If there in the world of trading and investing exists something like a free lunch, it is clearly and solely diversification. Essentially, there is nothing so extensive, extended and mathematically described as in the field of classic investments as portfolio theory. There are even studies that are trying to prove that success in investments of many investors (among the best known belongs for example Warren Buffet) is rather a coincidence than a high standart of skill.
In finance there is constantly present a considerable degree of uncertainty and randomness that it is entirely inevitable that one of the thousands/millions investors will be exceptionally lucky. Such a perspective on a world of finance is held by a mathematician Nassim Taleb who described his sceptic view on finance more in detail in his books.
Possible evidence of this fact is that Buffet‘s success has basically never been repeated by any other investor. That is the reason why most of the investment approaches focus less on the selection of instruments that are guaranteed to earn 30 % per year on the initial investment for 50 ears in the row (since obviously it is not simply possible), instead they focus on the correct portfolio diversification so that such portfolio is the least volatile and as a whole is at least as good as (comparing for example annual returns and drawdowns) as the entire market – for example a basket of US stocks in the S&P 500 index.
Basic investment «truth» which is based on the efficient market theory is that «one can not beat the market». Of course, some investors will have better results than the market, the other part of them will do worse. The problem is that no one knows in advance into which group they will finally belong. The proof of it is the performance of classic mutual funds, which year after year underperforms a broad index because of management fees. Thus it seems that if we do not have a crystal ball the best we can do is to diversify and be moderate – just like the market.
Why do I write this introduction in the spirit of investments instead of trading? Most retail traders were (about the same as I was) not satisfied with the returns of classic equity portfolios (beacause the average annual income is typically much lower than drawdowns) and they started to become interested in other ways how to make money – the active trading of stocks, commodities, currencies, indices, shares etc. with expectations of an fabulous annual yield with minimal drawdowns.
So common retail trader leaves the world of equity investments where the only activity is to properly buy a diversified portfolio of stocks and wait. Instead he runs into the world of seeking his desired market, timeframe and strategy. If he manages to do this correctly he will without a doubt become a millionaire. That is at least what some cheap books and websites claim.
Finding a suitable market, timeframe and strategy, however, becomes a search for the Holy Grail. Some traders are naive and copy a srategy designed by some strategy guru and starts to trade it without any further study and verifications. He can be lucky for a certain amount of time and end up in a profit. Then the trader gains a very high self-confidence about his abilities, gets too leveraged and sooner or later will lose everything. If he is unlucky he will experience losses immediately and will end his trading activities forever.
The not so naive trader takes a different approach — studies all the possible approaches, educates himself in finances, maths, statistics, economics and programs and backtests all kinds of strategies with the help of trading platforms.
One can also create strategies in a specialized sotfware, which thanks to genetic algorithms is able to find lots of strategies with nice equity curve in a short period of time with. Trader then expects that one of the millions of strategies will at least function – if applied on the right market and timeframe.
The fundamental problem of this – still a naive approach — is that there are still considerable risks that the discovered strategy is nothing but a cluster of meaningless rules that due to a completele coincidence and a significant number of attempts of finding the Holy Grail generated some stunning historical returns.
Such a strategy is completely not suitable for live trading. In short, historical returns do not guarantee future returns. A trader who is trading a single one strategy will most likely experience a reverse to mean – an actual performance of a strategy which is in a huge contrast opposed to what he expected from the historical results.
Setting up one or a few strategies into live trading is for so many reasons considerably risky. The greatest risk is, of course, a described fact that an impressive historical performance does not necessarilly indicate anything about a future behaviour of the strategy. This topic is covered by many many good books and articles. It is an area so well known that I will not describe it into further detail.
But if we believe that neither historical performance nor statistical tests do not necessarily mean future good performance, what are we actually looking for? In my opinion it is necessary to add another factor which is a sense and a logic of a traded strategy. Hence my request for a good strategy is primarily about what, how and why it trades. Do we trade a trend or, conversely any significant deviation from the norm or some fundamental factors? Is the logic of a strategy justifiable in a real world? Only then I am interested in a historical performance.
