Correlation, Causation, and Sports Narratives

"Teams that rush for 150 yards win 80% of the time." True, and almost entirely backwards.

Teams that are winning in the fourth quarter run the ball to burn clock. The rushing yards are a consequence of the lead, not a cause of it. Reverse the arrow and the entire strategic recommendation built on it collapses.

This shape shows up constantly: time of possession, turnover margin in a blowout, "clutch" free-throw attempts. In each case the winning came first.

• Reverse causation. The outcome caused the stat, not the other way around.
• A common cause. A good team both runs well and wins; neither causes the other. The talent causes both.
• Selection. If you only look at games that were close, or only at players who stayed healthy, the relationship you find may be an artifact of who made it into the sample.

The third is the sneakiest, because the data looks complete. It isn't — it's the survivors.

The useful discipline is to state the causal claim out loud and ask what else it predicts.

If running the ball caused winning, then teams that ran more *while trailing* should also win more. They don't. One sentence, and the story is tested rather than repeated.

Do this with any narrative you meet: name the mechanism, then find a place it should also show up. If it only shows up in the original claim, it probably isn't a mechanism.