The M6 Financial Forecasting Competition is the next installment of the renowned Makridakis’ Forecasting competitions. The M6 will be similar to the preceding five competitions, with the aim of obtaining empirical evidence that sheds further light on the causes and implications of the paradox of the Efficient Market Hypothesis (EMH). This hypothesis states that financial share prices reflect all available information, and investment outperformance relative to the market is therefore not feasible, even for legendary investors like Warren Buffett, Peter Lynch and George Soros, among others; as well as for celebrated investment firms such as Blackrock, Bridgewater Associates, Renaissance Technologies, and DE Shawthat, all of which have achieved phenomenal investment results, amassing returns impossible to justify by mere chance, and casting doubt on the validity of the EMH.

Forecasting competitions can only ever approximate reality, and their value and usefulness depends on how realistic such approximations can be made to be. From a motivational perspective, participants do not have “skin in the game” as they do not invest their own money, and we rely on modest prize money and professional and academic curiosity to encourage participants. To make the task manageable for a wide range of competitors, we limit the investment universe to 100 assets, but include both individual equity securities and asset classes via a range of ETFs. The M6 competition allows for forecasting and investment decisions to be made in a truly real-time context, mimicking the true environment that real-world forecasters and investors operate in, but is of necessity limited in terms of time horizon and duration. M6 will run for 1 year, with 12 rolling assessment periods across which forecasts and investments will be made. While this time period limits the ability of M6 awards to properly account for fat tails associated with market returns, the structure of the competition still allows addressing several critical types of questions, including among others the following:

  • What are the key differentiating factors associated with the good/poor forecasting and investment performance?
  • What is the interplay between objectively measured forecasting performance based on strictly proper scoring rules and investment performance measured using criteria associated with portfolio optimization like the information ratio? A key aspect of M6 is that it captures both return forecasts and portfolio decisions.
  • Are excess financial returns achieved by one or a combination of the following factors?
    • The ability to accurately forecast overall market returns, or those of individual assets.
    • The ability to properly model market or individual asset uncertainty.
    • The ability to formally optimize a portfolio when making investment decisions.
    • The ability to use judgment, model-based methods, or some combination thereof when constructing forecasts and making investing decisions.
    • The ability to adjust (or keep fixed) an investment strategy, over time.
    • Other factors, including judgmental and model-based prediction and investment decision biases and inefficiencies that can be exploited in forecasting and in investment allocation.

No major competitions have been previously conducted in the area of financial forecasting. The M6 competition aims to provide empirical evidence about how investors can improve the accuracy of their forecasts, mitigate the uncertainty involved in these forecasts, and exploit their findings to build robust, profitable portfolios. The M6 forecasting competition innovates in the following three directions (see Makridakis, Fry, Petropoulos and Spiliotis, 2021):

  • M6 is live, so that no data are concealed. Having a live competition allows participants to:
    • Search and use any available information for use in their forecasts and investment decisions.
    • Include judgmental inputs about the economy at both the macro and the micro (industry and firm specific) levels. This allows participants to utilize both judgmental and numerical (model-based) inputs in order to improve their performance.
  • Instead of using a single one, M6 uses multiple rolling origins to evaluate performance. This allows for participants to learn and to adjust their methods and/or models in real-time. More importantly, considering multiple evaluation rounds allows separating skills from luck and investigating the consistency of the participants’ performance over time.
  • M6 consists of two challenges (forecasting performance and investment performance). In this sense, the competition is a duathlon, and winners will be announced in both sub-competitions, as well as in the overall combined competition. Additionally, prizes will be awarded quarterly, as well as at the end of the competition, with the “Global” winner being the participant who provides superior performance in both sub-competitions over the entire horizon of the competition. A key metric will thus be the ability of participants to effectively exploit forecasting methods to mitigate uncertainty and to translate their forecasts and findings into meaningful, profitable decisions.

The M6 expands on the learning objectives of prior competitions by focusing on the interplay between forecasting and investing, as well as on the importance of forecast accuracy/uncertainty when used to support investment decision making.

The M6 competition will be a live competition, lasting for twelve months, starting in February 2022, and ending a year later in 2023. It will consist of a single month trial run and 12 rolling origins (every four weeks) for participants to provide their submissions and be evaluated once the actual data becomes available. The prizes, that will consist of a total of US$300,000,

For further details on the evaluation of the competition, please see here.

To participate in the competition, register now

The M6 Academic Committee
Spyros Makridakis, University of Nicosia
Anil Gaba, INSEAD
Ross Hollyman, University of Bath
Fotios Petropoulos, University of Bath
Evangelos Spiliotis, NTUA
Norman Swanson, Rutgers University

Share Post!

Subscribe to Blog