What My Z-Score Indicator Taught Me About Chasing Better Numbers
What moving from HMA to KAMA taught me about Z-scores, backtest overfitting, market regimes, and the limits of better-looking results.
The collected writing
The archive.
Technology, markets, and the ideas that connect them. Explore the complete collection.
AI infrastructure: separating investment risk from a bubble story
How to examine AI capital spending, depreciation, cash flows and power constraints without treating a plausible risk scenario as a proven collapse.
From market mood to Regime Analytics
The retired economic-dashboard prototype that helped shape Erika Barker’s interest in predictive analytics and statistical market regimes.
Palantir and the price of moving faster
My military procurement experience helps me understand Palantir’s appeal. It also leaves me asking what faster decisions cost democratic accountability.
What the data says about the American middle class
How to interpret changes in middle-income population share without confusing income, wealth, money velocity or a prediction of national collapse.
Gradient boosting: how the trees learn together
An explanation of boosting, loss functions and regularization, with practical limits on interpretability and financial-model evaluation.
Miami housing risk: what the data can actually show
How to evaluate Miami housing affordability, inventory and local economic conditions without inventing a crash probability.
Comparing the US and China: measure before forecasting
A framework for comparing economic output, financial markets and cross-border investment without treating them as interchangeable measures.
Composite market scores: useful summaries, fragile assumptions
How to normalize market indicators, expose weights and test a composite without confusing ETF prices, overlapping holdings or retrospective fit.
Quantitative analyst vs. data scientist: compare the work
How quantitative finance and data science overlap, where their objectives differ, and what evidence makes a portfolio useful to a hiring team.
Editorial standards and how this site uses AI
How Erika Barker’s site distinguishes sources, analysis, experiments and AI assistance, and how substantive corrections are recorded.
My HMA Z-score indicator: distance is not probability
A project note on standardizing a Hull moving average, the limits of normal-distribution interpretations and what a trading validation would require.
What running taught me about doing enough
Running helped me manage anxiety, ADHD and the isolation of the pandemic. Then I learned when a useful routine could become another way to avoid my life.
Conformal abstention: when a language model should decline
What the 2024 conformal-abstention paper proposes, how calibration enters the method, and why statistical error control is not a guarantee of truth.
Kernel methods in machine learning and finance
How kernels represent similarity, what an RBF kernel does, and how to evaluate a nonlinear model without confusing a flexible fit with predictive evidence.
Why I like the Hull moving average
How Alan Hull’s moving average combines weighted averages, where the square-root window enters, and why a responsive line is not a trading edge.
What the Riemann hypothesis actually says
An introduction to the zeta function, prime numbers and the limits of what a proof of the Riemann hypothesis would establish.
Markov chains, explained with a two-state example
A complete transition-matrix example, a stationary distribution and the assumptions that limit Markov models in economic and financial research.
Game theory and markets: what Nash equilibrium explains
A practical introduction to strategic incentives and Nash equilibrium, with a clear boundary between public mathematics and conjecture about private trading strategies.
Understanding the economy without misleading plumbing
A guide to GDP accounting, money, credit and inflation that separates useful metaphors from causal claims.
Making AI-assisted data work easier to verify
A practical workflow for dividing research tasks, checking generated code and distinguishing a model’s answer from evidence.
The law of large numbers—and the stories it cannot prove
Why sample averages can stabilize, why repeated trials do not erase selection bias, and how a perfect prediction record can be manufactured.
How binary represents numbers, text and instructions
Bits do not come with built-in meaning. A short guide to binary numbers, character encodings and the role of conventions in computing.
No articles found.
Try a different phrase or browse all subjects.