Momentum: Is a Hot Team Really More Likely to Keep Winning?
“They’re on a roll.”
“They’ve won seven straight.”
“This team has all the momentum.”
Few ideas in sports feel more intuitive than momentum.
When a team keeps winning, we naturally expect the success to continue. When a team loses repeatedly, it can seem as though everything is moving in the opposite direction.
But sports forecasting requires us to ask a harder question:
Is the team actually getting better—or are we simply observing a streak?
That’s an important distinction.
Because momentum can contain genuine information.
It can also contain a considerable amount of noise.
What Do We Mean by Momentum?
In everyday sports conversation, momentum can mean several different things.
There is in-game momentum:
A basketball team goes on a 15–2 run.
There is short-term momentum:
A baseball team wins six consecutive games.
And there is longer-term performance momentum:
A soccer club steadily improves over several weeks as a new tactical system begins working.
These aren’t necessarily the same phenomenon.
For forecasting purposes, we’re especially interested in whether recent performance provides information about future performance.
And that means we have to look beneath the winning streak.
Today’s Analytical Concept: Recency Weighting
Suppose we’re evaluating a basketball team that has played 50 games.
Should Game 1 and Game 50 receive exactly the same importance when forecasting Game 51?
Perhaps not.
The team may have changed considerably.
Players have been injured or returned.
Lineups have changed.
Young players have developed.
A trade may have occurred.
The coaching staff may have changed its rotation.
This leads to a useful forecasting technique called recency weighting.
Instead of treating every historical game equally, we give somewhat greater weight to more recent performances.
Conceptually:
Recent games = greater weight
Older games = smaller weight
But there’s a danger.
Give recent games too much weight and the model begins chasing streaks.
Give them too little weight and the model may react too slowly to genuine changes.
Finding the balance is the challenge.
A Five-Game Winning Streak
Imagine an NBA team has won five consecutive games.
At first glance:
Strong momentum.
Now let’s investigate.
During those five games:
three opponents were missing key starters,
four games were at home,
the team’s three-point percentage was unusually high,
opponents shot unusually poorly, and
three victories came by four points or fewer.
Does the winning streak still matter?
Absolutely.
But the underlying evidence suggests caution before assuming that five victories automatically predict a sixth.
Now consider another five-game winning streak.
This team:
defeated several strong opponents,
won both home and road games,
improved its defensive efficiency,
generated better shot quality,
reduced turnovers, and
won by consistently comfortable margins.
Same record:
5–0.
Very different evidence.
Winning Streaks Can Be Outcomes—or Symptoms
This distinction is extremely useful.
A winning streak can simply be an outcome.
Or it can be the symptom of an underlying improvement.
If a team begins defending better, creating higher-quality scoring opportunities and getting healthier, victories may naturally follow.
In that case, the winning streak isn’t necessarily the important information.
The underlying improvement is.
That’s what we want to identify.
Today’s Statistical Tool: Rolling Average
One simple way to examine recent performance is a rolling average.
Suppose a basketball team’s offensive efficiency over its last five games is:
112
116
118
121
123
Rather than concentrating on one game, we can calculate the average across a moving window.
When the next game occurs, the oldest observation drops out and the newest is added.
This creates a continuously updated picture of recent performance.
The same technique can be applied to:
scoring efficiency,
defensive efficiency,
expected goals,
run differential,
turnovers,
shooting percentage,
yards per play,
possession statistics, or
almost any measurable performance variable.
Rolling averages help smooth some of the noise created by individual games.
But Window Size Matters
Why use five games?
Why not three?
Or ten?
Or twenty?
There is no universally correct answer.
A three-game average responds very quickly—but can be heavily influenced by one unusual performance.
A twenty-game average is much more stable—but may respond slowly when a team genuinely changes.
This is a classic forecasting tradeoff:
Responsiveness versus stability.
Short windows react quickly.
Long windows provide more evidence.
A strong analytical system often considers multiple horizons rather than pretending one arbitrary window contains the entire truth.
Momentum Versus Regression
Now we encounter an interesting tension.
In our first SignalScoreSports blog, we discussed regression toward the mean.
Today we’re discussing momentum.
At first, those ideas might seem contradictory.
Momentum says:
Recent performance may continue.
Regression says:
Extreme recent performance may move back toward normal.
