Knowing how data pipelines work, how analysis happens, and how results are presented is a specialised skill that every PM working with data must develop.
Heckyl was an ambitious product from day one — a news analytics suite built out of India in 2010, when such ventures were rare and difficult. It evolved from news analytics to price analytics and fundamental analytics, combining all with open datasets to help investors and traders make informed decisions. The actual job of a PM in such a fintech analytics company is to understand the entire data flow — from collection through analysis to presentation — and ensure the insights truly serve the user’s needs.
The stakes are high. Trading is volatile. One event anywhere in the world can cause massive market swings. Heckyl’s value lies in curating massive volumes of unstructured data and turning it into actionable financial insights for business professionals. If you want to work in fintech or any data-heavy product, mastering these skills is essential.
Heckyl’s origin and mission
Heckyl Technologies was founded in 2010 by four former Merrill Lynch executives — Abhijit Vedak, Jaison Mathews, Mukund Mudras, and Som Sagar — who worked in fixed income. Having been on the investor and trader side themselves, they saw the opportunity to revolutionize the trading industry by adding real-time, data-driven insights for traders, investors, brokerage firms, and fund managers.
The inspiration came from the explosive growth of social media and open data sources. They realized these trends could add value to financial decision-making, moving beyond just stock prices to include news, sentiment, and global market trends.
Heckyl continuously scans millions of sources — news sites, blogs, forums — to aggregate financial data about companies and markets. This breadth of data collection is a core competitive advantage. Unlike Bloomberg or Thomson Reuters, Heckyl offers its services at a fraction of the cost, serving 14 clients with data from over 1.5 million sources. It recently secured $1 million in funding from Seedfund and maintains positive cash flow.
The fintech context in India
Fintech in India is on a strong growth trajectory, driven by mobile technology, cloud computing, and big data. In 2016, fintech startups attracted $17.4 billion in investments globally, with India contributing significantly. The sector is projected to grow at a CAGR of 17%, supported by government initiatives for financial inclusion and digitization.
This environment creates both opportunity and pressure for fintech analytics platforms like Heckyl. They must innovate rapidly while managing complex data challenges unique to India’s market.
The actual job: Managing data for financial insight
The core challenge Heckyl faced was connecting vast, unstructured data to meaningful financial insights. Collecting data from millions of sources creates enormous complexity and risk.
Data quality is paramount. The platform must filter noise and ensure only relevant, accurate information reaches users. Unstructured data from social media or forums can be misleading or erroneous if not properly validated.
Data velocity matters. Financial markets move fast. Heckyl must maintain the speed of ingestion and processing to provide real-time insights. Delays can make the data useless.
Security and privacy risks are real. Handling sensitive financial data from multiple sources requires safeguarding against leaks or breaches. The tools used for analysis must be robust and secure.
Heckyl’s offering includes news analysis, market analysis, sentiment analysis, global market trends, and predictive data analytics. Each of these relies on different data types and analytical approaches.
Defining the data to collect
If you were the PM at Heckyl, your first step would be to define the data necessary to deliver value. This includes:
- News data: Headlines, full articles, blogs, social media posts, and forums mentioning companies, sectors, or markets.
- Market data: Real-time price quotes, trading volumes, order book data for equities, commodities, and currencies.
- Sentiment data: Natural language processing to classify news and social media as positive, negative, or neutral sentiment regarding financial instruments.
- Fundamental data: Company filings, earnings reports, macroeconomic indicators, and other structured datasets.
- User behavior data: How clients interact with the platform, what reports they generate, and which alerts they subscribe to.
Collecting these data types requires integrating with APIs, web scraping, partnerships with data providers, and building internal ingestion pipelines.
Data collection strategies
The data strategy at Heckyl must balance coverage, quality, and timeliness:
- Source prioritization: Focus on high-impact, reliable sources for critical market-moving information.
- Automated filtering: Use machine learning models to discard irrelevant or duplicate data.
- Normalization: Standardize data formats across diverse sources to enable aggregation and comparison.
- Real-time streaming: Implement event-driven pipelines to process data with minimal latency.
- Privacy compliance: Ensure data collection respects legal constraints and client confidentiality.
A layered approach — starting with broad data ingestion, followed by filtering, enrichment, and finally presentation — is essential to manage scale and complexity.
Predictive analytics model variables
To build predictive analysis, Heckyl’s PM must identify independent and dependent variables:
-
Independent variables (inputs):
- News sentiment scores (positive/negative/neutral)
- Volume of news mentions per company
- Market price changes in related sectors
- Macroeconomic indicators (interest rates, inflation)
- Trading volumes and order book imbalances
- Social media buzz metrics
-
Dependent variables (outputs):
- Predicted stock price movement (up/down/neutral)
- Volatility forecast
- Risk scores for equities or sectors
- Expected trading volume spikes
Using these, a regression model can be constructed to forecast price changes or market trends. For example:
PriceChange(t+1) = β0 + β1 * SentimentScore(t) + β2 * NewsVolume(t) + β3 * SectorPriceChange(t) + β4 * MacroIndicator(t) + ε
Where β coefficients are learned weights, and ε is the error term.
Measuring success with web analytics
Success for Heckyl is not just delivering data but ensuring users find insights actionable and return frequently. Metrics include:
-
User engagement:
- Number of daily active users (DAU)
- Session duration and frequency
- Number of reports generated per user
-
Feature usage:
- Adoption rate of sentiment analysis dashboards
- Click-through rates on alerts or notifications
- Use of predictive analytics tools
-
Conversion metrics:
- Trial-to-paid conversion rates
- Churn and retention rates
-
Performance metrics:
- Data freshness (latency from source to display)
- System uptime and error rates
Using analytics to refine financial insights daily
Web analytics also inform product adjustments and data quality improvements:
- Track which data sources drive the most engagement and focus on improving their accuracy.
- Analyze user feedback on alerts to reduce false positives or irrelevant notifications.
- Monitor latency spikes to optimize data pipelines.
- Use A/B testing for UI changes or new features like heat maps or trend visualizations.
- Adapt the weighting of variables in predictive models based on user outcomes and feedback.
The PM’s specialized skillset in fintech analytics
The field is specialized. You are not expected to be a data scientist or analyst, but you must understand:
- How data pipelines ingest, clean, and transform data.
- How analysis algorithms produce outputs from raw data.
- How results are presented meaningfully to users.
- Trade-offs between data volume, velocity, and quality.
- Security and privacy implications of financial data handling.
These skills are critical for roles at companies like Heckyl and WebEngage, especially in India’s rapidly growing fintech ecosystem.
Test yourself: Heckyl data strategy challenge
You are the PM at Heckyl in 2017, responsible for expanding the platform’s predictive analytics capabilities. You have access to 1.5 million data sources but limited engineering bandwidth to process them all in real time. Your CEO wants to add a new feature showing real-time sentiment heat maps for equities, commodities, and currencies. You must decide which data sources to prioritize and how to balance data freshness with quality.
The call: How do you define the data collection strategy to support the new feature? What trade-offs will you communicate to leadership?
Your reasoning:
Where to go next
- Understand how to discover user needs and pain points: User Research Methods
- Learn to translate data insights into product decisions: Metrics and KPIs
- Explore product strategy frameworks for fintech: Product Vision and Strategy
- Develop skills in managing data privacy and security: Ethical PM