Use Cases of Alternative Data

【Use Cases of Alternative Data】

【Web Scraping: Pointing Negative Fundamentals for GoPro】

In this case study the data obtained from comparison sites such as Shopping.com with web scraping showed that GoPro's new product's sales wouldn't be high.

Eagle Alpha, a Dublin-based alternative data provider, published a report named "The Rise of Data - Leveraging Data in Your Investment Process Edition7", saying that data of pricing or ranking on the comparison site suggested that a new product at that time "Session" sold poorly. While many of the former products ranked No.1 for two terms in a row, "Session" ranked No.5 at the term when GoPro started selling it and No.6 at the subsequent term.

In fact, such trend showed up in its stock prices and the stock price plunged by about 60% (while S&P500 Index rised by 10% )

In conlusion, web scraping data will give you some information about fundamantals, for example, pricing and popularity on EC sites.

（Source：Eagle Alpha）

【SNS: Forecasting What Game Would Be Most popular】

In this case study the number and content of tweets (positive/negative) enabled you to understand what video game would be popular. The report Eagle Alpha published “The Rise of Data - Leveraging Data in Your Investment Process Edition7“ suggested that “Overwatch“, a new product of a US game publisher Activision, was likely to be more successful than other competitive products.

Eagle Alpha also pointed out the fact that the number of tweets about Overwatch was greater than that of other products, and that the proportion of positive tweets about Overwatch was also higher. Indeed, while the number of tweets about another product was only 1 million, that of Overwatch was as many as 1.2 million. In addition to this, the proportion of its positive tweets was 72% - the highest out of the set of products being compared

In conclusion, SNS data enables the understanding of consumer sentiment before/after purchase.

【Cosumer Transaction Data: Macroeconomy Analysis】

“Nowcasting of Consumption with Business Data”

Yoshiyuki Suimon, senior economist for Nomura Securities

This report is one that Mr. Suimon has contributed to us and uses our product JCB Consumption Now to estimate GDP consumption (JCB is a Japanese credit card issuer).

Forecast and actual value of real GDP consumption

Source：Excerpt from Nomura Securities Yoshiyuki Mizumon's contribution to the company, "Nowcast of consumption using business data" (September 2, 2019)

In the figure above, the blue line represents a measured value, the red line – an estimated value calculated with JCB Consumption NOW, the green line – ESP forecasts (economists’ consensus). He got a result that JCB consumption is 0.05% more accurate in RMSE (Root Mean Square Error), 8~10% more accurate in DA (Direction Accuracy) than the consensus. In addition, while the consensus is said to have a upside bias, the estimated value performs better when measured consumption goes down.

　Consumption statistics estimated with consumer transaction data is useful because they are quciker than official statistics, made mainly with surveys. This analysis takes advantage of that usefulness.

【Cosumer Transaction Data: Academic Use Cases】

The responses of consumption and prices in Japan to the COVID-19 crisis and the Tohoku earthquake

Tsutomu Watanabe, Dean and Professor of Economics, University of Tokyo (Founder and Technical Advisor of Nowcast)

This report poses a problem in which you should regard COVID-19 and the Tohoku earthquake in March 2011 as similar economic phenomenon, and argues about the difference in their mechanism.

The author points out that the responses of consumption and prices (food and daily necessities) to the COVID-19 are similar to that of the Tohoku earthquake in that it temporarily rises very soon after the shock breaks out and returns back to the normal level after a certain period of time. He used Nikkei CPI Now (daily sales and price index based on POS data in Japan), which is effective in terms of timeliness, and analyzed economic impact as early as April 2020.

The author, Tsutomu Watanabe, is the founder of the company and is currently a technical advisor.

Daily Sales and Price Rate of Change at Supermarkets in Japan when the COVID-19 shock and the Tohoku Earthquake happened.

Source：Extracted from the column "The responses of consumption and prices in Japan to the COVID-19 crisis and the Tohoku earthquake" by Tsutomu Watanabe, our founder and technical advisor

In this way, the academic world also accepts alternative data as useful. Promptness like this is one of the most attractive attributes of alternative data.

In addition to this, an essay on how the COVID-19 crisis could affect earnings inequality in Japan written by Kitao et al. (2020) uses JCB Consumption NOW for an analysis. She is a professor of Economics at University of Tokyo. More and more academic use of alternative data is expected in the future.

【Geolocation: Economic Analysis with Cell Phones’ Geolocation Data.】

The More Elderly, the Less Going Out, Reported by Nomura

Source: Nihon Keizai Shimbun 2020/3/5

Mr. Suimon at Nomura, who contributed the report which uses JCB Consumption Now, is enthusiastic about making use of alternative data.

In this article on Nikkei, he analyzed macroeconomy with geolocation data. He pointed out that the older people are, the less they go out during the COVID-19 crisis.

Other than this, there are an increasing number of geolocation data use cases out there.

For example, some data provider judges whether a Tesla's factory is in operation with geolocation data from factory workers' cell phones. Using this data, it is possible to analyze not only consumer behavior but also production activities.

【Satellite Images: Analyzing Oil Storages】

Understand the Economy with Satellite Images

Source: Nikkei Business 2017/11/25

Orbital Insights, as mentioned in the article, is a US-based leading startup in satellite images that analyzes them and judges, for instance, how oil production is going and how much people are traveling.

It is famous for grasping the amount of oil in storage tanks. While OECD member countries report accurate economic statistics such as oil deposits, non-OECD member countries like Saudi Arabia report inaccurate statistics or in some cases don’t report any statistics - such as how many tanks the country has or where they are. But the US-based startup has focused on floating roofs and the height of shadows, and estimated oil deposits.

This type of method is also used for other commodities like ores and LNG.

※1…Web scraping means automatically collecting useful information for data analysis from open web sites.