ZettaQuant

Our NeurIPS 2025 paper ("Words That Unite The World") introduced a unified framework for decoding Central Bank communications across 3 key dimensions. But at ZettaQuant, we realized 3 dimensions weren't enough. We’ve now expanded that research into 11+ specialized AI models for monetary policy labeling (with more shipping weekly). Instead of just telling you how the Fed sounds (Hawkish/Dovish), our API tells you exactly what is driving the conversation; from Inflation and Housing to Labor Markets and Money Supply. We're moving beyond simple sentiment scores to quantify the relevancy of every critical economic topic in real-time. If you're building a trading algo or macro dashboard and want to plug into this multi-dimensional analysis, DM me for a free API key. (Links to the models and paper in the comments 👇) #Fed #MonetaryPolicy #NLP #FinTech #AI #MacroEconomics Huzaifa Pardawala Siddhant S. Sudheer Chava

Based on this morning’s breaking news around Kevin Warsh being picked as the next Fed Chair nominee, many of us (at least me) are naturally wondering how his views actually differed from other FOMC members when the most market-consequential decisions were being made. So I went back to the data. I ran ZettaQuant’s new central-bank stance classification model on 40 FOMC meeting transcripts where Kevin Warsh participated as a Governor between 2006–2011, a period that notably includes the Global Financial Crisis. Key insight: - Kevin Warsh was, on average, more hawkish than the overall Fed consensus. While both are on the dovish side in absolute terms, the gap indicates Warsh consistently leaned toward tighter monetary policy relative to his peers. This divergence is especially visible around 2010, his final year on the Board. Curious to hear how others interpret this, especially in the context of today’s macro environment and forward-looking policy expectations. #FederalReserve   #MonetaryPolicy  #FOMC  #MacroEconomics   #CentralBanks  #FinancialMarkets Huzaifa Pardawala, Siddhant S., Sudheer Chava

Hiring momentum has cooled since the 2022 high. Lately, I’ve been hearing the same thing from friends across tech and finance: the job market feels unusually tough. This chart helps quantify. Using earnings call transcripts from S&P 500 companies, we track how firms talk about hiring. Normalized to 2022 (=100), hiring momentum is now ~21% lower. What’s striking: this analysis took just a few clicks using ZettaQuant’s app; no surveys, no manual labeling, just signals directly from earnings calls. This also lines up with recent macro signals. Markets are increasingly pricing in delayed rate relief (e.g., low odds of a January cut on Polymarket), alongside jobs data showing rising unemployment. What’s driving this? Overly tight monetary policy? AI changing hiring needs? Something else?

The world's most innovative startups run on Snowflake - just ask ZettaQuant. Built on research published at NeurIPS, the team offers financial intelligence at enterprise scale. Agam Shah Siddhant S. And Huzaifa Pardawala are building AI Powered Financial Intelligence and they have products already consumable on the Snowflake Marketplace ready for macro analyses, investment management and retail sentiment analysis. We're hanging out at NeurIPS all week, come meat the team!

Is AI a bubble? Before you answer, have a look at the work that, for example, ZettaQuant is doing. Accurate, fast, differentiated insights for quants, built to run securely and natively inside Snowflake. Some will use this technology to get an edge, others will dismiss AI as a bubble. It's a free country, follow whichever path looks more appealing. JR

I’ll keep this one short because the numbers speak for themselves. We just compared ZettaQuant’s ZQ_Classify (deployed as a Snowflake Native App) to Snowflake’s AI_Classify and Databricks’ AI_Classify on financial-domain classification tasks. Result: ZQ_Classify (on Snowflake) > Snowflake AI_Classify > Databricks AI_Classify across every dataset we tested. If you care about accuracy, cost at scale, and enterprise-grade reliability for financial text analytics, you’ll want to see this chart.

ZQ × Tariffs: Turning Tariffs Related Earnings Calls Discussions Into Signals Tariffs are front-and-center in markets right now, impacting markets and being mentioned in meetings, announcements and releases. We use the S&P 500’s companies’ latest earnings calls as our dataset. Using ZettaQuant’s (ZQ) pipeline and actionable intelligence platform, we quantified tariff sentiment and its impact across public companies: Methodology 1. Filtered our entire earnings-call corpus for tariff-related sentences using ZQ’s proprietary topic filtering model. 2. Labeled tariff-related sentences using ZQ’s proprietary classification framework: [Positive impact, Negative impact, No impact] 3. Constructed a company-level metric: TARRIF_DISCUSSIONS (Measures each company’s stated sentiment regarding tariffs) Insights: - Top 5 companies positively positioned on tariffs - Top 5 companies negatively positioned on tariffs - Compared each group’s YTD stock performance The results displayed below show our generated metric vs the company’s YTD performance and demonstrate how ZQ can help unearth actionable insights from unstructured data. Want to generate signals that validate your own market view? Reach out to us contact@zettaquant.ai. This is just one application of ZQ’s unstructured data-to-signal engine for financial research and investing. Disclaimer: This material is for informational purposes only and is not financial advice. Nothing herein constitutes investment recommendations or predictions. Past performance is not indicative of future results. Huzaifa Pardawala Siddhant S. Sudheer Chava #EarningsCalls #Tariffs #UnstructuredData #FinancialSignals #QuantResearch #EarningsAnalysis #FinancialAI #EquityResearch #DataDrivenInvesting #SnowflakeAI #EnterpriseAI #AssetManagement #MarketInsights