statistics.com

I am working with a very successful family office in their build out of a new multi-strat hedge fund (NYC). They are looking to hire a data scientist/analyst to be their fundamental PM's right hand person, influencing the best investment decisions as possible through data analysis, development, and discovery of market anomalies and opportunities to maximize revenue generation and of course, AUM. This person should have a strong professional (from a top hedge fund or trading firm) and academic background in math, analytics, and data science. Compensation will be very competitive including base and bonus. If this sounds like a strong fit, apply below and send me a message. https://lnkd.in/e52PD2Ui #hedgefund #datascience #quantanalyst #quanttrading #fundamentalequities #Longshort #dataanalyst #dataanalytics

Predictive models can fail quietly...and expensively. Dr. John Elder has spent 30+ years helping teams build models that actually work in the real world. On February 11, he’s sharing the hard-won lessons.  Join us in Charlottesville, VA, for a workshop packed with tips for real-world success in predictive analytics.  Topics include:   💡 How to choose the right modeling techniques and understand their strengths  💡 Tips to make your models more reliable  💡 Industry best practices   💡 Popular machine learning methods  💡 How to achieve results with new data  Plus, you’re in for some laughs as John shares engaging, relatable stories from over the years.  Whether you’re just starting out or looking to sharpen your skills, you’ll leave with practical tips you can put to work right away. Learn more and sign up: https://lnkd.in/eZpuAbEF  We’re proud to present this workshop in partnership with AFCEA CeVA and Charlottesville Business Innovation Council (CBIC).

If you're tired of AI webinars that overpromise and underdeliver, this one’s for you. Join a panel of analysts, scientists, and business leaders from Elder Research who have seen what works—and what doesn’t—when it comes to real-world AI. 🧠 Not Your Typical AI Webinar: 4 Practical Tips for Real Results 🗓️ January 21, 2026 | 🕑 2:00pm ET | 💻 Virtual 🔗 Register here: https://hubs.ly/Q03VzYlv0 👉 We’re skipping the buzzwords and getting straight to it: → Smarter questions to ask before your next AI build → ROI techniques that tie AI work to business outcomes → How to align business, data, and learning teams → Lessons from the field—what’s actually working in practice 🎤 Speakers from Elder Research include: • Stephen Assink, Technical Business Analyst • Shaylee Davis, Data Scientist • Josh Fairchild, Senior Technical Business Analyst • Leah Severance, Director of Commercial Business Development • Ramzi Ziade, Learning Engagement Manager If you’re leading or supporting AI delivery, this is 60 minutes worth showing up for.

We’re starting a new chapter. We’re proud to announce that Elder Research is joining MANTECH through an acquisition that amplifies our efforts to help clients see it solved. MANTECH is at the forefront of AI and mission-critical technology solutions, and we’re excited to combine our three decades of experience with theirs to continue helping clients solve their toughest data challenges. Hear what our founder John Elder and CEO Gerhard Pilcher have to say about this exciting new chapter: https://lnkd.in/eGA2iWbz

This past week at Elder Research: Hurricane predictions! 1. “How we're supporting better tropical cyclone prediction with AI”: https://lnkd.in/eAGmfreW Google DeepMind have a paper out and a new Weather Lab website (https://lnkd.in/eYBAAP3t) with interactive views into model predictions. I’ve spent a lot of time on the North Carolina coast, so this seems especially relevant! 2. “AI-assisted coding for teams that can't get away with vibes”: https://lnkd.in/ernJ_gna A useful guide for working with LLMs in more rigorous settings, not just for hobby projects. I’m still ambivalent about AI for writing code—I don’t think I’ve achieved anything like 10⨉ improvement in perceived efficiency, yet—but I find these kinds of practical write-ups helpful. 3. “Don’t Flip Out!”: https://lnkd.in/eNhC853E We occasionally pull puzzles from The Fiddler for our Commercial practice’s monthly “Datathon,” and I really liked this one we did recently. It’s an interesting example of how a game that looks fair on the surface can end up being unfair or asymmetric.

Are you still using linear regression for everything? Let’s rethink that. Linear regression is great—but not if: *Your response variable is a count *You're predicting a proportion *Or working with positive continuous values (like time-to-event) 🔍 Enter: Generalized Linear Models Example use case: A team analyzed years of air quality data alongside hospital records. Using Poisson regression, they discovered that a small increase in particulate matter led to a measurable spike in asthma-related ER visits. 💥 Insight like that doesn’t just win awards—it saves lives. GLMs aren’t just statistical tools—they’re engines of discovery in public health. 📌 Save this post for later if you're working on: >Clinical research >Environmental epidemiology >Chronic disease modeling 💬 Have a use case in mind where a GLM might work better? Drop it below!

🧠 What if your data doesn’t follow a bell curve? Not all outcomes are nicely distributed like a textbook Gaussian. Think counts, proportions, or even just yes/no outcomes. That’s where Generalized Linear Models (GLMs) shine! They allow us to model a wide variety of response types using a flexible statistical framework that adapts to the shape of real-world data. For example: 📊 Poisson regression can model the effect of air pollution on the incidence of respiratory disease (count data). ✅ Logistic regression assesses the impact of diet on coronary heart disease (binary outcomes). 🔁 Mixed-effects GLMs handle repeated measures, like tracking disease progression over time. Bottom line: GLMs are more than math—they are a critical tool for predicting health outcomes and informing public health decisions. 👉 Curious how GLMs could apply to your work? Let’s connect in the comments 👇

Last week at Elder Research: Using LLMs for what they’re good at. 1. “What I’ve learned about writing AI apps so far” https://lnkd.in/eG_9X7Z4 This matches my own experience with language models (including so-called “agentic” applications, like Cursor). The models are getting better and better at comprehending and distilling text, but it feels like we’re in the middle of a rush to replace more or less demonstrably correct (or at least deterministic) systems with shiny, but stochastic, AI. “Have the LLM do as little as possible” is a good take—and a good rule of thumb. 2. The Unsure Calculator https://lnkd.in/egSamj33 This is such a great idea, and with really good notation. I wish I’d come up with it myself. I wonder whether there’s a user interface that would help it make sense to generalize from Gaussian distributions. Also see the more powerful notebook version (https://lnkd.in/eP8Vn32u). 3. What do LLMs ‘value’? https://lnkd.in/eEiCPUZN From Matt Bezdek: Anthropic is doing quite a lot of interesting work to understand these language models from a higher-level, conversational perspective. The data set has also been made available under an open license: https://lnkd.in/ei6F55Cu 4. GPT 4.1 prompting guidance. https://lnkd.in/eCpRPFMU Lots of really good (and long!) examples of how to prompt GPT 4.1 well, taking advantage of its larger context size and wading into “agentic” orchestration and tool use. Really helpful as we’re doing some of this with web search. 5. Some Python regular expression tips https://lnkd.in/eNHU29Jc TIL you can pass functions into Python regular expression substitution calls!