Thank you for participating in Haafor Challenge 2018!
The results on this post are final with F-source. We hope you had enjoyed the challenge.

 Name  F-score
 Chanmin Jun/ Jay Choi
 Tien Le  0.9578
 Juwon Song
 Ilhwan Ko
 tap  0.8311
 isj  0.7040
 Pilkwang Kim
 yck  0.6576
 ykk  0.6103
 Doan Xuan Dung   0.5436
 Chan Kim
 hkl  0.4985
 hrk  0.2644
 cwi  0.1835
We will contact the top 5 participants for the award. Also, we will give gifts to some of participants.
We will personally contact participants who will earn gifts by email.
Thank you again for your participation and please hope next Challenge!

Due:26 Oct 2018


Annoucement:Winners will be contacted via email


HAAFOR Challenge 2018

HAAFOR Challenge is annual event through which we seek to expose participants to various areas such as applied/financial math, computational science, machine-learning and artificial intelligence that we utilize in quantitative trading. This year’s challenge will focus on natural language processing (NLP). NLP is an important research area in HAAFOR as many sources of information relevant to financial predictions exists in unstructured format, e.g news and financial reports. Research and development of NLP techniques allow us to tap into these valuable data and utilize them to generate diversified sources of returns.
This year’s challenge task is named entity recognition (NER). While general purpose NER is more or less a solved problem, recognizing company names in financial texts poses higher difficulty, due to heterogeneity in the structure of their names. For example, Apple Inc (Cupertino-based iPhone maker) may show up in many different forms as  ‘Apple’, ‘Apple Computers’, ‘Apple Computer, Inc’, ‘Apple Inc’, ‘Apple (Singapore)’, Apple (Korea). We would like participants to design and implement an algorithm to identify mentions of certain companies in the corpus provided. Along with the corpus, participants will be provided with a mapping of relevant CompanyID, OfficialName in which

  • CompanyID is a unique number identifying each company
  • OfficialName is the “official” name of the company as used by stock exchanges and brokerages, e.g. ‘Apple Inc’ 

Given the above mapping and the corpus, the algorithm should parse each article and return the CompanyIDs of all relevant companies mentioned in the article. An evaluation corpus containing 100K articles will be provided, for which participants will label using the algorithm they have implemented. Please see attachment for the details. If you would like to participate in HAAFOR Challenge 2018, please send your application form(attached) to
Winners will be selected based on the evaluation method above. Selected winners will be awarded with prizes as below.

  • 1st and 2nd winner : Apple iPad Pro, 256GB
  • 3rd to 5th winners : Mechanical Keyboard
  • 5 random participants : High-end Headset  

In addition, winners and top-ranking participants will be invited for job interviews at our Singapore or Korea office.
20th July 2018

Due:23 Sep 2018

Submission:Please email the answer file to with subject - [Challenge] Both Individual/Team submission are available. Please kindly include all the team members’ names for team submission.

Annoucement:Announcement will be in October 2018