Forecasting • Causal Inference • Python / R / SQL / Stata
This page tracks my ongoing forecasting practice across a range of domains. Some questions are recreational (e.g., chess tournaments), while others involve structured modeling and quantitative analysis. The goal is to demonstrate calibration, probabilistic reasoning, and consistent updating — the core skills behind forecasting in applied economics, decision science, and technical environments. Forecasts are updated weekly as new information arrives.
All forecasts are submitted to structured prediction platforms and updated weekly as new data arrives. Calibration record tracked on Metaculus.
Silver entered 2026 in an unprecedented structural regime — US Critical Minerals designation, China export controls, and AI/solar industrial demand converging simultaneously. With silver at $92.68 and gold at $5,230, a macro-adjusted Monte Carlo simulation (10,000 paths, gold β=2.29, R²=0.72) assigns only ~10% probability the April high stays below $97. The community median of $95.2 implies near-zero upside from current prices.
Full methodology and model output →
An 8-player double round-robin in Cyprus, March 29 – April 16. Rather than relying on Elo alone, this forecast uses a weighted scoring model emphasizing recent classical form (30%) and head-to-head record vs the field (30%). The biggest departure from the community: Nakamura down 10 points due to 8 months of inactivity against sub-2100 opposition. Praggnanandhaa up 6 points as the dominant classical performer of 2025.
Full methodology and model output →
No resolved forecasts yet. Once a question closes, resolved entries move here with a post-mortem.