Last 5 or last 35: separating form from noise

The last five games can quickly reveal changes in rotations or style, but they form a very small sample. The last 35 are more stable, yet may include an outdated playing pattern. These windows should therefore provide separate evidence, rather than one blended average.

Three time windows

The model compares short-term form over 5 games, medium-term form over 10–15 and a baseline of up to 35. Agreement across all three increases confidence. If the last five differ sharply, the system looks for an explanation: the roster, schedule, coaching decisions or unusual shooting.

Separate home and away records

Home and away periods are not pooled without an adjustment. Pace, opponents' eFG%, free-throw frequency and the way close games finish matter in the relevant venue role. A small home sample must be pulled towards the overall profile.

What the model checks

• The last five do not replace the baseline
• The home or away role matches
• A change in playing pattern has an explanation
• Statistical shrinkage is applied to small samples

Analysis describes probability and risk; it does not guarantee outcomes. This material is intended for readers aged 21+.