A minimum of 5 years of historical data updated daily with the ability to extract specific sections, e.g. MD&A. Public news and press releases collected from exchanges including:

All company filings for list companies have been captured across the exchanges we cover. Filings include reports such as:

We have a wide range of predefined functional/NLP/machine learning algorithms (English/Chinese) which enable insight from the selected datasets. Our algorithms include but not exclusive to: 

• Functional: Sentiment, Event detection, Theme detection, Summarisation, Article/Phrase Clustering, Keyword Extraction 
• NLP Pre-processing: NER, Part of Speech Analysis
• Machine Learning: BERT – (Bidirectional Encoder Representations from Transformers), Deep Learning, Integrated Training

China data coverage

• Order flow data
• Service Earning call transcripts
• Public disclosures
• Executive official response
• Issuers
• Instrument
• Company relationships
• Equity top 10 shareholders
• Fund Holdings

Social media coverage

We have extensive social media coverage including Baidu, WeChat, TikTok and Weibo.

Comprehensive and high-quality data

China public filings – We have full public filings coverage of Shanghai, Shenzhen and Hong Kong exchanges since 2016 which is updated daily. We retain all original PDF files with text versions also available.

China A share earning call transcripts – We have full coverage of Shanghai and Shenzhen stocks since 2013. We have all original documents, in structured JSON format, and translated into English also. See further down this page for details in our ‘Focus on’ feature.

China golden data – We have full coverage of Shanghai, Shenzhen and Hong Kong exchanges with key metadata for all listed companies, related entities, people and social media accounts.

We maintain search index APIs for Baidu, WeChat, TikTok and Weibo.

We provide comprehensive official/regulatory documents which link and impact securities both directly and indirectly.

Information extraction – we have a number of detailed NLP algorithms and sophisticated calculation platforms which accurately extract key required information from documents in a variety of ways.

The language barrier – We have developed an optimised Chinese – English translation engine specifically tailored for the financial industry to increase accuracy against non-industry specific generic platforms.

Focus on: China A Transcripts

Data types

When delivering files, each report (in JSON format), together with its metadata file (in CSV format) is included within a zip folder. English and Chinese versions of the same report are separated in two individual zip folders. Data files are put in folder directory by YYYY/MM.

Each transcript is in JSON format and takes on the following filename convension: DATE_TIME_TRANSCRIPT-UNIQUE-ID_LANGUAGE.json
With the metadata file being metadata.csv

Reports are then classified into one of the three types below:

ECM: Earning Call Transcripts – the transcripts of earning call meetings after annual or quarterly reports being released.

BRD: Public disclosures on broker onsite research – by regulation, if any broker/investment bank carry out on-site research to listed companies, they are under obligation to disclose such information publicly. Since it’s an ad-hoc activity, not every company has such disclosures. Statistically around a 1/3 of the whole universe has these disclosures.

OQA: Executive official responses on online platforms – there’s a small number of online forums that allow investors to directly ask directors of listed companies questions, and executives may reply selectively. Such communications can happen in real-time, which provides the most up to date insights of the listed companies.

Crypto reference data including:

• Token name
• Code
• Website URL
• Official Reddit
• Telegram accounts