Potential for generous excess returns in Inderes recommendations
Source: Inderes *The Inderes recommendation portfolio is constructed by including only the stocks with a Buy and Accumulate recommendation from Inderes, with a double weighting for the Buy recommendations. The assumptions are explained in more detail later in the article.
The return of an investment portfolio relative to an index is a simple but brutally honest measure of stock picking success. A similar measure, we believe, will also serve as a tool for assessing the quality and independence of equity research. It is easy to question the usefulness of equity research to the stock picker if equity analysts fail to pick investments with a good risk/reward ratio and to avoid those with a poor one. This is probably true, although research coverage in itself does have inherent informational value for the investor.
For this reason, the Inderes analysis team has long been tracking the returns generated by analysts' recommendations internally, using calculations from Bloomberg, among others. For the first time, we have now compiled for publication a data analysis of the returns of our recommendations for the period between 1/1/2013 and 12/31/2024. Inderes analysts, at least, have not yet lost their humanity and also make mistakes, so we think it is a reasonable starting point to track the accuracy of our team across the range of all the companies we cover.
Based on our data analysis, the weighted portfolio of Inderes equity research Buy and Accumulate recommendations ("Inderes Recommendation Portfolio") has returned 30.8% and 23.2% per year (CAGR) over the long term (2013-2024) and medium term (2020-2024), respectively, taking into account dividends. Moving the reaction to the portfolio's recommendations forward by one day (trading takes place after the "Inderes effect" that follows a change in recommendations), the annual returns are 22.6% and 14.3%, respectively. Adding in the trading cost assumption (1% per trade), the annual returns are 17.6% and 9.7%, respectively. Over the same periods, owning all the companies covered by Inderes would have produced corresponding annual returns of 10.5% and 4.0%, which all the above-mentioned portfolios built on the basis of Inderes' recommendations have far exceeded.
Source: Inderes *The Inderes recommendation portfolio is constructed by including only the stocks with a Buy and Accumulate recommendation from Inderes, with a double weighting for the Buy recommendations. The assumptions are explained in more detail later in the article.
The difference in returns becomes concrete when looking at the portfolio's change in value. In the long term (2013-2024), a portfolio of EUR 10,000 would have grown to around EUR 250,000 with a strategy based on the Inderes recommendations, and to EUR 70,000 after deducting the Inderes effect and trading costs, while owning a balanced index of all covered companies would have increased the value of the portfolio without costs to around EUR 33,000. In 12 years, the difference in returns would have increased by a factor of 3 to 10 in favor of the Inderes recommendation portfolios.
While the Inderes analysts' recommendations seem to have been of considerable value in the long term, there was also one disastrously bad year of 2022, when all the stocks we followed, regardless of the recommendation (Buy, Accumulate, Reduce and Sell), made a loss after the index of covered companies fell by 24%. Nevertheless, even in the short term (2022-2024), the portfolio built on recommendations has outperformed the index of all covered companies. However, this does not necessarily make the investor feel better, because after adjusting for the Inderes effect, or after adding in trading costs, the portfolios have still made negative returns over this period (albeit less than the covered companies as a whole).
Source: Inderes *The Inderes recommendation portfolio is constructed by including only the stocks with a Buy and Accumulate recommendation from Inderes, with a double weighting for the Buy recommendations. The assumptions are explained in more detail later in the article.
The Inderes recommendation portfolio has also generated significantly higher annual excess return for corporate customers paying for the research (2013-2024: Recommendation portfolio +35.1%, all corporate customers 11.0%) than for non-paying companies (2013-2024: Recommendation portfolio +12.8%, all non-paying companies 10.3%). At least from this perspective, we have been able to provide high-quality equity research that serves investors, regardless of whether a company pays Inderes for research coverage.
Below, we explain the methods used and the logic behind them. We have also included a downloadable summary of our recommendation history at the end of this article to make it easier for investors to review. If the figures or methods raise questions or criticism, we have a thread on our Finnish forum where things can be discussed further.
How returns are calculated: We build a daily rebalanced portfolio of recommendations
Changes in the value of the portfolios we create are calculated by rebalancing the position weights on a daily basis and assuming that the portfolio is always fully invested. In practice, portfolios live with stock movements, so this is a simplification. However, Inderes' recommendations are updated frequently (2013-2024: 100–300 recommendation changes per year), so we assume that the portfolio is rebalanced whenever there is a recommendation change that affects the portfolio. The calculation does not take into account exchange rate changes.
