Cashing-Out equity curve (your system's actual results).
The point to mine and other posts above was not to simply take "money off the table on any given trade as that’s how you miss the big moves". These points were directed toward handling open risk once you have a big more or perhaps MANY big moves in your portfolio. The correlated big moves in portfolios are perhaps a twice a decade thing. I think it was Eckhardt who said the main rule was knowing when to break your rules (like him closing out his large short Eurodollar position the day of the stock market crash in 1987).Chris67 wrote:UPDATE TO POST
Would never contemplate taking money off the table on any given trade as thats how you miss the big moves
currently in a 15 % draw down as one would expect - maybe it gets worse - maybe it all goes back !! who knows and if that bothers you then don't trend follow
C
It all comes down to what you want from your system and account. If you're in a trade that has already gone more than 40 to 50 ATRs from the entry point, creative risk management will help maximize the return to dd you achieve. Usually trades like these end in a blowoff move if they are long positions in hard commodities or short positions in stock futures. Proper position size scaling out based on any number of ideas as above will help capture more of the net move with less portfolio equity volatility, and at rare moments a bell will ring in your head like it did for Eckhardt. Whether you choose to listen to the bell or even hear it will be 100% a function of your mental state at that crazy point in time.
Oh, I hear that bell. Trouble is I hear it several times before the end of the nice nice phase. If I'd opened the door every-time temptation rang, i'd have never been able to see the mountaintop. Or something like that...Chuck B wrote:Whether you choose to listen to the bell or even hear it will be 100% a function of your mental state at that crazy point in time.
You're listening to the wrong frequency bell then...RedRock wrote:Oh, I hear that bell. Trouble is I hear it several times before the end of the nice nice phase. If I'd opened the door every-time temptation rang, i'd have never been able to see the mountaintop. Or something like that...Chuck B wrote:Whether you choose to listen to the bell or even hear it will be 100% a function of your mental state at that crazy point in time.
![Wink :wink:](./images/smilies/icon_wink.gif)
Ah. Ill have to revisit my yogi for a hearing tune-upChuck B wrote:You're listening to the wrong frequency bell then...RedRock wrote:Oh, I hear that bell. Trouble is I hear it several times before the end of the nice nice phase. If I'd opened the door every-time temptation rang, i'd have never been able to see the mountaintop. Or something like that...Chuck B wrote:Whether you choose to listen to the bell or even hear it will be 100% a function of your mental state at that crazy point in time.
![Cool 8)](./images/smilies/icon_cool.gif)
-
- Roundtable Knight
- Posts: 2071
- Joined: Fri Apr 25, 2003 3:33 pm
- Location: London
- Contact:
Scaling out seems to work very well indeed on a long term system.
My tests have tended to show that for a given CAGR, a system using scaling out has a considerably lower risk in terms of depth and length of DDs, std dev of return, higher Sharpe etc than the identical system which does not employ scaling out.
I confess that in the testing I have done, I have seen little amelioration as far as losing trades are concerned. The effect is concentrated on the winners: much as Mark and Chuck describe above.
When I say "identical system", perhaps I should add the caveat that in order to achieve the same CAGR it often seems necessary to increase the bet size on the "scaling out" version.
Another interesting exercise, often alluded to by Mark, is to run tests on short only trading with "Earn Interest" set to "False". It is interesting to note what trading short adds or does not add to your system in terms of return and volatility.
My tests have tended to show that for a given CAGR, a system using scaling out has a considerably lower risk in terms of depth and length of DDs, std dev of return, higher Sharpe etc than the identical system which does not employ scaling out.
I confess that in the testing I have done, I have seen little amelioration as far as losing trades are concerned. The effect is concentrated on the winners: much as Mark and Chuck describe above.
When I say "identical system", perhaps I should add the caveat that in order to achieve the same CAGR it often seems necessary to increase the bet size on the "scaling out" version.
Another interesting exercise, often alluded to by Mark, is to run tests on short only trading with "Earn Interest" set to "False". It is interesting to note what trading short adds or does not add to your system in terms of return and volatility.
This type of phenomenon is "analyzed" (in my opinion, bungled quite badly) by George Pruitt in the current (#2-2008) issue of Futures Truth magazine. He runs the simulations, he's got the results data right in front of him, and what boneheaded conclusion does he draw? That a long-only mechanical trading system beats buy-and-hold.AFJ Garner wrote:exercise: ... run tests on short only trading with "Earn Interest" set to "False". It is interesting to note what trading short adds or does not add to your system in terms of return and volatility.
Wow.
I've seen this too. However decided that it was another element of fitting which I didn't want in my case. Wasn't comfortable cranking up the leverage to compensate for added rules. I chose a lower bet with less tuning.AFJ Garner wrote:Scaling out seems to work very well indeed on a long term system.
My tests have tended to show that for a given CAGR, a system using scaling out has a considerably lower risk in terms of depth and length of DDs, std dev of return, higher Sharpe etc than the identical system which does not employ scaling out.
-
- Roundtable Knight
- Posts: 1436
- Joined: Mon Dec 22, 2003 12:24 pm
- Location: Des Moines, IA
- Contact:
I attempted this a few months ago, prior to stumbling across this post. I did not consider the exponential nature of the equity curve at that time. I am not a super-duper math guy, but I think that if i take the ln of the equity curve numbers and calc the bbands on that i'll get the correct numbers? But that got me thinking, some trading instruments also exhibit exponential growth behavior, why dont we adjust the bands in this same manner for them?Mark Johnson wrote:There's no big problem here. Just make "taking money off the table after a huge run-up" a part of your fully mechanical system. Try modifying your system as follows:
Calculate Bollinger Bands on the system's equity curve. When you are in a period of enormously-bigger-than-normal profitability (i.e. above the upper Bollinger Band), reduce positionsize of all trades (existing and new) by X percent. Take money off the table by exiting a portion of each trade ("cashing out"). If and when the equity curve crashes back below the upper band, your reduced exposure means your %drawdown will be smaller.
Your Bollinger Band calculation probably needs to include the fact that the equity curve has geometric growth and is plotted on a log scale.
Now you've got a new, fully mechanical system, that you can backtest. If you like it then you can follow Tim's excellent advice to trade what you test, test what you trade.
Your system code probably need to keep track of two equity curves, the No-Cashing-Out equity curve (upon which you calculate BBands), and the With-Cashing-Out equity curve (your system's actual results).