Sport is full of people working for an opportunity. Every rating, ranking and recommendation describes part of that work. We owe them care in how we collect the evidence and explain what it means.
We believe a conclusion should come with enough evidence to question it.
Our principles
01—10We begin with a real question.
Who deserves a closer look? What does this team need? What has a player demonstrated, and what remains uncertain? The purpose of an analysis determines what we measure and what we can reasonably conclude.
We make the evidence traceable.
Sources, identities, dates, competitions and playing volume belong to the result. Official inputs are a starting point. The evaluation we build from them remains our responsibility.
We give performance its context.
League, role, workload and sample size affect meaning. Season performance, recent form, club fit and future projection answer different questions. We explain which one we are addressing.
We preserve what is unknown.
Missing data does not become poor ability. An unmeasured skill does not become a fact because a model can generate a number for it. Gaps should tell us what to investigate next.
We make uncertainty usable.
Readers should know when a result is provisional, when a sample is thin and when a conclusion depends on an assumption. Confidence must have a stated meaning. Precision must be earned by evidence.
We make the work repeatable.
We retain the inputs, versions, settings and outputs needed to revisit a calculation. We distinguish what can be explained publicly, repeated internally and independently reproduced. A proprietary engine still owes the reader an intelligible account of its claims.
We keep a record that can challenge us.
Forecasts must be evaluated against outcomes and relevant baselines. Misses, exclusions and inconclusive comparisons belong beside successes. Complexity earns its place by demonstrating value.
We correct the record.
Better evidence should change our conclusions. Corrections need to be identifiable, while original forecasts remain available for evaluation. Accountability is part of maintaining the product.
We use AI with responsibility.
It can help us research, calculate, inspect and communicate. The people producing an analysis remain accountable for its facts and reasoning. A persuasive paragraph cannot establish the truth of its own claims.
We respect human judgment.
Models can organize evidence and focus attention. Coaches, scouts and players bring observations our datasets may not contain. Our work should help them ask sharper questions and make better-informed decisions.
We want clubs to understand why a recommendation was made. We want players to understand the evidence behind an assessment. We want scouts to see where confidence ends and observation needs to begin.
That is the advantage we intend to earn: analysis people can inspect, challenge and use.
The same standard applies to every claim we make about Statbase.
Show the work. Earn the trust.