Risk:Reward & Drawdown - The questions new traders should be asking?

[LiamH]

My point is, that many experienced traders (i'm including myself) are trading multiple systems for equity curve smoothing purposes, therefore it's impossible to answer your 5 questions with a single definitive response for each. You did ask traders to respond, so this may be why you're not getting any answers...

I should add though, that this topic, which you've brought up, is exactly what new traders need to be thinking about - so i agree with your reasons for posting it.

Fair enough Perhaps you could expand on your experiences in general then?

Perhaps you could take one of your strategies, as I assume they're all net profitable, and answer the questions based on that?

Your input will be appreciated

 

[VielGeld]

Well, the point of MM is not entirely position sizing -- it's also more simply the now almost cliché phrase "cutting losses and letting your profits run".

Position sizing can be anything so long as a DD doesn't hurt your account. The usual figure you'll find is 3% of account, and there's some truth to this. What it comes down to is your own performance, timeframe (case in point, are you risking 3% on a 10 pip or a 100 pip stop?), and how much you're willing to risk.

The Leopard has taken to saying 'round these parts that making just 1 or 2 points per day on ES can amount to very large yearly returns once scaling comes into play. I very much agree. I would add that just making a consistent 1% per day = 10x return per year, so in reality you needn't risk all that much to make substantial returns. It all depends on the consistency of your performance.

Also, imo, Psychology > MM > Method, though all 3 are important, of course. MM may become more important if the psycho aspect isn't affecting you, though.

Lastly, what size you settle on is ultimately personal. It's a combination of managing emotions, DDs, your own performance, and your account size.

Anyway, that's my take.

 

[tony1979]

Fair enough Perhaps you could expand on your experiences in general then?

Perhaps you could take one of your strategies, as I assume they're all net profitable, and answer the questions based on that?

Your input will be appreciated

If a new trader, or even not so new trader, starts trading a system, which he hasn't tested and verified for himself, then I can guarantee he will fail. The only way to get an idea for how bad things could get in the future, is to look back over a large sample size of trades from the past (But remember, your biggest drawdown is ahead of you!). In doing this, the individual is able to see their method's worst performing periods in the past, and how trading a certain % per trade influences the equity swings - both positive and negative. This is the difference between the pro who can trade through the drawdowns to get to the profits, and the amateur who's only worried about what they can make, and quits when they lose 45% of their account because psychologically they're finished.

 

[the hare]

This is the difference between the pro who can trade through the drawdowns to get to the profits, and the amateur who's only worried about what they can make, and quits when they lose 45% of their account because psychologically they're finished.

I agree with everything you say, but would you tolerate a 45% drawdown ? I can think of specific situations where I might, but that seams on the high side even for a mechanical strategy.

 

[tony1979]

I agree with everything you say, but would you tolerate a 45% drawdown ? I can think of specific situations where I might, but that seams on the high side even for a mechanical strategy.

Aiming for the gains I do, I understand that those size drawdowns, and sometimes more, can and do happen. Even if you're trading only half of your method's optimum position size, the drawdowns can be astounding! Newer traders with only a couple of years experience should be trading much, much smaller stakes, and should stay well away from Kelly and Optimal F derivatives. Whatever drawdown % they "think" they can handle, they should slash it in half at least, and trade at the position size which would result in this figure.

There should be no hesitation in execution, and if there is, you're (pl.) trading too much size.

 

[the hare]

Aiming for the gains I do, I understand that those size drawdowns, and sometimes more, can and do happen.

Thats good advice.

For years I used to retrospectively analyse previous returns, looking at what I could have made using a variety of different position sizing methods. I did a fair amount of statistical resampling on that data to try and determine worst case drawdowns, and every year I almost pursuaded myself to use an optimal f type approach on at least a couple of sytems. In retrospect I'm glad I didnt, although I'm still inclined to allocate a couple of percent of the account each year and treat it as a lottery ticket.

I suppose the justification is that depite the drawdown, the equity curve will still be higher than it would have been using a less agressive approach.

I just need to replace my testicles with balls of steel and I'll be fine.