We first build portfolios containing only stocks of a particular recommendation to get an idea of the overall accuracy of the recommendations. Inderes’ recommendation policy currently includes four recommendations: Buy, Accumulate, Reduce and Sell. We use a risk-adjusted expected return in our recommendation policy:
- Buy - The 12-month risk-adjusted expected shareholder return of the share is very attractive
- Accumulate - The 12-month risk-adjusted expected shareholder return of the share is attractive
- Reduce - The 12-month risk-adjusted expected shareholder return of the share is weak
- Sell - The 12-month risk-adjusted expected shareholder return of the share is very weak
In addition, in rare special situations (e.g. competitive bidding in tender offers), there have been Hold recommendations, but these play a minor role in the overall picture. These recommendations have been in place throughout the review period, although the policy has been updated along the way.
Source: Inderes
Over a longer period of time, the returns of the recommendations have remained in line, as the returns have progressively decreased with the recommendations (Buy > Accumulate > Reduce > Sell). At the same time, all returns, regardless of recommendation, have been lower in recent years, in line with the weaker returns of all covered companies over this period. Years are also not made equal, and in some years the returns are more in line with each other. A particular blemish for the recommendations was the year 2022, when the shares with Buy and Accumulate recommendations also posted losses.
Source: Inderes
Of these recommendations, Buy and Accumulate are "positive" and represent the analyst's view that the stock is worth having in the portfolio. Similarly, Reduce and Sell recommendations are "negative," and the analyst at least questions whether it is worth buying more shares. On the one hand, a Reduce recommendation is often given to a stock with a positive expected return when the expected return is not sufficiently attractive on a risk-adjusted basis (typically the expected return is below the cost of capital). A Sell recommendation, on the other hand, is stronger in its wording and is generally given in a situation where the stock's expected return is negative.
Based on this interpretation, we create illustrative portfolios based on the following principles:
|
|
Inderes weighted long portfolio ("Inderes recommendation portfolio") |
Inderes weighted long/short portfolio |
Inderes index of covered companies |
|
Portfolio weightings |
|
|
|
NB! "Long" corresponds to ordinary ownership of a share, where the investor benefits from the share's positive return. "Short" corresponds to short selling, where the investor benefits from the negative return of the share.
Source: Inderes
We believe that the returns calculated for these portfolios provide a good measure of the accuracy of our analysis. Both recommendation-based portfolios have outperformed the "index portfolio", which includes all covered companies, in the short (2022-2024), medium (2020-2024) and long term (2013-2024).
In terms of the accuracy of the recommendations, all recommendations, including Sell and Reduce, are relevant. As a general rule, it would be valuable for an investor to be able to keep stocks with a lower expected return out of the portfolio. In the case of small companies, however, it may be virtually impossible, or at least very difficult, to short sell their stocks (= an investment that pays off for the investor if the stock declines in value) in order to take a negative view.
Therefore, we believe that a weighted long portfolio, where one simply selects the stocks to own and their weightings, is a more realistic portfolio construction principle from an investor's point of view. The returns of this weighted long portfolio ("Inderes recommendation portfolio") have been better than those of the weighted long/short portfolio, i.e. in practice Inderes Buy and Accumulate recommendations have been able to generate better returns than shorting the Sell recommendations would have done.
What about the Inderes effect, which can prevent one from buying a stock at the price prior to the recommendation being updated?
That's a good point. Inderes analysts update recommendations when the market is closed, so when the market opens the next trading day, the stock price may react to the recommendation before the investor has time to trade the stock (the so-called Inderes effect).
For this reason, we have also calculated the returns of the above weighted long portfolio by delaying the effective date of the recommendation by one business day. In this way, the return on a recommendation is calculated from the closing price of the stock (including any dividends) after the recommendation has been in effect for one trading day, allowing the investor to trade at that price. This change moderates the profitability of the strategy but only takes away a portion of the excess returns.
