Academics and investment institutions have published many papers in the recent years about financial markets and how to outperform the equity market average (benchmark).
Some of the most notable papers that have attracted our attention are Momentum Investing from London Business School (LBS), Seasonal Effect in financial markets by Ben Jacobsen from Massey University, and some research papers and books published by the major investment institutions (e.g. Standard & Poor’s).
Some of these documents demonstrate that it is possible to consistently outperform the market in the long-run and they were prepared by some of the brightest minds of the financial industry with the highest financial skills. Those which most appeal to us have in common the use of quantitative analysis.
Benjamin Graham[1] said: “I should greatly welcome an effort by security analysts to deal intelligently with speculative operations. To my mind the prerequisite here is for the quantitative approach, which is based on the calculation of the probabilities in each case, and a conclusion that the odds are strongly in favour of the operation’s success”.
We favour the use of a quantitative approach rather than methodologies that have been proved not to work. We believe that a quantitative approach is more rational and objective than any other approach that has not been back tested. Because we completed the first part of our research/modelling into the US equity markets, we have decided to compare our findings with those of the above mentioned papers.
The challenge
Although future performance cannot be predicted accurately, one of the challenges of our approach is to demonstrate that a sector rotation approach based on the Intermarket analysis can achieve a long-term capital growth (~18%) by far more than the most well-known benchmark in the world, the S&P 500 Total Return (~8.9%).
However, the greatest challenge for our models is to compare them with well-known top performers such as Warren Buffet (~21% since 1980, assuming we can benchmark Warren Buffet’s performance to Berkshire Hathaway) and the LBS Momentum investing model (~17.8% from 1980 and ~15.2% from 1900).
The results
The results of our research/model demonstrate that investing in sectors or industries is by far more profitable and less risky than stock picking a few companies. The evidence suggests that buying a whole sector of the economy is less risky than buying a few selected shares.The extra transaction costs due to the higher number of shares to be bought and sold (almost zero using sector ETFs) are more than compensated by short, medium and long-term extra premium.
The long-term returns of our models are highly competitive with the LBS model, Warren Buffet and most of the top performing Hedge Funds.We have developed four core models: Business Cycle, Intermarket, Ranking & Timing and the Conservative models. All the models are rotation models but they have four different types of buy and sell signals. The models are invested in sectors/industries, bonds and commodity ETFs.
Sector Rotation
There are a lot of charts available on sector rotations (i.e. Merrill Lynch’s investment clock and S&P’s sector rotation chart) but we have not yet found any study which explains how investors should rotate their money, what the drivers are, what the performance of a model would look like and mainly the timing of entry into and exit from the market.In our research we have used what we would call the “drivers” of the Business Cycle borrowed from the Intermarket analysis and we have created our timing sector rotation models. The results of our models are based on the entire period from 1900 to 2011. However, when we developed our models we had to consider from 1926, 1936 and 1942 to date only due to the lack of data on sectors and industry groups prior to those dates.
Some numbers
One of our back tested models has 80% probability of outperforming the benchmark within 18 months, 90% probability within three years, 95% probability within four years. On those few occasions that the models under perform the benchmark, they do so for a short period of time with a small negative premium. One of the sector rotation models has 100% probability to beat the benchmark within five years. Most of them have 95% within four years.When performances are measured in absolute returns, the results are even more impressive with some models being able to achieve a positive return within 12 months with 95% probability and a compound annual rate of growth of ~18% since 1942.
The four models compound annual rate of growth ranges from 12% to 20%. However, there are models that, when invested in industries rather that sectors, can achieve a 25% compound annual rate of growth.Most of the models have an annualised risk-adjusted return of between 0.9 and 1.5 and some of them have a historical maximum drawdown of 10%-12%. This means that at any point in the past 70-80 years the value of the portfolio has never gone below 10%-12% of its peak value. All the models consider end of month prices only and transactions are assumed to be executed at the close of the last trading day of each month.
Back testing methodology
All the above is based on past performances from 1926, 1936 and 1942 to date and clearly there is no absolute guarantee it is going to happen again. However, while some models are a very good example of data mining (i.e. what would have been the best asset allocation in the last 70 years?) and can be used only as an indicator for future performances, other models are robust and have been properly back tested.For example the data has been modelled from 1926 to 1980 and back tested from 1980 to 2011 or modelled from 1942 to 2000 and back tested from 2000 to 2011. Those results are available on our website pages with equity curve data available in a spreadsheet for your own analysis and for the sake of transparency.
The evidence suggests that our sector rotation models are less expensive than the LBS Momentum investing model which, for several reasons, is a good benchmark for comparison. Firstly, it comes from one of the most prestigious institution in the world. Secondly, the model covers more than 100 years of history. Thirdly, its performance is similar to that of the best Hedge Funds.However, in the LBS momentum investing model, there is an high turnover of shares (the authors admit that transaction costs can seriously compromise the performance), and sometimes momentum strategy can earn negative returns for five to seven years and under perform the benchmark for much longer.
Conclusions
Sector rotation works as an investment strategy firstly because it is more likely that the same sector or industry performs similarly in the same economic conditions. Secondly, there is a business correlation between most of the sectors/industries of the economy. Bonds and Commodities are then added as a hedge to economic fluctuations and financial market volatility.Our quantitative and probability based approach helps to map the relationships between asset classes and to allocate money to the right sector/industry at the right time in the right quantity at any stage of the Business Cycle and thereby we believe that the results obtained with our models are able to enhance the manager’s fund performance, return them extra performance fee and attract more capital into their funds.
[1] Graham is considered the first proponent of value investing. He began teaching at Columbia Business School in 1928 and subsequently refined with David Dodd through various editions of their famous book Security Analysis. Warren Buffet is the most famous investor who follows this approach.