adase-api

Adalytica Sentiment Explorer API

Introduction

Alpha Data Analytics ("Adalytica") is a data analytics company specializing in sentiment analysis. Its core product, Sentiment Explorer, uses sentiment monitoring technology to analyze news and social media content, transforming it into machine-readable data. The platform provides insights into public sentiment that can be applied across various fields, including capital markets, demand forecasting, political analysis, marketing strategies, and more.
This package powers the Fear and Greed Index, a key tool for assessing market sentiment.
For a detailed explanation of how it works, read the article: How Fear and Greed Using Sentiment AI Works?

This packages is made for data proficient users, like quantitative analysts, data engineers, data scientist or analysts

Package is build around two search engines keyword and topic, as explained below. Data contains two indicators:

• score — sentiment polarity score in the range of [-1, +1]
• coverage — ratio (proportion) of query hits divided to total number of data
  • naturally, different queries will have varying levels of popularity. To make them comparable over time, z-score normalization is recommended and supported by the package
  • indicator shows for most of the queries a strong weekday seasonality

To install

pip install adase-api

Credentials

In case you don't have yet the credentials, you can Sign Up for free 14 day trial

Sentiment Open Query

To use API you need to provide API credentials and search terms

from adase_api.schemas.sentiment import Credentials, QuerySentimentTopic
from adase_api.sentiment import load_sentiment_topic

credentials = Credentials(username='youruser@gmail.com', password='yourpass')

search_terms = ["inflation rates", "bitcoin"]
ada_query = QuerySentimentTopic(
  text=search_terms,
  credentials=credentials
)
sentiment = load_sentiment_topic(ada_query)
sentiment.tail(3)

indicator                     score        coverage     score  coverage
query               inflation rates inflation rates   bitcoin   bitcoin
          date_time                                                              
2025-02-04 06:00:00        0.112417        0.002583  0.011417  0.001750
2025-02-04 09:00:00        0.103167        0.002750 -0.155000  0.001500
2025-02-04 12:00:00        0.047500        0.002750 -0.050375  0.003375

Returns two indicators, coverage and score, in a Pandas DataFrame indexed by timestamp objects, with columns organized as a multiindex

More advanced queries

You can supply a dictionary of query keys, each containing multiple comma-separated sentiment queries along with the corresponding weight for each subquery:

ASSETS = {
    "NXPI": {
        "queries": "(+NXPI), (+semiconduct*)",
        "weights": [1.0, 0.5]
    },
    "PLTR": {
        "queries": "(+Palantir), (+PLTR)",
        "weights": [2.0, 1.0]
    }
}

ada_query = QuerySentimentTopic.from_assets(
    ASSETS,
    credentials=credentials
)

In this case, each search term is queried individually, and then a weighted average is calculated

indicator               score  coverage     score  coverage
query                    NXPI      NXPI      PLTR      PLTR
          date_time                                                  
2025-02-04 06:00:00  0.056472  0.000333 -0.010778  0.000417
2025-02-04 09:00:00  0.051278  0.000194  0.039500  0.000417
2025-02-04 12:00:00  0.016750  0.000042  0.058500  0.000583

You can also apply z-score sentiment normalization. In this example, the z-score is calculated based on a 35-day rolling window.

from adase_api.schemas.sentiment import ZScoreWindow

ada_query = QuerySentimentTopic.from_assets(
    ASSETS,
    credentials=credentials,
    z_score=ZScoreWindow(window='35d')
)

By default is queried last 35 days of live data collection, but you can switch to historical data

ada_query = QuerySentimentTopic.from_assets(
    ASSETS,
    credentials=credentials,
    z_score=ZScoreWindow(window='365d'),
    live=False,
    start_date='2020-01-01',
    languages=['de', 'ro'],
    freq='1d',
    on_not_found_query='warn'
)

This query will retrieve data from 2020, in German and Romanian languages, apply a 1-year rolling z-score and return a daily data granularity

More about search query syntax

1. Plain text
   • Unlike keyword search, plain text relies on topics to query data based on broader concepts. It works best when 2-5 words describe a particular concept. Examples include:
     • "stock market", it might also analyse terms as "Dow Jones", "FAANG" etc.
     • "Airline travel demand"
     • "Energy disruptions in Europe"
     • "President Joe Biden"
   • analysed scope depends on how words normally co-occur together
     
     
2. Boolean search
   • Search for exact keyword match
   • Each condition is placed inside of round brackets (), where
     • + indicates a search term must be found
     • and - excludes it
   • For example "(+Ford +Motor*), asterix * will include both Motor & Motors

This query will do a boolean search on historical data starting from Jan 1, 2010 and include only data in specified languages

Key stats

Data coverage, missing or insufficient data

• Curated from over 5,000 sources
• Approximately 1,000,000 unique stories daily
• Available in 72 languages; both original and translated versions for querying in base English
• To ensure sufficient data per query, you have the filter_sample_daily_size option (refer to the code docstring for more details). If not enough data is found, those rows will be set to NaN. The calculation uses a rolling window, meaning some periods may have sufficient data while others may not.
• If no results are found, you’ll receive a Server HTTP 404 error. For multiple sub-queries, any missing ones will be ignored, and the results will be based on the found sub-queries.
• For more details, refer to the code docstrings

Data history

• Data available since January 1, 2019

API rate limit

All endpoints have a set limit on API calls per minute, with a default of 10 calls per minute.

Chat with data

You can also interact with data using LLM and integrate live news data feeds into your systems, although this is outside the scope of this particular package.

Questions?

• For package questions, rate limit or feedback you can reach out to info@adalytica.com
• You can also follow us on LinkedIn
• Or check some of our public research powered by this package data
• If this feels too complex, there's also a lightweight web app solution that provides access to sentiment data.