Source: Inderes
The effect is also visible in the overall recommendations, since for all recommendations, the correction for the "Inderes effect" moderates the accuracy of the recommendation. Thus, in practice, for Buy and Accumulate recommendations, the market has reacted positively on average on the first trading day after the recommendation, while for Reduce and Sell recommendations, the market has reacted negatively on average on the first trading day.
Source: Inderes
What about the costs of constantly tweaking the portfolio?
That's another good point. From an investor's perspective, it makes sense to assess the rationality of stock picking in terms of total return after costs, since index investing through funds and ETFs is typically low-cost. There are ways to manage capital gains taxes, for example through share savings accounts (ISK/OST) and investment insurance setups, so we only assess the impact of trading costs here.
In a weighted long portfolio (Buy 2x Long, Accumulate 1x Long), the number of recommendation changes affecting the portfolio is quite high (2013–2024: 75–250 recommendation changes per year), so we assume that the portfolio is rebalanced whenever there is a recommendation change that affects the portfolio. Since the portfolio is always fully invested, when a new investment is added to the portfolio, other investments are first sold for the same amount, and then the new investment is bought for that amount. Similarly, when shares are sold from the portfolio, the money freed up can be used to buy underweight shares in the portfolio. For example, Nordea has a 1% cap on trading costs for Nordic shares, so we use this as a cost assumption. However, especially for larger share trades, much lower trading costs (0.1-0.3%) would often be available, so this is a conservative assumption.
Source: Inderes *The Inderes recommendation portfolio is constructed by including only the stocks with a Buy and Accumulate recommendation from Inderes, with a double weighting for the Buy recommendations.
Under these assumptions, trading costs represent 4-7 percentage points of portfolio size each year. Thus, a weighted long portfolio constructed using Inderes' recommendations would still have outperformed the index by 7.2 percentage points per year (2013-2024), despite its costs. Over a 12-year investment period, this would mean that the value of an initial portfolio of EUR 10,000 would increase to EUR 70,300 instead of EUR 33,100. Furthermore, owning the recommendation index portfolio would also entail some costs that we have not taken into account here (e.g. the purchase of new covered companies and portfolio rebalancing).
How accurate are the equity research recommendations for companies that pay for the research?
Inderes' research coverage is mainly based on research commission agreements with listed companies, whereby listed companies pay Inderes to cover their shares. This has allowed the analysis to be made widely available to investors, which is not the case with typical bank equity research. But there is always the question of the independence of the commissioned analysis: is the analyst providing a quality service to investors even if the company being covered is picking up the bill?
Inderes also provide free coverage of certain large companies and companies of particular interest to investors. We can therefore compare the return of a portfolio built on Inderes recommendations for companies that pay for the analysis (Commissioned) and for companies that do not (Non-Commissioned). We have placed a company in the paying category if, at any point in our coverage history, the company has paid for ongoing research coverage.
Source: Inderes
The data shows that Inderes' recommendations have actually been significantly more accurate for commissioned companies (2013-2024: excess return of 24.1 pp/year) than for non-commissioned companies (2013-2024: excess return of 2.5 pp/year).
Non-commissioned customers are in principle limited to larger companies, where companies are often followed by 5-30 analysts and the market can be expected to be more efficient due to high investor interest. In this case, mispricing is likely to occur less frequently and it will be more difficult to generate excess returns. Companies that pay for research often have less analyst coverage and investor interest. In this case, mispricing can be expected to occur more often and it will be easier to generate excess returns.
The numbers are consistent with this and, at least from this perspective, we have been able to provide high-quality equity research that serves investors, regardless of whether a company pays Inderes for research coverage.
Notes on the data used
We used Inderes' own recommendation history data, which can be found in our own systems in aggregated form from January 1, 2013. Therefore, the calculation does not include Inderes' early years of operations from 2009 to 2012. We also used Bloomberg total return data, which takes into account dividends in addition to share price changes. Due to the large amount of return data (over one million data points), we performed spot checks on the data, which show that share price changes and dividend payouts appear to be largely correct in the data. However, we have corrected individual broken return series in the return data, for example due to listing transfers.
The recommendation data is of course Inderes' own and can be downloaded here. We do not have the right to share the return data, but it is available from a number of sources, so the recommendation data we provide can also be used to independently assess our calculations and the returns of our recommendations. If the figures or methods raise questions or criticism, we have a thread on our Finnish forum where things can be discussed further.
