Objectives of the service
The service provides accurate and meaningful car count data to financial practitioners for 65+ European store chains, for them to ingest into their financial back-testing, trading, equities portfolio research and data platforms. The data can be used standalone or in conjunction with other data-sets to predict and validate predictions of earnings, stock price and other key retailer performance indicators
Users and their needs
For the initial product release, RetailWatch as the go-to-market product extension of RWatch services quantitative and technical participants within hedge funds, asset managers and their service providers (such as research providers and data vendors) by providing meaningful satellite-derived car count data from retail stores. That data can and be:-
- Incorporated as part of a data universe, usually as part of a back-testing system
- As part of a database or a data class
- Alongside market (stock price), fundamentals (company data) and alternative (other new) data sources
- Accessed through a csv file over ftp or over a RESTFul API
- Point-In-Time (time-stamped)
- Normalized
- Sampled over key time durations
The service targets European firms, but is useful to anyone trading European cash equities.
Service/ system concept
The car count process, originating from earth observation satellite, applies automated deep and machine learning to count at scale.
This in turn underpins store-specific data and more thematic index-based analyses of store financial performance, supporting multiple trading styles and strategies
Compared to other “alternative data” sources, for example credit card or consumer transaction data, or geolocation (within retail outlet) data, satellite-derived data is smaller and simpler. Quoting one satellite data user, “it can fit on the back of a postage stamp”, so is easier to ingest and analyse. It also incurs less compliance risk, particularly in the GDPR-sensitive EU. It can coexist with those other alternative data sources, either as part of a boosted blended data-set, or as a means of validation.
Space Added Value
The Kore Platform, used by Deimos to provide services based on subscriptions in various EO applications and the DigitalGlobe GBDX Platform, providing big data processing and access to the archive of WorldView imagery, provide the backbone of the hundreds of thousands of acquired images. Deep learning techniques would underpin the image segmentation to form the basis of the automated count methodologies.
Current Status
As a result of the Kick-Starter Project, we are releasing the RetailWatch product in Q119, further expanding product capability and coverage with Deimos UK and Geospatial Insight working together to deliver a combined service as follows.
- A product building on the work conducted through this project is going to be released early in 2019
- There is a financial services market demand for a Europe-focused satellite car counting capability, as part of the wider alternative data universe, and the initial results indicate that the data output is useful.
- Deimos UK and Geospatial Insight are collaborating and expanding on the capabilities created during the course of this KickStarter Project, activities not limited to:-
- A Demo Project Application
- Continued hiring
- Operational collaboration in producing and expanding RetailWatch
- We shall collaborate on seeking additional demonstrator funding.
Prime Contractor(s)
Subcontractor(s)
