Automated Portfolio Level Risk Management

[AyePee]

Hi there,

I have been working on the risk management rules of my trading system recently. I have made some improvements within the risk management rules at an individual trade level.

I have tried many different risk management rules at a Portfolio level, without any significant improvement over the trade level risk management rules.

I am wondering if anyone else has had any similar experiences, or has any insight into this issue? If anyone has anything to offer, it would be appreciated.

Cheers,

AP

 

[amnonco]

hi AP,
first of all you need both individual trade level AND portfolio level risk management.
of course, it could be that with your individual rules combined with # of opportunities you have, you already have a robust portfolio level risk managment scheme.

however, it's really a matter of deeply understanding your system in order to know how portfolio level risk management should be applied.

second, you may not see the results of your portfolio level risk management on your backtesting if you don't have enough data...

what methods of portfolio risk management you already tried?

 

[AyePee]

I have 15 years of back adjusted data across over 30 markets.

I have tried reducing position sizes after a drawdown of X%, I have tried reducing position sizes after a bad week of Y%. I have tried using moving averages of my equity curve to reduce position sizes as they crossover in a negative fashion. I have tried creating a giant variance-covariance matrix across all 30 securities (that was a ball tearer) and measuring the standard deviation of the daily PL (as a % of my equity) and reducing position sizes if it gets above a certain threshold. I have tried capping my number of positions to a certain number. And probably a few more that dont spring to mind immediately.

I consider myself to have a very deep understanding of my risk. I know what % of the trades make money, how much they each risk, how bad a poor day, week, month, quarter, year can be. The probability of making money in a week, month, quarter, year. The potential size and duration of a drawdown. And....I know my expected retrn and the variability of that return, across all time frames.

Combining both of the previous two paragraphs......all the attempts in the first paragraph, dont seem to lead to any improvement in the statistics mentioned in the second paragraph.

It could be that as you say, my trade level risk management rules are, on their own, providing good risk management for the entire portfolio. However, I find it a little odd that I am unable to find anything at the portfolio level that can reduce risk relative to return.

Do you have any suggestions that I could try?

Cheers,

AP

 

[amnonco]

I have 15 years of back adjusted data across over 30 markets.

I have tried reducing position sizes after a drawdown of X%, I have tried reducing position sizes after a bad week of Y%. I have tried using moving averages of my equity curve to reduce position sizes as they crossover in a negative fashion. I have tried creating a giant variance-covariance matrix across all 30 securities (that was a ball tearer) and measuring the standard deviation of the daily PL (as a % of my equity) and reducing position sizes if it gets above a certain threshold. I have tried capping my number of positions to a certain number. And probably a few more that dont spring to mind immediately.

I consider myself to have a very deep understanding of my risk. I know what % of the trades make money, how much they each risk, how bad a poor day, week, month, quarter, year can be. The probability of making money in a week, month, quarter, year. The potential size and duration of a drawdown. And....I know my expected retrn and the variability of that return, across all time frames.

Combining both of the previous two paragraphs......all the attempts in the first paragraph, dont seem to lead to any improvement in the statistics mentioned in the second paragraph.

It could be that as you say, my trade level risk management rules are, on their own, providing good risk management for the entire portfolio. However, I find it a little odd that I am unable to find anything at the portfolio level that can reduce risk relative to return.

Do you have any suggestions that I could try?

Cheers,

AP

ok, the chances for data problems are slim to none.

how do you position size an individual trade? fixed per equity? fixed fraction? fixed risk?

personally I use a correlation matrix combined with position size cap combined with total portfolio risk cap.

another point, the portfolio risk management rules maybe tested just once in a decade or more (in black swans events such as black monday...) but that only means you should use them even though your backtesting didn't show any actual difference...

 

[AyePee]

ok, the chances for data problems are slim to none.

how do you position size an individual trade? fixed per equity? fixed fraction? fixed risk?

personally I use a correlation matrix combined with position size cap combined with total portfolio risk cap.

another point, the portfolio risk management rules maybe tested just once in a decade or more (in black swans events such as black monday...) but that only means you should use them even though your backtesting didn't show any actual difference...

I size my positions by risking a constant amount of equity over a measure of volatility I use. This means all trade risk the same % of capital, in the more volatile markets I trade less with wider stops, and in the quieter markets I trade more with tighter stops.

I agree with your point about the portfolio risk management rules only being needed occasionally. However, I find that I lose a lot of performance when I try to implement portfolio level risk management rules.

 

[amnonco]

I size my positions by risking a constant amount of equity over a measure of volatility I use. This means all trade risk the same % of capital, in the more volatile markets I trade less with wider stops, and in the quieter markets I trade more with tighter stops.

