Mitsubishi UFJ showcases five prominent startups from its first FinTech accelerator batch



See the original story in Japanese.

Bank of Tokyo-Mitsubishi UFJ held a Demo Day for MUFG FinTech accelerator in Tokyo early this month. Though it has previously hosted a FinTech Challenge or a hackathon program in relation to the startup community, this is the bank’s first attempt at running an accelerator program.

Among the participants gathered from the end of 2016 through January of 2017, five teams took part in the program after clearing the selection process. The teams had been based at a working space named Garage for four months from this April. Operated by MUFG FinTech accelerator and located inside Tokyo Bankers Association Building alongside the FINOLAB FinTech hub, their Garage sojourn was spent making product improvements or brushing up services while receiving instructions from mentors.

During this Demo Day, the results of four months’ efforts were unveiled to persons in charge at MUFG group companies, venture capitalists and the media. Teams evaluated highly by the jurors     received a business assistance bonus or other goods as supplemental prizes.

Jurors at the Demo Day were:

  • Masaru Murai, Supreme Advisor, TX Entrepreneurs
  • Takane Hori, Partner, Mori Hamada & Matsumoto
  • Hironori Kamezawa, Managing Executive Officer, Mitsubishi UFJ Financial Group
  • Eiji Sumi, Executive Director, Mitsubishi UFJ Research & Consulting
  • Muneki Handa, President and Representative Director, Mitsubishi UFJ Capital

Additionally, the five participating startups were awarded the PR Times prize (complimentary use of press release distribution service PR Times gratis for a fixed period, from August of 2016 to July of 2017).

Top Prize winner: Xenodata Lab.

  • Mentoring by Minoru Imano, Globis Capital Partners
  • Supplemental prize: 2 million yen (about $19,700) as business assistance bonus


Among 3,600 listed companies in Japan, only 500 companies account for 14% of the total stocks have their financial analysis reports issued by securities companies for use by individual investors as investment decision reference material. Most mid- / small-scale companies with stocks targeted mainly by individual investors do not issue financial analysis reports, so investors must collect statements of account, timely disclosure documents or financial results documents, then analyze those documents by themselves.

Xeno Flash developed by Xenodata converts information on documents related to financial reports into table data by XBRL (eXtensible Business Reporting Language) analysis, PDF table analysis and PDF chart analysis. Specifying important points from table data with its own original algorithm, it extracts background sentences from the enormous text data behind the specified number data by natural language processing. The total process from collecting documents to analyzing information will be completed within one minute per stock.

Its value proposition is that users receive information about every company quickly after a financial results announcement, when only important parts are proposed, and offered excellence in viewing information using infographics. The team plans to launch service at Securities within 2016. In the future, it will explore possibilities of information provision for individual investors via retail-type brokerage firms as B2B2C (business-to-business-to-customer), information distribution to new media, service expansion to cover overseas stocks, or service development to wholesales-type brokerage firms having financial analysts as B2B (business-to-business).



Runner-up: Alpaca DB

  • Mentoring by Jun Nakajima, Archetype
  • Supplemental Prize: 1 million yen (about $9,800) as business assistance bonus


Alpaca, which had developed a trading platform Capitalico in the past, developed anew Alpaca Scan during the accelerator period. This tool analyzes relevance between stock price fluctuation and various data in real-time through continuous monitoring of 7,000 stocks listed on US exchanges and by using AI (artificial intelligence) technology focused on market data. Based on past price fluctuation data or news, it helps users understand which stock and how much of this to own, how much the stock price is likely to rise in the future, or when to trade.
Simultaneously, the firm developed its own database named MarketStore as a back-end infrastructure for Capitalico or Alpaca Scan focused on financial time-series data. It is more cost effective compared with Oracle or Kx; in addition, the firm also succeeded in reducing the amount of RAM required as the database works with just a percentage of the resource. The time taken for algorithm analysis on Capitalico was dramatically shortened to about 10 seconds, although it had taken 30 minutes as of this March. On the platform, 7,000 types of ‘model’ have been created so far.

Henceforth, the firm plans to launch a service to support foreign currency reserve-building at optimal timing, based on MarketStore or Alpaca AI Engines for the Japanese internet bank Jibun Bank. This bank was established upon joint capital investment by Bank of Tokyo-Mitsubishi UFJ and the telecom carrier KDDI. Also, the firm will develop jointly a trading tool utilizing AI for Securities.




See also:

AWS Award winner: Smart Idea

  • Mentoring by Akira Kurabayashi, Draper Nexus Venture Partners
  • Supplemental Prize: Amazon Fire Tablet and other Amazon goods not for sale


Smart Idea had already developed an account book app named Quick Money Recorder which ‘can be input only in 2 seconds,’ and the app has been downloaded 3.5 million times since its launch back in August of 2012. Among smartphone users living in Japan, 10% of the women in their 20’s to 30’s are Quick Money Recorder users, according to the team. For what kind of services targeting its own wider range of woman users does the team propose to cooperate with financial organizations?

