Free research lab · No account needed

Put a streak
in perspective.

Bring a player’s game log. Compare the recent window with the games before it, see how one result moves the average, and separate observed frequency from a prediction.

Your entries stay in this page. This tool does not upload or save them.

Explore a game log

Enter 2–100 nonnegative numbers separated by commas, spaces, or newlines. Use one statistic and unit throughout. Omit games the player did not play; enter 0 only for an actual zero result.

The initial numbers are a fictional teaching example, not a real player’s record. This tool does not load live sports data.

Worked example · 10 games

Your sample, explained

Average23.00
Median23.00
Recent average25.60
Prior average20.40

The last 5 games average 5.20 more than the preceding 5 games.

Oldest → newestOrange = recent window

5 of 10 games were above 22.5, 0 were equal, and 5 were below.

Range: 15 to 31.

Observed frequency describes these games. It is not the probability of the next result. A change in minutes, health, opponent, or role can make the next game unlike this sample.

See the individual results
GameResultWindowvs. threshold
Worked example

A rising average needs an explanation.

Imagine a fictional basketball player with these ten scoring results, in chronological order: 18, 22, 15, 27, 20, 24, 31, 19, 26, 28. The first five total 102 points, for an average of 20.4. The last five total 128, for an average of 25.6. The full sample averages 23.0.

The recent average is 5.2 points higher, but that difference does not tell us why. More minutes could explain it. So could a temporary shooting streak, weaker opponents, or an unusual overtime game. The next research step is to compare opportunity and conditions between the two windows—not to assume a new scoring baseline.

Stress-test the headline

Replace the 31 with 51 and run the tool again. The full average rises from 23.0 to 25.0, while the median stays at 23.0. The recent average rises to 29.6. One exceptional game now makes the trend look much stronger, although the other nine results are unchanged. This is why the tool shows the median, individual observations, and range alongside the average.

A threshold is a description, too

In the original example, five results exceed 22.5 and five fall below it. That is a 50% observed frequency in this ten-game sample. It is not a calibrated 50% forecast. Selecting a player because of an eye-catching streak, switching thresholds after viewing the results, or excluding inconvenient games can all distort the conclusion.

How the calculations work

Average
Sum of all entered results divided by their count. Every entered game has equal weight.
Median
The middle result after sorting. For an even number of games, it is the average of the two middle results.
Recent and prior windows
The recent window uses the last 5, 10, or 20 entered games. The prior window uses up to the same number of games immediately before it. The windows do not overlap. If fewer games are available, the actual counts are shown; if no prior games remain, no prior comparison is made.
Above, equal, and below
Each result is compared directly with the threshold. Equal results remain separate; they are not counted as above or below. These categories describe the numbers and do not apply sportsbook settlement rules.
Sample standard deviation
The square root of the sum of squared differences from the sample average, divided by n − 1 before taking the square root. It measures dispersion in the statistic’s units. It is not a prediction interval or a confidence score.

Use the result to ask a better question.

Did opportunity change?

Review minutes, attempts, plate appearances, routes, or ice time in the same windows. A result without an opportunity measure leaves a major part of the explanation unresolved.

Read the recent-form workflow →

Will those conditions continue?

Re-check lineup status, teammate availability, and opponent context. The calculator cannot know whether the conditions behind the sample still apply.

Review roles and availability →

This public lab is a descriptive calculator. It does not reproduce the app’s projections, recommend selections, or guarantee outcomes. Keep the raw game log and explain any exclusions when sharing your analysis.

Published November 6, 2026 by TAP Props. Found a calculation issue? Contact support with the inputs and the result you expected. Learn more about sample size and variance.