The Effects of Increasing System Number: The Deterrents of Big Portfolios

A few weeks ago I wrote a post about big portfolios and some of the possible problems that might arise from the combination of a large number of systems into a portfolio without a proportional reduction in risk. Along the problems exposed within this post are the probabilities of large losses due to the large number of trades open at any given time and the possibility of an under or over estimation of profit and draw down targets due to broker dependency issues which get magnified as the number of trades – especially the number of conjunctively open trades – increases. On today’s post I am going to talk about some additional observations of this matter based on my analysis of an Asirikuy challenge account as well as what I believe are the main deterrents for the trading of large portfolios to aim for what seems to be unrealistic profit and draw down targets.

Certainly the addition of systems into a portfolio has several different effects. The first one is to increase the number of trades open at any given time and the second is to better hedge other systems by offering a better coupling between profit and draw down periods. Interestingly large system conglomerates seem to have a limited average compounded yearly profit to maximum draw down ratio of about 10 which is very unrealistic considering that it would imply a profit to draw down ratio higher than what anyone currently achieves on the market in the long term on an audited basis. However, even if this target was realistic, there seem to be some very interesting deterrents to the use of such accounts even if such large long term targets were achievable.

An Asirikuy challenger decided to test this hypothesis by trading a real account using 15+ Asirikuy system instances which have been trading for about one month. The expected average yearly profit to maximum draw down ratio is about 10, which is in line with the observation I have given before. The portfolio trades several types of strategies on several symbols including trend and counter trending strategies of different types which include breakout, price action, indicator-based and time-based strategies. After several weeks of what seems to be a very entertaining rollercoaster trading ride I decided to do a very in-depth analysis of this system and what its probable problems might be.

When you look at the statistical characteristics of this portfolio one thing immediately jumps out of the picture. The Ulcer Index of this portfolio is above 50. When you consider that portfolios like Atinalla No.3 and No.4 have Ulcer Index values below 15 (sometimes even below 10) there is a very strong sense that this value makes large portfolios VERY different from regular small portfolios and that this very high Ulcer Index may be the key as to why these portfolios are tremendously hard to trade. A careful analysis of how such a portfolio trades during a ten year period reveals that the frequency in which the maximum draw down – or values close to it – are reached is actually extremely high, something which means that a trader using this may be subject to strong psychological pressure.

When you consider the possible variations caused by broker dependency it also becomes clear that draw downs that might be closer to the Monte Carlo WC scenario might be reached a few times during a ten year period, pointing out that trading such portfolios is bound to be very difficult. Another thing is that the reaching or not of a deep draw down can depend on a few trades meaning that missing trading days on a setup like this can be critical since missing the “profitable side” of such a volatile portfolio can easily cause a draw down to double over the course of a bad trading period.

In the end it seems that trading a “big portfolio” has an upside (very large AMR to max DD ratios) which is compensated by the fact that the portfolios have a huge ulcer index when compared to smaller ones. If you reduce the Ulcer Index to make the trading of these portfolios as “easy” from a psychological perspective as a smaller one then you get a trading setup which has overall similar results. That means that if you want to trade a large portfolio without getting a stomach ulcer you might need to reduce your risk to the point where your reward will be similar to that of a smaller portfolio. Of course the benefit of higher diversification would still prevail but your capital requirements will be much higher than for a smaller portfolio (because the same capital is “allocated” in smaller portions as the number of systems grows).

In the end I believe this is a quite important discovery because it answers some of the important questions we have had about running “big portfolios” and why – if such high AMR to max draw down ratios are possible – no one seems to be trading in this way. The answer is that such portfolio have less certainty regarding their maximum draw down values, have a tendency to go into sharp draw down and recovery periods, are sensitive to the missing of trades and – because of these facts – have a tendency to drill an ulcer into its owner’s stomach! When large portfolios are traded with high profit targets it seems to be warranted that a very high Ulcer Index will be generated.

If you would like to learn more about my journey in automated trading and how you too can earn an education on this field, simulating and building portfolios for your particular risk aversion and trading style please consider joining Asirikuy.com, a website filled with educational videos, trading systems, development and a sound, honest and transparent approach towards automated trading in general . I hope you enjoyed this article ! :o)

Print Friendly, PDF & Email
You can leave a response, or trackback from your own site.

11 Responses to “The Effects of Increasing System Number: The Deterrents of Big Portfolios”

  1. McDuck says:

    Daniel,
    I think there is a related question which must be also addressed. What’s the difference between i) running 5 small portfolios in 5 different brokers with respect to ii) run these 5 portfolios in the same broker?.

    Case i) is what is seemingly done currently by many members, with accounts opened around the globe. Case ii) is what you deal with in this article.

    Apart from debatable compensation of broker dependency in case i) I don’t see why case ii) is essentially different.

