Join Vinesh Jha, founder of Extract Alpha and former director at PDT Partners, as he dives into the tumultuous lessons from the 2007 Quant Quake. He reveals the inside story of PDT's $500M loss and the cultural differences with other hedge funds. Vinesh discusses the rise of alternative data, how smaller firms can thrive against giants, and the impact of AI on quant finance careers. He also explores the intricacies of predictive modeling and the importance of adaptability in today's trading landscape.
47:03
forum Ask episode
web_stories AI Snips
view_agenda Chapters
auto_awesome Transcript
info_circle Episode notes
volunteer_activism ADVICE
Build Investor-Ready Alternative Data
If you want to avoid building the same models as everyone else, seek alternative data sources and learn how investors need data served.
Build datasets with point-in-time, tickerized, and featurized formats so funds can use them immediately.
insights INSIGHT
LLMs Will Mature, But Not Overnight
LLMs and NLP will likely commoditize over time as implementation converges, but currently many model-source combinations remain unexplored.
Signals tend to get crowded slowly, then may experience sudden liquidation events like the QuantQuake.
insights INSIGHT
Earnings Transcripts Predict Long-Horizon Returns
Earnings-focused text models can predict returns at multi-week to multi-quarter horizons when features and model combos are engineered carefully.
In this episode of Odds on Open, Ethan Kho sits down with Vinesh Jha, founder of Extract Alpha and former director of PDT Partners, to unpack lessons from the 2007 Quant Quake and how systematic investors can adapt in today’s crowded landscape.We cover:- What really happened inside PDT Partners when the firm lost $500M during the Quant Quake- Why so many quant hedge funds blew up in 2007 — and the key financial crisis lessons still relevant today- Inside the culture at PDT Partners vs the siloed world of other hedge funds- Why Vinesh Jha left the buy side to start Extract Alpha — and how alternative data reshaped quant finance- The rise of earnings transcript models, analyst accuracy signals, and Estimize’s crowdsourced forecasts- Will today’s LLMs and NLP models in finance get commoditized like old factor strategies?- The trade-offs between running a hedge fund and building a data company- How smaller systematic funds can still compete with giants like Citadel, Millennium, and DE Shaw- What it’s really like to work as a quant — and the traits that make a good quantBonus: - How quant hedge funds find alpha using alternative data and NLP- How hedge funds use earnings expectations and post-earnings drift to trade- What lessons can quants learn from market crashes and black swan events?00:00 Intro01:00 Inside PDT Partners during the Quant Quake05:11 How quants decide when models fail08:49 Culture at PDT vs other hedge funds10:38 Why Vinesh founded Extract Alpha15:25 Financial crisis lessons: crowded quant trades16:20 Will LLMs and NLP in finance get crowded?18:53 Best alternative data sets for alpha24:54 Do Estimize crowdsourced forecasts make money?28:19 Can buzzwords like AI predict returns?32:02 Why Vinesh didn’t start a hedge fund35:37 How quants should reinvent mid-career38:51 AI disruption vs creativity in quant finance40:48 Can small funds compete with Citadel, Millennium, DE Shaw?43:30 What makes a good quant stand out46:54 Closing thoughts on longevity in quant financeWhether you’re deep into quant finance, researching hedge fund strategies, or simply curious about what makes a good quant, this conversation offers rare insight into how edge is found—and lost—in modern markets.
Odds on Open