Analyst forecasts - separating the wheat from the chaff

Analyst's earnings forecasts and earnings revisions are well-known drivers of stock price movements and are widely applied in both academic research and quantitative stock selection models. This generic investment signal recommends to go long the stocks that recently have been upgraded by analysts and to short the stocks that recently have been downgraded.

Robeco continuously works on improving its models' signals and definitions. This internship's objective is to dive deeper into a detailed analyst forecast database, preferably from different angles, and test some innovative ideas presented in both the academic literature and broker reports. Examples of these ideas include, but are not limited to:

  • Star analysts: Do certain analysts outperform others?
  • Fresh vs. stale forecasts: Do more recent analyst forecasts contain more information?
  • Modeling the analysts' forecasts using innovative machine learning concepts.

This project involves all steps of the quant model development cycle. For this project, we envision the following steps:

  • Literature review:
    We are aware of several papers that were published on this topic. Having a comprehensive and up-to-date overview of the available literature helps to shape the rest of this project.

  • Data preparation:
    Since Robeco has access to massive amounts of data related to analysts' forecasts, you will work with state-of-the-art big data tools that will help you aggregate this data into valuable insights. No prior knowledge of big data tools or database programming is required. With good programming skills, you will be able to work with these tools in no time.

  • Replication of past results:
    In the past, Robeco has conducted research that relates closely to this project. By replicating the results of Robeco's past research, you will be able to test whether these results hold in an out-of-sample context as well. On top of that, by replicating these results, you might find.

  • Predicting Analysts Revisions using Machine-Learning:
    Bew, Harvey, Ledford, Radnor, and Sinclair (2019) describe a novel way to use Bayesian Machine Learning in predicting future analysts' revisions. We are interested in seeing whether these results can be replicated or improved upon using a wide array of other machine-learning techniques.

This project's preferred candidate has a solid background in econometrics, statistics, or mathematics, with know-how or willingness to dive deep into machine-learning techniques. Additionally, we require good programming skills (preferably in Python) as well as interest and enthusiasm for financial markets.

Are you interested?
Let us know your motivation and send it together with your top-3 favorite internship topics, your CV and list of grades to
Available projects Previous projects


Chan, Jegadeesh, and Lakonishok (1996) “Momentum strategies”, Journal of Finance 51

Bew, Harvey, Ledford, Radnor, and Sinclair (2019) “Modeling analysts' recommendations via Bayesian machine learning”, Journal of Financial Data Science 1(1)