    Best.

    • admin says:

      Hi McDuck,

      Thank you for your comment :o) I believe that the question here is not whether or not you use different brokers but whether or not you run them on the same ACCOUNT. When you run several small portfolios on different accounts (whether on the same or on a different broker) there is no problem as Ulcer Index values are small and broker dependency and other such issues are not that important. The problem is generated when you run a large amount of system son the SAME account as they generate a portfolio with a very high Ulcer Index. It is NOT about whether or not you have many accounts but about how many systems you run on a SINGLE account. The best choice – which is what I do – is to have several accounts for each portfolio across several brokers (to compensate for broker dependency) but this has nothing to do with the above article which deals with the size of portfolios run on each DIFFERENT account (how much systems are run on a SINGLE account). I hope this clears it up!

      Best Regards,

      Daniel

      • Franco says:

        Hey Daniel,

        As I see it, running a big portfolio of say 8 systems with all eight using a risk percentage of 1% will give the same risk properties as using one system with a risk percentage of 8%.

        Although the portfolio’s risk is more equalised over different systems, more or less the same drawdowns/Ulcer Indexes will be expected for both.

        What makes more sense to me is to use eight different portfolios, divide the risk of each portfolio by 8, and run them on eight different accounts.

        Hope what I just said made any sense :D

        • admin says:

          Hi Franco,

          Thank you for your email :o) That assumption about portfolios and single systems is not true (to the best of my knowledge), try running an 8 system portfolio with a risk 1 and then any of its systems with a risk 8. You will notice that in all cases the yearly profit to draw down ratios will be MUCH smaller on the portfolios. The average compounded yearly profit to maxim draw down ratio improves dramatically for portfolios against risk increases on single systems. The best way to trade systems as far as I have been able to see is to run large portfolios of many instances with very reduced risk levels (what large money managers often do) these means that you have the same Ulcer Index as a smaller portfolio with much higher diversification. Obviously this increases capital requirements greatly as I mentioned on the article.I hope this makes sense :o)

          Best regards,

          DAniel

          • Franco says:

            Hey Daniel,

            Thanks for reply, your time management is something out of this world :) There is a new series on Discovery/History channel called Superhumans, think you should give them a call for an interview :)

            Anyways I agree with you completely, think I did not communicate my thoughts accurately enough. My assumption is wrong though as you said….

            Regards
            Franco

          • admin says:

            Hi Franco,

            Thank you for your reply :o) Superhuman? Lol! Thank you very much for the compliment :o) I regard myself as being efficient but nothing beyond that ;o). This year’s projects will require some “super human time management skills” but thankfully I have a great Asirikuy member team that helps me leverage my very “sub-super” condition. Thanks again a lot for posting !

            Best Regards,

            Daniel

  2. Corezo says:

    Daniel..you have pointed out comprehensively in several posts the possible benefits and drawbacks of trading big portfolios. I will list your main points and possible ways we could perhaps approach account management in these circumstances. Some of these suggestions probably go against what you would do in your own account.

    1. Margin and execution requirements can become an issue after an account reaches a certain balance (ie $100K+) – could we withdraw profits and start up a smaller account, perhaps with a smaller risk..and repeat the process.

    2. Broker dependency – Once our accounts grow large enough..why not pool accounts and open up an institutional account at a prime broker. A bank feed from Barclays Capital or Goldman Sachs, or even a second tier broker like London Capital Group certainly will give Asirikuy traders direct access to Currenex or Hotspot feeds. Large portfolios will always increase risk of broker dependency, but if we reject the use of retail brokers altogether, we will be trading from the same feeds as hedge funds and traders with very large capital.

    3. You might get an ulcer – drink a concoction made of licorice..it will help soothe the pain :)

    • admin says:

      Hi Corezo,

      Thank you for your comment :o) Well let me comment on some of the things you have said:

      1. Well I would in fact say that trading at quantities around 100K+ is no problem since most Asirikuy systems are not very leveraged and in most cases positions will not reach more than 1 million (which is easily filled in Forex). Of course we would need to start implementing some things to do partial fills, etc around low liquidity times but I do not think that 100K is a high limit for retail.

      2. Absolutely agree :o) Asirikuy is starting to look like something which will become like a hedge fund in the future. Looking at the capital currently traded by members it might be in excess of 1 million so would it make sense to build a corporation and put everything together into an account we can trade with Goldman ? Stay tuned for some about this on tomorrow’s post! You’ll see that we first need to walk some intermediate steps which I will be discussing across the following few weeks.

      3. I have now reached a state of Zen-like trading which I hope will help me avoid an Ulcer :o) However I will keep some licorice just in case ;o)

      Thank you very much again for your post,

      Best Regards,

      Daniel

      3.