 

[tony1979]

Thats good advice.

For years I used to retrospectively analyse previous returns, looking at what I could have made using a variety of different position sizing methods. I did a fair amount of statistical resampling on that data to try and determine worst case drawdowns, and every year I almost pursuaded myself to use an optimal f type approach on at least a couple of sytems. In retrospect I'm glad I didnt, although I'm still inclined to allocate a couple of percent of the account each year and treat it as a lottery ticket.

I suppose the justification is that depite the drawdown, the equity curve will still be higher than it would have been using a less agressive approach.

I just need to replace my testicles with balls of steel and I'll be fine.

The optimal f/kelly path is fraught with danger, and I'm glad too that you made the right decisions, retrospectively. However, it's useful, if not imperative, to know the figure in order to undertrade it at the minimum. Most new traders who are risking weighty %'s on a few markets have no idea that they're probably already exceeding it.

Yes, in theory the equity curve will be higher, however it's unlikely you'll (pl.) know if you can ever deal with a 50%+ drawdown until you're actually there. It's worth remembering too, that i'm referring to these sized drawdowns in the context of doing EVERYTHING perfectly, with no mistakes at all.

 

[the hare]

The optimal f/kelly path is fraught with danger, and I'm glad too that you made the right decisions, retrospectively.

In my case I was aware of what might happen from backtests and resampling etc, but year after year, knowing the probability of these worst case drawdowns occurring was low, and then seeing the potential benefits, when applied retrospectively, almost convinced me to pursue this further. Intellectually I knew better, but I still tried to justify the approach to myself

A couple of years ago, I went through the same retrospective exercise, on around 8 years worth of data, and at long last finally encountered a series of trades where things went extremely ugly, very quickly. Its hard to explain, but although neither instance was real life, there's bit of a difference between seeing extreme drawdowns in a theoretical equity curve simulated on monte carlo testing, and the same drawdowns on an equity curve based on your own trades !

The really frightening thing for me was that the sequence of events that led to what would have been a blown account where nothing really out of the ordinary, or anything that caused me concern at the time when I was trading the system. These sorts of position sizing methods are just inherently unstable, and no more than a lottery really.

 

[tony1979]

A couple of years ago, I went through the same retrospective exercise, on around 8 years worth of data, and at long last finally encountered a series of trades where things went extremely ugly, very quickly. Its hard to explain, but although neither instance was real life, there's bit of a difference between seeing extreme drawdowns in a theoretical equity curve simulated on monte carlo testing, and the same drawdowns on an equity curve based on your own trades !

The really frightening thing for me was that the sequence of events that led to what would have been a blown account where nothing really out of the ordinary, or anything that caused me concern at the time when I was trading the system. These sorts of position sizing methods are just inherently unstable, and no more than a lottery really.

At least you found out the cheap way that these algorithms aren't for you. They do have their merits, however, the application of them is very subjective. If we really went into more depth about how extreme the drawdowns can be (both in size and length of time), while executing perfectly, most would think you insane or just plain stupid for even considering trading at a significant value of the optimum number. Saying all this though, I use a derivative of them to size my trades, but I know I have a high tolerance of risk.

 

[piggybridges]

I'm a winning player at NL50 - I make about £20 per hour playing poker...

Then I'd just stick with the poker then, personally.
Even better, start a poker journal here. I'd be interested to watch the results of a consistent winner.
Good luck

 

[wekim]

Poker is way easier than trading. The problem is scaling up. I haven't played online in a while but the games at NL1k and up were a sit-out fest. NLHE is definitely dying, you need to learn PLO to find new edges. Also if you live in the US you're pretty screwed with the online situation. I've had to grind at the live cardrooms for 6 months but I've made a good bankroll for this trading stuff. We'll see.

I don't know how you guys can use Kelly for trading because I'm finding it near impossible to quantify your edge.

 

[the hare]

I don't know how you guys can use Kelly for trading because I'm finding it near impossible to quantify your edge.

you hit the nail on the head. the magnitude of the edge varies through time.