In retail-type financial business, demands for fund occur in response to life events of customers, so that every financial organization expects to approach the presumed users at the time of their life events, but that is not easy due to lack of information or methods of approach. On the other hand, some surveys revealed that women in their 20’s and 30’s, mainly targeted by Smart Idea, have worries such as ‘cannot save up money’ or ‘have no idea about money, have nobody to ask.’

So, Smart Idea suggests a concept called ‘Fintainment’ which combines Fintech and Entertainment, and developed two apps: Okane Navi and a chat bot app. Okane Navi is an app adopting quiz factors related to financial services into a love simulation game in which 70% of women in 20’s and 50% of women in 30’s are absorbed in Japan. Users play a quiz game to obtain rewards, but they are not allowed to continue on without watching a movie for every five questions. The chat bot app allows users to receive proposals about ideal financial services by answering questions generated by AI.

Smart Idea plans to hold a life planning seminar which targets women in 20’s and 30’s together with Mitsubishi UFJ Trust and Banking. By leading Okane-reco users to Okane Navi or the chat bot app, the firm aims to increase the number of Okane-reco users to 3.5 million and to grow the entire service to a 10 million user-scale.



See also:


  • Mentoring by Kengo Ito, Genuine Startups


Zerobillbank, started last year in Israel and financially supported by Samurai Incubate, is developing ZerobillCore which enables the issuance of original virtual coins based on the blockchain, and an app Z-Wallet which enables receipt of the issued coins based on position, time and information acquired from sensors.

In the pitch, the Zerobillbank team demonstrated how to issue and provide virtual coins called ‘Marunouchi Coin’ as an example (the Demo Day was held in Marunouchi, Tokyo) to users holding the Z-Wallet in the Marunouchi area. It allows for issuance of coins to users within a fixed distance from a certain base point using the Geo-fence idea. Since Z-Wallet works as a sensor / beacon as well, various settings are available such as providing coins to users who have visited a certain place or who walked a certain number of steps.

As for assumed cooperation cases, the team suggested issuing ‘life insurance coins’ according to the users’ number of steps or exercise amount, by tying up with life insurance companies in order to develop health encouraging-type insurance products, or issuing ‘automobile insurance coins’ according to the insured cars’ track record or linking with telematics data by tying up with non-life insurance companies.

Moving forward, test operations of the reward system at Bank of Tokyo-Mitsubishi UFJ and implementation of a point system for third-party websites at Securities are scheduled, and with Mitsubishi UFJ NICOS, Zerobillbank looks to find ways of utilization for reward development with its cooperative partner companies.



Knowledge Communication

  • Mentoring by Yasuhiro Yoshizawa, Inclusion Japan


Knowledge Communication had launched a cloud service Nare-com AI supporting easy use of AI or deep learning this March. Nare-com AI shortened the time needed for algorithm selection for machine learning to 2 weeks, equivalent to one-fourth that for conventional technologies, and also realized automation of the selection process, which formerly required data scientists, using the roundrobin technique for both algorithm and parameters. The team has begun considering means of utilizing Nare-com AI for banking services during the accelerator, and set a goal to develop a service enabling use of data mining and the modeling process for less than 500,000 yen (about $5,000) / within a day, while such kinds of service generally require tens of millions of yen / more than two months minimum.

Knowledge Communication had modeled a judgment process based on 20,000 SME (small and medium-sized enterprise) financial statements and information about judgment by persons in charge of financing at Bank of Tokyo-Mitsubishi UFJ regarding whether each SME is appropriate as an investment target or not. Inputting SME financial statements for this year into the model made from last year’s data as trial, the team succeeded in creating reference information on some 4,000 companies for investment decision within 30 minutes. It is valuable to semi-automate a part of the judgment process under conditions actually used until last year, although conventionally persons in charge of financing had to judge based on data of target companies such as capital adequacy ratio, plus his / her experience and intuition.

In the future, the team expects the service to be used for stress test of financial works at banks, sampling of potential customers of new services having the same characteristics asw other service users or list sampling of candidate borrower companies that are likely to have financial demands.


It is unclear with what kind of KPI (key performance indicator) MUFG FinTech accelerator is managed, but it is easily assumed that the number of suggested cooperation projects with MUFG’s group companies is one of the most important evaluation criteria. Some of the cooperation projects would generate revenue sharing or sales immediately after graduation from the program. That may be one of the attractions for startups to participate in the operational company-type accelerator.

Although initially future plans of MUFG FinTech accelerator had been undecided because it was the first attempt for Bank of Tokyo-Mitsubishi UFJ, at the closing ceremony the manager of the accelerator Eiichi Kashiwagi (General Manager of Global Innovation Division, Bank of Tokyo-Mitsubishi UFJ) expressed his willingness to start accepting application for the 2nd batch from this autumn or within the year, upon considering the results of the 1st batch.

See also:

Translated by Taijiro Takeda
Edited by “Tex” Pomeroy