Traders' Roundtable

I'd be curious to know how many people use this function.

ColdFact wrote:I'd be curious to know how many people use this function.

Don't use in system development -- But in real life I use...
For me Not easy to trade full size if more than 10% loss...
But flip side is -- it takes longer to recover from drawdown

You can reduce the number of contracts you trade when the drawdown exceeds, say, 10% and expand the number back when the drawdown is less than 10%. The duration of the drawdown then is approximately the same, and the recovery time will be approximately the same. The drawdown will be less deep. In case of an extreme drawdown, especially if you are starting up a new system, you have a higher change of surviving and keep the system running.

condor - if only it were as simple as that then im sure everyone would use it and it would holy grailish in its application - a bona fide way of avoiding deep drawdowns - so can you think why that isnt the case ?
i dont use

Back in the day my colleague and I explored a whole raft of DDrm. All the ones with a discrete binary state switching gate(or multiple) were understandably crap or curvefitting exercises. We had some success with non-linear continuous drop-off functions and successfully used a couple of variants, but at the end of the day it was still quite a subjective choice whether or not to bother at all.

Nowadays I find that Draw down duration/recovery is just as important a performance parameter as absolute depth(provided leverage is sensible), and all the useful DDrm typically work against you in that regard, so aren't very useful.

Following on from that train of thought, trying to martingale your dd sounds nice in theory but in practice is a fast road to system fragility and failure for obvious reasons.

I've noticed in a lot of books that interview managers (various Market Wizards books, Inside the House of Money, etc), many of the managers cut position sizings during drawdowns. The big missing part is that many don't talk about how/when they ramp back up.

IF YOU use a fixed frac money manager then "sort of" by default you cut position size when losing ?
Its probably still the best way
I think problem with D/D reduction thresh is that if an equity curve hovers around the threshold for say 3 months then you can find yourself in and out of additional positions so often that you end up losing more bloody money than if you hadnt bothered in the first place !!

mojojojo wrote:I've noticed in a lot of books that interview managers (various Market Wizards books, Inside the House of Money, etc), many of the managers cut position sizings during drawdowns. The big missing part is that many don't talk about how/when they ramp back up.

Which is perhaps the most important part. You could be negatively affecting your portfolio by prolonging the time it takes to get out of the DD by cutting down the size of your potential wins. Particularly if those wins are of a larger than normal size...

Aaron01 wrote:Which is perhaps the most important part. You could be negatively affecting your portfolio by prolonging the time it takes to get out of the DD by cutting down the size of your potential wins. Particularly if those wins are of a larger than normal size...

If your returns are serially correlated, and TF returns often are, then this is like a stop on your strategy, and would make sense.

My takeaway from the interviews with people who cut size after losses by more than the reduction in account size was

A. Turtle trainees: Possibly people with big losses are messing it up and need to not trade so much of my money.

B. Others say something like "I cut my size until I feel in tune with the markets again". This can be good advice because a lot of things can mess up trading. Who has traded well through a divorce for example?

If you believe there is no serial correlation in your strategy, and your strategy is going to work for a long time - no need to reassess if it is still valid after a drawdown - and you are not going to subtly undermine it for any reason, then cutting size a lot doesn't make sense.

stopsareforwimps wrote:
If your returns are serially correlated, and TF returns often are, then this is like a stop on your strategy, and would make sense.

If you believe there is no serial correlation in your strategy, and your strategy is going to work for a long time - no need to reassess if it is still valid after a drawdown - and you are not going to subtly undermine it for any reason, then cutting size a lot doesn't make sense.

Well, let's use a hypothetical. Assume your TF strategy comes to an equity peak, then the dynamics of the markets shift into prolonged period of choppy, oscilating, sidways trading. Your equity takes a hit from all of the whipsaw action; lots of losses, some small wins or break evens, watching a lot of open equity give back, etc... So you cut back on your position size at your threshold to ameliorate your drawdowns. Now the markets break out of their channels and start trending, your portfolio goes to "fully" loaded in a matter of days.

If you had cut back additionally on your size with a DD threshold your wins would have less of a positive impact on your equity curve than they otherwise would have in a "normal" situation. This is the same as when Ralph Vince talks about being on the left side of the peak of the f curve can be just as dangerous as being on the right side of the peak of the curve.

Of course, it's always best to test and draw your own inferences...

I've seen guys who cut size in drawdowns do mipuch worse. They whip saw their account into much larger overall drawdown.

I've also seen the opposite.




As far as I know, ralph Vince never really made it.


This entire issue needs to be reframed.....instead of same old proposition.

Bravochico wrote:I've seen guys who cut size in drawdowns do mipuch worse. They whip saw their account into much larger overall drawdown.

I've also seen the opposite.

This entire issue needs to be reframed.....instead of same old proposition.

hear hear!

starting with "total equity" perhaps... It always seemed dumb to me that when using a TF strat that lets onside winners run free with a nice wide trailing stop, that intra-week or intra-month favourable price-spikes that then retrace quite far (perhaps before going onto new spiky highs again and again) penalise the system performance metrics versus a equity curve that goes absolutely nowhere at all....