Does that seem like some kind of philosophizing without any grater importance? For me the decisive factor is neither a historical performance nor a beatiful equity curve, but a future performance. Future performance cannot be forecasted by any method available, we can only speculate and hypothesize, probably badly. Under these circumstances it is necessary to rely primarily on the fact that the traded strategy has some sense to it, the environment with regard to strategy fundamentals has not changed and that is why we can expect a future performance similar to the historical.
The paradox is, and that is the whole point of this article that sthe strategies that have such a sense to it and we can expect some future performance, often does not perform very well in the backtests. The resulting equities resemble a roller coaster ride, whole years are accompanied by losses and even in some cases the losses are long term.
If we look at the most widely used and familiar strategy in financial sector – a trend following system, by definition it is clear that such strategy works best if there is a trend present in the market and will perform poorly if the trend is not present. In some markets the trend may not be present for decades and this strategy can lose money year after year. Would we stop trading such a strategy and shut it down because we lost money on it? Yes, if we have a reason to believe that in the current market conditions the trend will not be established in the near future. Can we predict that? In some cases yes (of course there exist some markets which by its nature are not trending that much), in most cases we do not know. However, trading such a strategy alone is a bare foolishness. If, however, we manage to compile a portfolio of many uncorrelated markets, and run there the same trading strategy, our chances of success will increase. Eventually every large hedge funds are doing so. Trading of a diversified portfolio is here as well as in the field of classic equity investments that above mentioned free lunch. We will not worry that much that some strategies on some markets are doing poorly at some time, other strategies are in fact smoothing out these periods of drawdowns. The whole portfolio will demonstrate a smoother equity line.
However it is not just a matter of trading one strategy on multiple marktets. The aforementioned trend following has been showing an enduring stagnation or even a loss in the last 5 years, even though there are several dozens of futures contracts traded. In order to be profitable during these kind of years we have to trade other strategies in parallel – for example mean reversal on stocks, selected currencies or market neutral strategy options. By combining all these inherently very different strategies we can expect further improvement of efficiency and further smoothing of equity curve. In short, in the project QuantOn Solutions (asset under management company) where I am co-founder we have to be prepared for any market conditions because our benchmark is not some equity index like S&P500 or some other fund, but simply an absolute number – the net annual yield.
The reasons why we as an asset under management company must trade broadly diversified portfolios of markets and strategies were described above. In short, we do not think that we can predict the market or pick currently hot stocks which will rise 100% in one year since the day of our purchase and we will successfully sell at the top. We also do not think that it is possible to a find a strategy which historically showed perfect results with sharply increasing equity and so it will perform the same way in the future.
Conversely, it appears that functional strategies which have long been profitable are not that attractive in historical simulations on different markets (especially in terms of volatility). From such historical results it can even be difficult to determine whether the profits of a tested strategy are statistically significant or not. However we can get a partial confirmation when we apply such strategies on a long historical periods of time (for example 15 years) and on multiple markets with the similar expected dynamics of behavior at the same time. In the case of testing the strategy in one market we get the results for the period of 15 years. In the case of testing the same strategy for 50 markets we get results for 50*15 years = 750 years. It is not quite true but the principle is the same. Thus the statistical results are more credible.
To compile a broadly diversified portfolio we need relatively complex tools. Most of the commercially available platforms does not offer a possibility of portfolio testing. Some exceptions we know represent — Tradestation Portfolio Maestro and Multicharts Portfolio Trader. Both programs include basic options for building a and tetsing of the portfolio. Both tools on the other hand have their limits. For example it is not possible to simply set up your own dynamic position sizing model. By that I mean for example a limit to trade just some strategies which meet certain parameters (depending upon the others). Other limits concern the inability to take into account the size of the positions opened according to correlations between marktets. Not to mention the testing of an options portfolio which is completely impossible. Furthermore there are other limits of the software, especially in the field of automated export of outputs and archivation of outputs, tests, randomization, comparing the results etc.
Last but not least there is a problem with live trading. Portfolio Maestro does not offer it at all and Multicharts Portfolio Trader offers it to a limited extent. Therefore we are currently evolving our own approach how to deal with those execution issues.
Previous: 10 Golden Rules of a Successful Trading