Portfolio & tool shed

Selected project / Personal research

Sports Research & Validation

An evidence-first research workflow for recording forecasts, checking inputs, and keeping evaluation honest.

Cole Hover · AI-assisted development & review · Public showcase updated November 8, 2026

The question

Can the result be evaluated fairly?

A forecast is only useful to research if its inputs, timing, and outcome can be reconstructed. This spreadsheet workflow records forecasts before an event, checks the evidence, and keeps predictions separate from sporting results and cash records.

What was built

Versioned research records, exact event and market matching, missing-input and timing checks, and an evaluation design using Brier score, log loss, and paired comparisons with a market benchmark.

What has been checked

The reviewed workbook reports 28 passing synthetic formula checks. Four direct-form tests are also documented. Those records support bounded calculation and control behavior; they were inspected for this showcase, not rerun here.

What remains unproven

The current validation record contains zero eligible paired observations and is marked not validated. There is no demonstrated forecasting accuracy or profitability. The public walkthrough below illustrates a few controls; it is not the original engine or a prediction tool.

Try the controls / fictional example

Evidence before evaluation.

Change one condition and see why the research record passes a check or needs work. No live data, odds, stakes, or recommendations.

Fictional research record A

Event and market
Matched
Required inputs
Complete
Source timing
Before review
Original forecast
Recorded before event

Control result

Paper research qualified

The illustrative input and timing checks pass. This permits a research record to move forward; it does not validate the forecast or authorize a wager.

A passing control is a process check, not an edge.

Take the process with you.

The reusable brief and review checklist live in the same library as the models and trackers.

Educational research only. No investment advice, betting recommendations, or promised outcomes. Disclosures.