What investment strategy would have given you the highest returns in Europe over the 13 year period from July 2000 to July 2014?

In this article I summarised the master’s thesis of Andreas Hennes (completed at Goethe University Frankfurt am Main on 14 September 2015) where he set out to test exactly that.

Get The Timeless Reading eBook in PDF

Get the entire 10-part series on Timeless Reading in PDF. Save it to your desktop, read it on your tablet, or email to your colleagues.

Results and findings

Before I get to all the details here are the summarised results:


Click image to enlarge

  1. There is a value premium in the European stock markets. This means value (undervalued companies) outperformed the market and glamour stocks (expensive or highly valued companies)
  2. Earnings / Price generated the highest return among the single-factor strategies
  3. Despite a lower geometric mean return of 10.9% Shareholder Yield had the highest Sortino ratio of 0.77
  4. The annual average value premium compared to the market ranged from 3.8% to 5.9%
  5. The annual average value stock premium compared to the glamour stock deciles (value companies did better than glamour companies) ranged from 11% to 17%
  6. The value deciles of all the multi-factor strategies outperformed their respective glamour deciles
  7. These results confirmed that previous research studies (mainly US studies) that investing in undervalued companies also works in Europe
  8. Combining value and momentum (the best strategy) to find undervalued companies that are already recovering (share price moving up), is a successful tool you can use to separate winners from losers
  9. Reducing portfolio diversification to 25 companies can increase your returns.

Why did he do the research?

I like asking this question to see if the researcher may have any bias that may influence his or her results.

For Andreas it was a requirement of his master’s degree which he wanted to finish (and get a job afterwards) so he had to write a good paper. It’s very unlikely that he had a bias he wanted to prove.

What did he want to find out?

Andreas wanted to answer the following questions:

  1. Given that a lot of studies about the value premium only looked at the US stock market he wanted to find out if value strategies also outperformed in the European stock markets?
  2. What single- or multifactor value investment strategies were successful over this period, that is, had better risk-return properties compared to glamour stocks (companies that are expensive or over-valued) and compared to the market?
  3. Can these strategies also be used by institutional investors or are the results only theoretically achievable?

To do this he analysed the performance of 11 quantitative value investing strategies in Europe over the 13 year period from July 2001 to July 2014.

Market performance - basically flat with a lot of volatility

This is what the STOXX European 600 index did over the period of the research paper:

STOXX European 600 price

STOXX European 600 price index Source: bigcharts.marketwatch.com

As you can see it was a period with a LOT of volatility but the index did not go up much – it only went up about 3.6% over the 13 year period.

His “market” looked a lot different

The results of this research paper look a lot different because financial companies (which completely collapsed during the financial crisis) were excluded and dividends were reinvested.

In the paper the average return of the market over the 13 years was 7.5% and the geometric average 6.0%.

This means €10,000 invested in the market would have grown to €22,675 over the 13 years. Over the same 13 year period the maximum market drawdown was 55.9%.

What time period - Jul 2001 to June 2014

Andreas put the first portfolio together on 1 July 2001 and tested returns up till 30 June 2014. Portfolios were formed on a yearly basis (holding period one year).

Over the 13 year period he tested 11 investment strategies using 143 yearly formed portfolios.

The back test investment universe

He used the STOXX Europe 600 a stock market index instead of all the listed companies on the selected stock exchanges as the investment universe.

When he did the study the STOXX Europe 600 Index had a fixed number of 600 companies that represented large, mid and small market value companies from the following 18 European countries: Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and UK.

His reasons for choosing this index were:

  • It covers 18 stock markets of developed European countries. Accounting for roughly 89% of the total market value of the European stock markets.
  • The availability of index constituent’s lists allowed him to exclude survivorship bias  (this is VERY important), as the exact composition of the index is known for the complete back test period. If he used only the current constituents’ list, as is common practice, it would have led to a survivorship bias, because returns from companies that were delisted during the back test period due to failure, acquisition or going private would not be taken into account
  • The index allocation of 200 small, midsize, and large companies ensured an equal exposure to all sized companies and the back test is not dominated by large companies.
  • To be eligible as for the STOXX Europe 600 index, a company must have a minimum average daily trading volume (ADTV) of USD 1.5 million, even for small companies. This is an important point to make sure all strategies can also be implemented in the real world, also by institutional investors.
  • Due to changes in the index during the back test period, 1141 companies were included in the index.

Companies excluded

Financial companies were excluded from the back test universe because their highly levered balance sheets distorted many ratios that were part of the analysis.

This reduced the universe over the back test period to 884 companies.

The data source

Fundamental and daily price data were obtained from Thomson Reuters Datastream.

In the case of different currencies than the Euro for fundamental data (UK companies for example), the accounting data was converted using the respective exchange rate of the day of portfolio formation.

All returns in Euro

Because Andreas aimed his research at European investors - all price and dividend data were converted to Euro.

The daily total returns were calculated for all companies in the investment universe in the period from 1 July 2001 until 30 June 2014.

Returns for companies that delisted were calculated based on their actual returns. These intra-year returns were only invested on the next date when portfolios were formed.

Portfolio construction

For the calculation of accounting ratios fundamentals are lagged by six months to prevent look-ahead bias.

This is done because companies release their financial reports several months after their financial year end. This approach is conservative as other researchers have lagged results only three or four months.

This means that, for companies with a December year end portfolios were only formed in July using December year-end financial data.

Invalid companies excluded

Companies with invalid ratios were excluded from the investment universe for that year.

Ratios were invalid if there was at least one missing data point or if the value did not pass certain plausibility criteria. For example, companies with negative book value, sales, dividends or market value.

Each year, for every investment strategy tested, decile portfolios and a benchmark portfolio was created.

The benchmark (market) portfolio consisted of all the companies for the respective year that had all

1, 23  - View Full Page