top of page

Our Selection Model

We apply quantitative research
to identify high-quality stocks.

During our conversations with clients and partners, questions about our market approach and editorial policy frequently come up:
 

  • How do we identify outperforming companies?

  • Why do we favor a dynamic approach (1year duration)?

  • How do we manage risk?


The answer to these questions all lie in the statistical selection model that we designed and use on a daily basis. All our decisions are based on quantitative indicators computed through a 4 steps methodology.

 

Here's how it all works, starting with a visual representation:

The European Investor Model.PNG

Step 1:
Aggregating Consensus Data

Our first step in evaluating the potential of a stock or an economic sector is to compute accurate consensus indicators. This means we collect all opinions and previsions recently published regarding a company in order to uncover what market professionals think about it and how those opinions evolve.

 

Concretely, we collect two different type of consensus data that will be combined in our analysis:

 

  • First, we source all the expressed previsions published by financial analysts regarding a stock or a sector. This is the simple part, as most research can be obtained through financial data providers.

  • Then, we shift our focus to implicit opinions. This means we uncover stocks and sectors that institutional investors are actively buying. It differs from expressed previsions in the sense that we are no longer merely tracking published opinions, but the actual buying (and selling) decisions from large investors (like banks, hedge funds and money managers). Implicit opinions are much harder to collect, but offer better forecasting power.
     

  • Last but not least, we assign a "weight" to all the opinions (expressed or implicit) we obtained. This means that we are actively giving more relative importance (weight) to some data over the others. Why? Because not all opinions are created equal, and we want our final results to be influenced by the most reliable experts first.

At this stage, we already have (i.) collected previsions published by analysts, (ii.) integrated buying decisions from large investors and (iii.) assigned consistent weights to all these data. Now, we can finally compute our consensus metrics, in the form of 12-month price forecasts range. Those price forecasts form the bedrock of our decision making and our editorial policy.

Step 2:
Integrating Dividends Expectations

As we've just seen, consensus computation is the main building block of our selection methodology. However, other factors play a critical role in our decision process, and the dividend policy of a listed company is among them.

We favor companies distributing generous and stable dividends. Why? Because dividend-focused portfolios tend to be less risky and resist better during markets drawdown. There are several expalnations to this:
 

  • A generous distribution policy is generally an indication that a company is commercially mature, beneficiary and can serve its debt. Qualities logically associated with resilient organizations.
     

  • Dividend stocks generate income regularly, even during bear markets. As a results, many investors prefer to hold on to them, even during difficult periods.
     

As risk management is central to our editorial policy, those dividend-induced qualities are critical to us and align with our objectives.

 

Note that this doesn't mean that all stocks we recommend have generous dsictribution policies (many don't), but it is an important decision factor in our models.

Step 3:

Incorporating of Momentum Metrics

Momentum is the measure of the the speed of price changes in a stock. It is a metric mostly used to characterize the trend of a company in terms of direction (bullish, bearish, neutral) and quality.
 

Continuously computing those trend indicators on hundreds of European stocks allow our models:

  • to rate companies  on variables like in terms of quality, duration and probability of continuation.
     

  • To compare stock (and sectors) and rank them to identify those with the best probability of trend continuation in the next 12 months.

 

Also, momentum is mechanically correlated with consensus (see step 1), as buying decisions from institutional investors should result in bullish trend.

In short, this momentum angle offers us a new and complementary dataset, allowing us to refine our analysis.

Step 4:

Setting Targets using Historical Volatility

Once one or several promising companies have been identified, time has come to make another important decision: where to set the intermediate price targets?  Embedded in all our recommendations, price targets are here to encourage our readers to exit winning positions in a progressive and optimized manner.

Defining those targets is a balancing act, as it must be high enough to justify the risk taken the investor (profit maximisation), but realistic enough to maximize the probability it will be reached within 12 months (risk reduction).

Here again, relying on quantitative indicators helps: we define those partial exit levels by analysing the historical volatility of the studied stock or sectors. In other words, we analyse the recent past behaviour of the stock to set exit targets that combine a high probabllity of being reached while offering sufficient profit.

And finally:

Promoting Sectorial Diversification

Through the first 4 steps we just described, we have:

  • Identified high-potential stocks through consensus aggregation, dividend expectations and momentum analysis
     

  • Defined optimized price targets based on historical volatility


One critical variable to integrate now is diversification. A classic pitfall in quantitative selection is sectorial over-exposure. Put simply, this means that statistical models left unattended can sometimes lead to excessive concentration in one corner of the economy.

 

For example, when gas and oil prices are high, oil and energy companies perform well, and their stocks all give positive signals at the same time.

 

A tendency we have to balance out to ensure optimal diversification within our recommendation list (and our reader's portfolios). Our models natively limit the amounts of companies selected within the same sector, and actively hunt the best opportunities accross all major industries.

Now, will all this computing markedly improve the performance and risk management of your portfolio?

Yes, without a doubt. But don't take our words for it, and form your own opinion by checking our complete track record.

bottom of page