Quantitative research blog
Notes on machine-learning stock scoring, factor signals and what our daily model output shows across 1,400+ global equities.
- Your Purchase Price Is Not a Signal
— The price you paid for a stock tells you nothing about what it's worth today. Here's why anchoring on cost basis quietly destroys portfolio returns. - Sector Rotations Often Look Like Strategy. Sometimes They're Just a Stampede.
— Herding behavior is one of the most persistent forces in equity markets — and sector rotations are where it shows up most visibly. - Confirmation Bias Hits Harder When Everything Is Going Up
— Bull markets don't just reward optimism — they punish skepticism. Here's why that feedback loop makes confirmation bias uniquely destructive in rising markets. - Self-Directed Portfolios and the Control That Isn't
— Managing your own portfolio feels empowering — but the sense of control it creates is often more psychological than real. - Last Month's Chart Is Already Leading You Astray
— Recency bias quietly distorts investment decisions by making recent price action feel more predictive than it actually is. Here's the mechanism — and how to counter it. - Holding Losers: The Cognitive Trap That Quietly Destroys Portfolios
— Loss aversion doesn't just feel bad — it systematically distorts how long investors hold declining positions, often at serious cost. - Daily Re-Scoring vs. Quarterly Rebalancing: One of These Is Living in the Past
— Quarterly rebalancing made sense when data was scarce. In an era of daily ML scoring, holding stale signals for 90 days is a choice with real costs. - Overfitting: The Silent Killer of Quant Models
— A model that looks brilliant on historical data can fall apart in live markets. Here's how overfitting corrupts quant signals — and why it's so hard to catch. - More Data, Worse Predictions: The Feature Bloat Problem in Stock Models
— Adding more variables to a predictive model feels like progress. Often, it isn't. Here's what actually happens when stock models get overloaded with inputs. - Backtesting Looks Backward. Markets Don't.
— Backtesting is a necessary tool, but it has hard limits. Here's what the historical record genuinely cannot tell you about a strategy's future. - Wrong vs. Early: A Distinction That Separates Quants from Gamblers
— A score that doesn't pay off immediately isn't necessarily a bad score. Understanding the difference between wrong and early is fundamental to using quantitative signals well. - No Factor Works Forever — And That's the Point
— Value, momentum, quality — every factor model eventually hits a wall. Understanding why regimes break factors is more useful than chasing the one that worked last cycle. - Signals from Two Worlds: How a Hybrid Score Actually Works
— A hybrid ML score isn't just technical plus fundamental data averaged together. Here's the real architecture behind combining two very different signal types. - Data Isn't the Edge Anymore. Processing Is.
— When every investor has access to the same information, the advantage shifts to whoever can process it faster and more rigorously. - Institutional Research Costs Six Figures. Here's What That Gap Actually Means.
— Hedge funds spend hundreds of thousands annually on quantitative research tools. The gap between them and retail investors has always been about more than money. - Institutional-Grade Analysis: What Changed in 15 Years
— The analytical tools that once cost hedge funds millions are now accessible to private investors. Here's what that shift actually means. - Financial News Is Engineered for Clicks, Not Clarity
— 24/7 financial media is built to hold attention, not improve decisions. Here's what that costs you as an investor. - Momentum Is Real, But You're Probably Measuring It Wrong
— Momentum is one of the most robust factors in quantitative finance — and one of the most misapplied. Here's what the evidence actually says. - Why Your Brain Is a Terrible Portfolio Manager
— Cognitive biases don't just affect bad investors — they're hardwired into all of us. Here's what the research says, and what to do about it. - Why Most Retail Investors Are Flying Blind on Sector Rotation
— Sector rotation drives some of the biggest return differentials in equity markets — yet most retail investors only notice it after it's already happened. - Why Most Investors Misread Momentum — And What the Data Actually Says
— Momentum is one of the most documented phenomena in finance, yet most investors apply it backwards. Here's what quantitative research reveals. - Mean Reversion vs. Momentum: Why Your Model Needs to Know Which Game It's Playing
— Two of the most powerful forces in markets directly contradict each other. Here's how quantitative models learn to tell them apart. - Mean Reversion vs. Momentum: Why Both Camps Are Right
— Two of quant finance's oldest forces seem to contradict each other — but understanding when each dominates is what separates disciplined investors from the rest. - Why Your Brain Is a Terrible Stock-Picking Machine
— Human cognition evolved for survival, not markets. Here's why our instincts systematically fail us as investors — and what quantitative models do differently. - Why Most Investors Mistake Noise for Signal — and Pay for It
— Markets generate endless data, but most of it is irrelevant. Understanding the difference between signal and noise is one of the most valuable skills in modern investing. - The Tyranny of Recency: Why Your Brain Is a Terrible Backtester
— Recency bias is one of the most costly cognitive traps in investing. Here's how quantitative models are built to fight it. - The Tyranny of Recency Bias: Why Your Brain Is a Terrible Quant
— Recency bias makes investors overweight the latest data and underweight the long arc of evidence. Here's how systematic models fight back. - Why Quant Models Struggle When Markets Change Character
— Markets are not static equations. When volatility regimes shift, models trained on yesterday's patterns can quietly lose their edge — here's how adaptive scoring helps.
Content on this site is general and impersonal research — not investment advice. Past performance is not indicative of future results.