iTranslated by AI
The Story Behind Founding Makedeco: A Market API Developer Community
Introduction
I created a Developer Community for Market APIs called "Mkdeco."
We plan to operate on Discord, and the launch is scheduled for the kick-off event on September 15th.
I’d like to reflect on and write about why I decided to create such a community (of course, based on content already presented at events and other venues).
J-Quants Project
It all started with the J-Quants project, hosted by JPX Market Innovation & Research (JPX at the time). This project aims to create a venue for competitions and API utilization by opening up data to individual investors.
AlpacaJapan, where I serve as CPO, and JPX had established a relationship through various data analysis projects, and it started when they consulted me. Initially, the goal was to hold a stock analysis competition.
When I heard the topic of a "stock analysis competition," I recalled the major failure of the first Two Sigma competition held on Kaggle. In the data scientist community, these competitions were recognized as mere "luck games"—competitions decided by luck rather than skill (often referred to as "terrible competitions" in some circles, making Kaggle competition organizers tremble).
When I was consulted, I felt that since the competition would be held under the "JPX" name—the biggest name in Japan—the success of this competition would be a project that determines the future relationship between Japan's financial markets and data scientists.
Therefore, I organized the best team within the company. I became the main PM, Hio (who is among the current operation members) served as the data engineer, and for data scientists, we brought in specialists from our team as needed, forming the best possible team composition.
The first thing we did was research the problem design of stock analysis-related competitions from all times and places. There were probably about 20 competitions. Analyzing their problem designs, we found that successful competitions involved things like volatility prediction, which are relatively easier to predict in the financial industry. From there, we designed about 20 problems, had data scientists actually solve three of them, and verified which ones worked as competitions. I hope to share more about this story in the future.
From this, the concept of "Competition design where the legitimate efforts of data scientists are rewarded" was born. The two subsequent competitions were very successful, with skilled participants ranking at the top, positive feedback from participants, and over 2,000 data scientists participating in total.
Furthermore, the problem we designed for Kaggle (currently in the evaluation period) captures the excitement of financial market prediction. It received high praise as a "best competition design" from the head of equity investment at Millennium and from Kaggle Grandmasters. Writing about the Kaggle problem design would be quite long, so I'll summarize it in another blog post.
APIs Open Up the Future
Once the competitions settled down, the beta release of the API for obtaining stock and financial data through J-Quants began. UKI-san previously summarized the potential of J-Quants in a blog post, and it was obvious that stable data provision via API is essential to realize the possibilities described in that entry.
Coincidentally, AlpacaJapan became a wholly-owned subsidiary of AlpacaDB, Inc., which is headquartered in California, USA, and develops fintech-related services and products globally.
Within this, there was a mission to deploy a US stock Trading API in Japan, and this mission overlapped perfectly with J-Quants' mission. Following the success of the competitions, we needed to aim for the success of the API. I realized that the most important thing for that was to create a wonderful community, much like what we did for the competitions.
However, having just been acquired, AlpacaJapan is not yet at the stage of providing a Trading API. Nevertheless, believing we could help broaden the base, JPX Market Innovation & Research agreed to support us as an official sponsor, leading to the launch of this community.
Future Plans
Now, there are various interesting topics at key points, so I would like to summarize them in another blog post next time. However, the Market API Developer Community (Mkdeco) was born from these developments, and I consider it the most important thing at present in my personal mission to create the future of financial markets and the data scientist community.
If you've read this far and are interested, we have plenty of exciting content lined up, such as a Q&A session with UKI-san, one of Japan's top system traders, and running the incredibly powerful model that won second place in the J-Quants competition. We would love for you to join the kick-off event!
Discussion