What a win-loss record hides
Standings are the first thing you look at and the last thing you should trust in September. Three things move a record besides quality.
A record is an outcome, not a description
The standings answer one question exactly: who has won more games. Every other question people ask them — who is good, who is improving, who will still be there in January — they answer only by accident.
This matters most in the weeks when everybody is looking. A 4-0 start earns a top-ten ranking, a run of prime-time slots and a month of coverage written to fit the record. Quite often the record is describing a schedule.
Who they played
Strength of schedule is the largest single distortion in any early record, and in college sport it can be the whole of it. Four wins over teams that will finish under .500 is roughly one piece of evidence, not four.
The fix costs a minute: look up the combined record of the opponents beaten, and take out the games those opponents played against the team in question. If the number sits near .400, the résumé is thinner than the record. Ranking systems that adjust for opposition exist for exactly this reason, and they routinely disagree with the standings by several places.
How close the games were
One-score games are close to coin flips. Across every major league, teams' records in them regress hard toward .500 the following season, because the things that decide them — a bounce, a whistle, a 48-yard field goal — are not skills a team can hold onto.
So a 7-2 team that won five games by a field goal and a 5-4 team that lost four of them by a possession are not nine and six games apart in quality. They may not be apart at all. When a team's record is far better than its point differential, the most likely next event is not a collapse of character; it is arithmetic.
The same logic runs in reverse, which is why the “unlucky” team nobody is writing about in October is often the one to back in December.
Point differential, and its equivalents
Points scored minus points allowed predicts future wins better than past wins do. This is one of the most reliable findings in sports analysis and one of the least used by the people reading standings.
Every sport has its version: goal difference in soccer, run differential in baseball, net rating in basketball, goals-for percentage at even strength in hockey. Each one is a bigger sample than the win column — every possession contributes instead of every game — and bigger samples are steadier.
What a record does tell you
None of this means the standings are a lie. Wins are the object of the exercise, they determine seeding and qualification, and a team that keeps winning close games is at minimum a team that keeps being in them.
The distinction is between describing the past and forecasting the future. For the past, the record is definitive. For the future, it is one input, and usually not the best one.
A quick honesty check on any hot start
- What is the combined record of the opponents they beat?
- How many of the wins were by one score, and how many of the losses?
- What is the point differential, and what record would that normally produce?
- Has anything structural changed — a new coach, a healthy starter back, a different scheme — or only the results?
- Would this team be ranked here if the same games had been played in a different order?
Four of those five are answerable from a standings page in under two minutes, which is less time than it takes to read one column arguing the opposite.