  3. Forexman1972 says:

    Hi Daniel,

    Thanks for your article. I have been questioning the issue of trading a large portfolio on the forum recently.

    Am I missing something here? If I combine 5 portfolios into 1 large portfolio on the SAME account it will have a high Ulcer index. Yes I agree with that.

    However if I trade the 5 portfolios individually on separate accounts then the Ulcer index will be small for each portfolio.

    But…surely the Ulcer index is additive for each portfolio? If you look at the 5 porfolio’s combined you will still see a similar overall drawdown (just on different accounts).

    What am I missing here?

    Skdoust

    • admin says:

      Hi SkDoust,

      Thank you for your comment :o) The issue here is capital. If you trade 25 systems on one account you only need X capital and will have a very high Ulcer Index, if you however divide them in groups of 5 and trade them in 5 accounts you will need 5X capital and therefore you will suffer the same overall draw downs but across a total of at least 5 times the capital (each portfolio is working a SEPARATE amount of X capital). This is the same as if you traded all the systems on one accounts with 5X more capital and 1/5th the risk (each system is working on 5X capital but with 1/5th the risk, therefore you are experiencing the same draw downs you would on X capital although now they represent a much lower percentage as your capital is 5X). As I highlighted on the post the ideal way to trade large portfolios is to drastically reduce risk to make the Ulcer Index lower but this greatly increases capital requirements. I hope this helps :o)

      Best Regards,

      Daniel

  4. erick says:

    This article and your other article analyzing and discussing Big portfolios are excellent and important. Would like to see an ebook on this subject.

    The Big question is: “if such high AMR to max draw down ratios are possible – why does it seem that no one is trading this way?”

    That is the Question. Either there is some fatal flaw with big portfolios or they are unable to be configured optimally – maybe due to a lack of algorithms that adequately diversify each other.

    Consider a successful discretionary trader. One trade, one win most of the time. No matter how good he is, he does not have the stacked/additive returns and diversification of a portfolio. Any Asirikuy trader with a smart six algo portfolio will leave this successful trader in the dust.

    Real world algo traders closely match the AR/DD of discretionary traders – why is this? Are they using single algorithms, rather than configured portfolios. This seems unlikely.

    The big question is: Why do we not hear of consistent 100+% portfolios with low maximal drawdowns? And: Can the reasons be overcome?

    Regarding the answers to the big Question above: “less certainty regarding maximum draw down, sharp draw down and recovery periods, sensitive to missing trades, high capital requirements, and broker dependancy,” it seems to me that the first two reasons can be dispensed with by configuring the portfolio to have an optimal AR/DD with a low Ulcer.

    A fifteen+ algo (ten-year backtested) portfolio optimally configured should be able to have an Ulcer close to 20. One just has to have the right assortment of algorithms. I do not believe that a large algo portfolio must have a large U. It would be helpful to have a summary portfolio statistic like AR/DD/U.

    I do not see the need for a higher (dollar value) capital requirement since lot size adjusts according to the global balance. And if the dollar capital requirement does not need to be higher, the individual missed (smaller lot) trades of a big/low U portfolio will have relatively smaller losses, which can be ignored.

    Since there are multiple articles on broker dependancy, it must be an important component of risk, but I think I do not understand this.

    Dependancy seems to mean that different brokers’ price series make one less confident about the validity of the portfolio backtest. But would not different price series have an equal probability of benefiting or hurting the portfolio’s statistical profile; are they not simply a different MC simulation with an equal chance of benefit or loss? Could this not be managed with a forward test that switches to brokers with consistently “good” dependancy? Is the direction of broker dependancy consistent? Is dependancy risk truly material?

    My idea is to configure a Big portfolio to max AR/DD/U subject to a reasonable maximum total risk capital requirement percentage (which is not the absolute/dollar capital requirement mentioned above); this capital requirement percentage seems to be the most critical risk component of a Big/Low U portfolio.

    I always wonder if a statistic is good enough – are there any benchmarks? A ten-year backtest 20U portfolio is better than a 30U portfolio, but is a 20U portfolio good enough? Can we live with it, or must we reconfigure down to 15U? Is 15U psychologically tradeable?

    I have no basis (does anyone?) for these arbitrary ten-year backtest portfolio numbers, but I try for a 12 EA or less portfolio, AR/DD greater than 6, U less than 20, maximal DD of less than 25%, a Total maximum risk percent less than 25%, and an AR greater than 150%. I think these numbers are possible but do not know if such a portfolio would be psychologically tradeable.

    But the Big question remains: Why does it seem that no one is trading this way? I think I may have a basic and fatal misunderstanding of Big portfolio risk.

Leave a Reply

Subscribe to RSS Feed Follow me on Twitter!
Show Buttons
Hide Buttons