OddsCrew is an independent sports analytics project focused on transparent, data-driven tools for analyzing sports markets. The current research program begins with NFL player projections and ongoing forward testing.
OddsCrew is being built as an independent sports analytics platform focused on transparent model projections, market comparison, and measurable forward performance.
Build the model. Publish the projection. Freeze the record. Grade the result. Learn from the evidence.
Start with historical performance, current market information, and measurable player-level data rather than narratives alone.
Test ideas against historical and forward results. Features earn their place through evidence rather than assumption.
Freeze projections before outcomes and preserve the results afterward, including observations that do not make the model look better.
A projection is only meaningful when it can be evaluated against something. OddsCrew uses the sports market as a benchmark rather than pretending it does not exist.
Ask whether the model was more accurate than the market benchmark. That comparison is central to how OddsCrew evaluates its research.
Measure model projections alongside the frozen market.
Test the system on results that were not known when projections were created.
Use accumulated evidence to guide future research rather than rewriting past results.
OddsCrew's forward-testing process is designed so that published observations remain inspectable after the games end.
Projections and qualification status are recorded before results.
Official results are compared with both the model projection and the market line.
Historical observations remain in the permanent record instead of disappearing after a poor result.
Performance is judged across the growing sample rather than isolated outcomes.
OddsCrew does not present qualified projections as guaranteed outcomes. The purpose of forward testing is to determine what the evidence actually supports over time.
OddsCrew is starting narrow so the underlying research process can be tested before the platform expands.
Forward-tested player projections, market comparison, qualification criteria, and permanent performance tracking.
Expand the analytics framework beyond the initial player markets as sufficient evidence and infrastructure develop.
Adapt the research process to additional sports only after the core workflow is established.
Build richer research, comparison, tracking, and visualization tools around the growing dataset.
The model, methodology, and historical results are designed to be inspected rather than hidden behind claims.