So which one is correct?
Potentially both.
The key is understanding why the recent performance changed.
Suppose a baseball hitter suddenly improves dramatically because he has experienced unusually favorable outcomes on balls in play.
Regression may be the stronger explanation.
But suppose he changed his swing mechanics, increased exit velocity and reduced strikeouts over a meaningful sample.
Now there may be evidence of genuine improvement.
The statistical result looks similar.
The underlying mechanism is different.
Look for Supporting Evidence
This is why forecasting shouldn’t rely on streak length alone.
If a team has strong recent momentum, ask whether other indicators confirm it.
For example:
Is offensive efficiency improving?
Is defensive efficiency improving?
Are key players healthy?
Has the lineup changed?
Has the schedule become easier?
Are underlying performance metrics improving?
Is the team winning close games or dominating?
Are unusually extreme shooting or scoring rates driving the results?
The more independent evidence supports the improvement, the more interesting the momentum becomes.
Losing Streaks Deserve the Same Treatment
The same logic works in reverse.
Suppose a soccer club loses four consecutive matches.
The immediate narrative:
They’re in terrible form.
But imagine the club:
faced four elite opponents,
created more expected goals than its opponents in three matches,
lost twice on late goals,
had a starting goalkeeper injured, and
now has that goalkeeper returning.
The four-game losing streak is real.
But blindly extrapolating it into the next match may be a mistake.
Now consider another club that has lost four straight while:
being consistently outplayed,
generating fewer scoring opportunities,
conceding more chances,
losing key players, and
showing declining underlying performance.
Again:
Same streak.
Different analytical story.
Psychological Momentum Is Harder to Measure
What about confidence?
Team chemistry?
Frustration?
The feeling that a team has “forgotten how to lose”?
These factors may exist.
The difficulty is measurement.
Sports forecasting works best when variables can be defined consistently and tested against historical outcomes.
“Team confidence” is much harder to quantify than:
offensive efficiency over the last 10 games.
That doesn’t mean psychology is irrelevant.
It means analysts should distinguish between:
something plausible
and
something measurable.
Those aren’t always the same thing.
Beware of Narratives After the Fact
Humans are excellent storytellers.
If a team wins eight consecutive games, we can usually construct a compelling explanation.
“They finally learned how to win.”
“The locker room believes.”
“They’re peaking at exactly the right time.”
But suppose the same team loses its ninth game.
A different story appears immediately.
“They became overconfident.”
“The streak exhausted them.”
“The pressure finally caught up.”
This is a warning sign.
If almost any outcome can be explained after it happens, the explanation may not have much forecasting power.
Good analysis tries to establish the variables before knowing the result.
Momentum Can Be Sport-Specific
Recent performance may also have different significance depending on the sport.
Baseball contains substantial game-to-game variation.
Basketball provides many scoring possessions, which can make underlying team strength emerge differently.
Football has relatively few games, making each observation influential.
Soccer’s low-scoring nature means individual events can have outsized effects on results.
Tennis involves individual competitors rather than teams and introduces surface, opponent and tournament effects.
A forecasting system should therefore avoid assuming that “momentum” means exactly the same thing everywhere.
SignalScoreSports and Momentum
Momentum is one of the components considered in the SignalScoreSports analytical framework.
But the important word is:
component.
Momentum shouldn’t overpower everything else.
Recent performance should be evaluated alongside variables such as:
injuries,
opponent quality,
rest and fatigue,
longer-term team strength, and
other matchup-specific information.
A hot streak can matter.
But understanding why the team is hot matters more.
SignalScoreSports Takeaway
The next time you hear:
“They’ve won six straight—they have tremendous momentum,”
don’t automatically disagree.
But don’t stop there either.
Ask:
Who did they play?
How did they perform?
Did the underlying statistics improve?
Did the roster change?
Are the results being driven by something sustainable?
Or are we looking at normal short-term variation?
The streak is the beginning of the investigation.
It isn’t the conclusion.
Because in sports forecasting, the most important question isn’t:
“Who’s hot?”
It’s:
“What evidence tells us that the improvement is likely to persist?”
That’s the difference between following a streak and analyzing one.
SignalScoreSports provides sports forecasting, statistical analysis, and educational information. Predictions and analytical outputs are probabilistic estimates, not guarantees of future results.