I agree with your point about the portfolio risk management rules only being needed occasionally. However, I find that I lose a lot of performance when I try to implement portfolio level risk management rules.

an ATR stop of some kind, o.k - this means you risk, as you said, the same amount of % of capital.
so, let's say you trade 50 markets, without regarding their correlations. and you take only 0.5% risk in 30 of them simultaneously. this means you have portfolio risk of 15%.

let's imagine a black swan, n.Korea throws a nuclear bomb, the U.S goes bankrupt, or whatever. markets is gaping at the open 7%....you lose on average (cause you maybe liquidaed the trades sooner or whatever), 105% of you capital.

now that story is a fiction, but if you don't account for that kind of story - what will happen then? you'll have great result over the next 4.5 years - really amazing results, and then one day. puff - it's gone.

unless you limit your total portfolio "heat" as seykota calls it, you may find yourself in a situation like that.

portfolio risk rules are a measure of defense not for improving results. your ATR stop as a % of equity does that for you.

 

[Adamus]

let's imagine a black swan, n.Korea throws a nuclear bomb, the U.S goes bankrupt, or whatever. markets is gaping at the open 7%....you lose on average (cause you maybe liquidaed the trades sooner or whatever), 105% of you capital.

That's a great reason not to hold your positions overnight. I suppose you'll say there's still a risk that some looney terrorist will fly a 747 into your exchange in which case it will close immediately... I guess the answer for me is just to set myself a total profit target and to stop trading if I'm lucky enough to reach it before my account-annihilating Black Swan arrives.

Isn't the alternative just to make the normal returns that you could make on a normal well-diversified investment portfolio?

 

[amnonco]

That's a great reason not to hold your positions overnight. I suppose you'll say there's still a risk that some looney terrorist will fly a 747 into your exchange in which case it will close immediately... I guess the answer for me is just to set myself a total profit target and to stop trading if I'm lucky enough to reach it before my account-annihilating Black Swan arrives.

Isn't the alternative just to make the normal returns that you could make on a normal well-diversified investment portfolio?

I actually wasn't going to say that there's a risk your exchange will blow up...I'd say holding position overnight can be very profitable IF you handle risks accordingly.
think about all the running gaps (that are way more probable during a trend) that day traders miss.

what would you do if your profit target is one you know should take 10 years of trading to achieve? the chance for a 2nd lehman, a 3rd world war would increase dramatically...

 

[Adamus]

what would you do if your profit target is one you know should take 10 years of trading to achieve? the chance for a 2nd lehman, a 3rd world war would increase dramatically...

I take your point and I think the main difference between us is that you are quite happy to get stuck in hedging your risks appropriately, but I would rather avoid and can avoid a lot of those risks you cite by choosing to day trade the futures and forex markets.

I wouldn't be affected by a 2nd Lehman Bros or a 3rd WW - unless it took out my market physically - the worst effect would be some horrendous slippage as my stops get hit by a free-falling market.

And if I wanted even less exposure, I would trade long options only, although there you have a different cost associated with premium decay to factor in.

So I think that's my response to AyePee - try reducing the risk instead of hedging it. Probably not feasible though if transaction costs are too high.

By the way have you read Trading Risk by K L Grant? It's sitting on my bedside table waiting to be read but I haven't got to it yet.

 

[amnonco]

I take your point and I think the main difference between us is that you are quite happy to get stuck in hedging your risks appropriately, but I would rather avoid and can avoid a lot of those risks you cite by choosing to day trade the futures and forex markets.

I wouldn't be affected by a 2nd Lehman Bros or a 3rd WW - unless it took out my market physically - the worst effect would be some horrendous slippage as my stops get hit by a free-falling market.

And if I wanted even less exposure, I would trade long options only, although there you have a different cost associated with premium decay to factor in.

So I think that's my response to AyePee - try reducing the risk instead of hedging it. Probably not feasible though if transaction costs are too high.

By the way have you read Trading Risk by K L Grant? It's sitting on my bedside table waiting to be read but I haven't got to it yet.

first of all haven't read it - if it's any good - let me know.

second, of course you can day trade to eliminate (or minimize) the risks above, however there are downsides to day trading which are of course the long term advantages.

if we're talking managing money - day traders often encounter a liquidity problem with large amounts of money.
moreover, noise - day traders have to account for a helluva lot market noise, that won't help their trading...

to sum up, there are numerous ways to make money trading the markets, the best one is the one that fits you the most. apparently, we agree on that one.

 

[thierry911]

Have you tried to add some markets in order to diversify. have you tried your system only long or short... that could help as well.