Baseline
Season-long efficiency and player history tell us what usually happens and what deserves attention.
Analytics
Analytics
Start with a football question. Use the numbers to frame the answer. Then check what actually happened on the field.
MELBOURNE • REALITY CHECK49ers 27 • Rams 7100,021 at the MCGStrong pregame evidence still has to survive one gameBack to current NFL coverage →
WHAT THE GAME REMINDED US
A model is a question, not a result
Los Angeles entered with a stronger 2025 team profile and Myles Garrett's elite pressure résumé. San Francisco still won 27–7. That does not make the pregame data useless; it shows why probability and certainty are not the same thing.
27–7Final scoreSan Francisco over Los Angeles
100,021AttendanceMCG
205Purdy pass yardsThree touchdown passes
0Garrett recorded statsAP game report
Current result context: Associated Press game report, 11 September 2026.
THE ANALYTICS JOB
Three layers to keep separate
Matchup
Protection, coverage, injuries and personnel tell us where that baseline might bend in a specific game.
Outcome
One game can land far from the centre of the probability range. The right response is to update, not pretend the range never existed.
ARCHIVED CASE STUDY
Open archived analysis →The Garrett protection problem
The pregame analysis is still useful as an example of how to translate pressure data into protection choices. It is no longer presented as a live Melbourne prediction.
Pressure rate
How often the defence creates quarterback pressure on pass plays. It captures disruption that never becomes a sack.
Chip / double team
Extra protection attention. The offence is spending another player's time or assignment to slow one rusher.
On/off split
A comparison of team performance with a player on the field versus off it. Useful context, but not proof that one player caused the entire difference.