https://winstonbartle.com Building the intelligence layer for modern finance — applying Machine Learning, NLP, and Deep Learning to reshape how institutions understand and manage risk. en Wed, 23 Sep 2026 12:36:37 GMT Next.js Static RSS https://winstonbartle.com/blog/roc-auc-pairwise-vs-threshold https://winstonbartle.com/blog/roc-auc-pairwise-vs-threshold Wed, 19 Aug 2026 00:00:00 GMT The pairwise interpretation of ROC AUC — 'probability a random positive outranks a random negative' — isn't just a nice intuition pump that happens to give the same number as the curve. It's mathematically the same computation. Here's why, worked through an 8-point example both ways. Machine LearningStatisticsModel EvaluationROC AUCClassification https://winstonbartle.com/blog/logistic-regression-class-imbalance-intercept-correction https://winstonbartle.com/blog/logistic-regression-class-imbalance-intercept-correction Mon, 27 Jul 2026 00:00:00 GMT Fraud, default, churn, cancer — the outcome you actually care about is almost always the rare one. SMOTE and class weights fix the training problem, but they quietly break your predicted probabilities unless you correct the intercept back afterward. Here's the fix, from King & Zeng (2001). Logistic RegressionClass ImbalanceSMOTEStatisticsRisk ModellingMachine Learning https://winstonbartle.com/blog/cohens-h-effect-size https://winstonbartle.com/blog/cohens-h-effect-size Thu, 02 Jul 2026 00:00:00 GMT Wilson confidence intervals are a great governance tool for monitoring behavioural assumptions — but at large sample sizes they get so tight that trivial, immaterial drift starts failing the check. Cohen's h fixes that by measuring the size of the gap, not just whether it's detectable. StatisticsEffect SizeRisk ModellingBehavioural ModellingIRRBB https://winstonbartle.com/blog/mle-optimizer-comparison https://winstonbartle.com/blog/mle-optimizer-comparison Thu, 02 Jul 2026 00:00:00 GMT Fitting a model by maximum likelihood almost always means calling scipy.optimize under the hood. BFGS, L-BFGS-B, Nelder-Mead, Powell, and basin-hopping all claim to find 'the' maximum likelihood estimate — but they make very different assumptions, and picking the wrong one silently hands you a local optimum instead of the answer you wanted. StatisticsOptimizationMaximum LikelihoodPythonSciPy https://winstonbartle.com/blog/dijkstra-algorithm https://winstonbartle.com/blog/dijkstra-algorithm Sun, 07 Jun 2026 00:00:00 GMT In 1956, Edsger Dijkstra designed an algorithm in his head, in 20 minutes, without pen or paper. It still powers Google Maps, internet routing, and game AI today. A complete guide to the mechanics, intuition, and limitations of one of computing's most elegant ideas. AlgorithmsGraph TheoryComputer SciencePathfindingData Structures https://winstonbartle.com/blog/monitoring-static-behavioural-assumptions https://winstonbartle.com/blog/monitoring-static-behavioural-assumptions Tue, 02 Jun 2026 00:00:00 GMT Static behavioural assumptions power ALM and IRRBB models, but they drift. Wilson score intervals give you a statistically sound, auditable way to monitor when they need recalibration — without building a full dynamic model. IRRBBALMRisk ModellingBehavioural ModellingStatistics https://winstonbartle.com/blog/irrbb-pass-through-rates https://winstonbartle.com/blog/irrbb-pass-through-rates Sat, 23 May 2026 00:00:00 GMT When market rates move, customer rates rarely follow one-for-one. Pass-through rates explain the gap, the timing, and the IRRBB risk hiding inside margin forecasts. IRRBBRisk ModellingBanking https://winstonbartle.com/blog/pandas-outdated-polars-pyiceberg-2026 https://winstonbartle.com/blog/pandas-outdated-polars-pyiceberg-2026 Tue, 31 Mar 2026 00:00:00 GMT After years of being the default, Pandas is hitting its limits on modern data lakehouse workloads. Here's why I switched to Polars + PyIceberg — and why you might want to as well. PolarsPyIcebergData EngineeringApache IcebergPerformanceAWS Glue https://winstonbartle.com/blog/pca-explainer https://winstonbartle.com/blog/pca-explainer Sat, 15 Nov 2025 00:00:00 GMT PCA rotates a cloud of correlated variables into a new set of uncorrelated axes ordered by variance explained. A complete geometric, mathematical, and practical guide with live interactive 3D projections, standardization demos, scree plots, loading heatmaps, biplots, and more. PCADimensionality ReductionMachine LearningStatisticsFinanceRisk Modelling https://winstonbartle.com/blog/variance-inflation-factor https://winstonbartle.com/blog/variance-inflation-factor Mon, 20 Oct 2025 00:00:00 GMT The diagnostic that reveals when your predictors are linearly dependent — and why it quietly destroys the reliability of your standard errors, confidence intervals, and t-tests. EconometricsRegression DiagnosticsMulticollinearityStatisticsVIF https://winstonbartle.com/blog/quantum-inspired-gradient-descent-qigd https://winstonbartle.com/blog/quantum-inspired-gradient-descent-qigd Fri, 12 Jul 2024 00:00:00 GMT Traditional optimization methods often get stuck in local minima. Quantum-Inspired Gradient Descent (QIGD) offers a promising way forward by borrowing ideas from quantum computing — without needing actual quantum hardware. OptimizationQuantum ComputingMachine LearningPortfolio OptimizationRisk ManagementGradient Descent