University of Nicosia, Institute For the Future
Apply TodayApply

Institute For the Future

M4 Conference

10-11 December 2018

Tribeca Rooftop, NYC

The findings of the M4 Competition that ended on May 31, 2018, have provided a wealth of practical information for improving the accuracy of forecasts and the correct assessment of uncertainty based on a huge data base of 100,000 series covering six application domains (macro, micro, demographic, industry, financial and others) and six time frequencies (yearly, quarterly, monthly, weekly, daily and hourly). M4 Conference elaborated on the findings of the M4, discussed their practical implications as well as how they can benefit business and other organizations in their effort to improve forecasting accuracy and correctly assess uncertainty.

The M4 Competition extended and replicated the results of the previous three competitions, using an extended and diverse set of time series to identify the most accurate forecasting method(s) for different types of predictions. It aimed to get answers on how to improve forecasting accuracy and identify the most appropriate methods for each case. To get precise and compelling answers, the M4

Competition utilized 100,000 real-life series, and incorporates all major forecasting methods, including those based on Artificial Intelligence (Machine Learning, ML), as well as traditional statistical ones.

M4 competition

M4 Conference Organizers

M4 Conference Sponsors

View Gallery

M4 Competition Winners & Prizes

PrizeDescriptionAmount
1st PrizeBest performing method according to OWA9,000€
2nd PrizeSecond-best performing method according to OWA4,000€
3nd PrizeThird-best performing method according to OWA2,000€
Prediction Intervals PrizeBest performing method according to MSIS5,000€
The Uber Student PrizeBest performing method according to OWA5,000€
The Amazon PrizeThe best reproducible forecasting method according to OWA2,000€
The winners of the prizes areAffiliationAmount of award
1st Prize Slawek SlymUber Technologies€ 9,000
2nd Prize Pablo Montero-Manso and teamUniversity of Coruna and Monash€ 4,000
3rd Prize Maciej PawlikowskiProLogistica€ 2,000
4th Price Slawek SlymUber Technologies€ 5,000
5th Prize Pablo Montero-MansoUniversity of Coruna€ 5,000
6th Prize Slawek SlymUber Technologies€ 2,000

The M4 Conference agenda

Monday, December 10, 2018
8:00 AM, 9:00 AMBreakfast
9:00 AM, 9:15 AMWelcome / Opening Remarks
9:15 AM, 9:45 AMInvited Talk Scott Armstrong (Wharton School, University of Pennsylvania) Data Models versus Knowledge Models in Forecasting
9:45 AM, 10:15 AMKeynote Speech Nassim Nicholas Taleb (New York University) Forecasting and Uncertainty: The Challenge of Fat Tailedness
10:15 AM, 10:45 AMCoffee Break & Networking
10:45 AM, 12:00 PMThe Three Most Accurate Forecasting Methods of the M4 Competition Slawek Smyl (Uber Technologies, USA) A Hybrid Approach to Forecasting Pablo Montero-Manso (University of A Coruña, Spain) Combining Statistical and ML Methods Maciej Pawlikowski (ProLogistica, Poland) Combining Statistical Forecasting Methods
12:00 PM, 13:30 PMBuffet Lunch & Networking
13:30 PM, 14:45 PMForecasting Challenges Robert Winkler (Duke University) The Value of Combining Forecasting Methods Tim Januschowski (Amazon) Statistical and ML Methods: Boundaries and Overlaps Michael Harris (Price Action Lab) Reflexivity in Financial Market Forecasting
14:45 PM, 15:15 PMCoffee Break & Networking
15:15 PM, 17:00 PMPanel Discussion ML Versus Statistical Forecasting Methods: Advantages and Drawbacks Chairperson: Peter Carr (New York University) Jocelyn Barker (Microsoft) Chris Fry (Google) Mike Gilliland (SAS) Maciej Powlikowski (ProLogistica)
17:00 PM, 18:30 PMFREE TIME
18:30 PM, 19:00 PMInvited Talk Forecasting at Uber: A survey Andrea Pasqua (Head Intelligence Decisions Systems, Uber)
19:00 PM, 21:00 PMGala Dinner & Award Ceremony of the M4 Competition Prizes
Tuesday, December 11, 2018
8:00 AM, 9:00 AMBreakfast
9:00 AM, 9:15 AMOpening Remarks
9:15 AM, 10:00 AMKeynote Speech Spyros Makridakis (University of Nicosia) The contributions of the M4 Competition to the Theory and Practice of Forecasting
10:00 AM, 11:30 AMPanel Discussion Can Deep Learning and ML Methods Improve Forecasting Accuracy? Chairperson: Dimitris Drikakis (University of Nicosia) Maria Kolitsida (Winningminds) Slawek Smyl (Uber) V. Assimakopoulos (NTUA) Yael Grushka (Harvard Business School) Jay Yonamine (Google)
11:30 AM, 11:45 AMCoffee Break & Networking
11:45 AM, 12:30 PMPanel Discussion Incorporating Judgmental and Quantitative Forecasts Chairperson: Antonis Polemitis (University of Nicosia) Robert Winkler (Duke University) Tao Hong (University of North Carolina) Pablo Montero-Manso (University of A Coruña) Warren Hatch (Good Judgement Superforecaster)
12:30 PM, 12:45 PMClosing Remarks and Announcing the M5 Competition
12:45 PM, 13:30 PMBuffet Lunch

Presentations

  1. Armstrong, Data Models vs. Knowledge Models in Forecasting: The Way Forward
  2. Harris, Reflexivity in Financial Market Forecasting
  3. Makridakis, The Contributions of the M4 Competition to the Theory and Practice of Forecasting
  4. Montero, Feature-based FORecast Model Averaging
  5. Pawlikowski, Weighted Ensemble of Statistical Models
  6. Smyl, Problems when using NN for time series forecasting
  7. Taleb, THE PROBLEM OF FORECASTING
  8. Winkler, The Value of Combining Forecasts
  9. Januschowski, Statistical and ML Methods: Boundaries and Overlap
  10. Pasqua, Data Science Manager, Intelligent Decision Systems, Uber