How to Tell Whether Your Forecasting Model Beats the Market: A Python WalkthroughBrier scores, the Murphy decomposition, a clustered bootstrap and a blending test, run on an open dataset of 5,654 timestamped model forecasts against Polymarket prices. If you build a probabilistic m7 min read·neuportal.hashnode.dev1d ago
7,426 forecasts against Polymarket: exactly where the market beat our modelsSince September our statistical models have made 7,426 probability calls on Polymarket questions. Each call is stored next to the market price at the moment it was made, and the file holding it is tim6 min read·neuportal.hashnode.dev4d ago
Nobody Pays For The Prompt: What Turns An AI Demo Into A Billable SystemA working chatbot demo and a support agent a company will pay a monthly invoice for are separated by operations, not by model quality. Here is the specific engineering that sits in that gap, and why your twentieth delivery is worth more than your first ten.7 min read·neuportal.hashnode.devAug 4
AI and Volatility: Forecasting How Much a Market Moves, Not Which WayAsk almost any machine pointed at a market the same question and it will answer confidently: where is the price going next. It is the question with the screenshots and the viral threads, and it is the8 min read·neuportal.hashnode.devJul 24
Calibration: the metric your ML dashboard is probably missing · tags: machinelearning, python, datascienceMost ML dashboards show accuracy. For any model that outputs probabilities, accuracy hides the thing that matters: is a "70%" really a 70%? That's calibration — of all the times a model said 70%, do ~1 min read·neuportal.hashnode.devJul 17
