“He averages 25 points. The line is only 24.5. Easy over.”
It is a neat argument. It is also missing the part where you count how often he actually gets there.
An average tells you how much a player produced across a group of games. A prop asks whether he crosses one exact target tonight. Those are different questions, and one huge game can make the gap surprisingly large.
Research checked October 24, 2026, at 8:06 a.m. EDT. Every score, sample, line and price below is a made-up teaching example—not a real player's history or a live sportsbook offer.
Same average. Very different tickets.
Imagine two players who each scored 250 points over ten games. Both average 25 points per game. Now compare their results against the same hypothetical over 24.5 points.
Hypothetical player A
A big finish lifts the average.
161819202122232435 ✓52 ✓
Hypothetical player B
More nights near the target.
242424242426 ✓26 ✓26 ✓26 ✓26 ✓
Each tile is one hypothetical game. A check mark means the score cleared 24.5. These samples are illustrations, not forecasts.
Player A finishes below the target eight times, even though his average sits above it. The 35- and 52-point games pull the average up. Player B clears it five times, with the exact same average.
That does not make B a bet. It shows why “average above line” is not enough to make either player one.
Average, median and hit rate: three useful numbers
The average adds all the results and divides by the number of games. For A, that is 250 ÷ 10 = 25.
The median finds the middle of the sorted results. With ten games, average the fifth and sixth values. A's middle pair is 21 and 22, so his median is 21.5. B's middle pair is 24 and 26, giving him a median of 25.
NIST's statistics handbook explains why large extreme values can pull the mean away from the middle. The median is less sensitive to their size because it depends on the order of the results.
The hit rate counts how many games met the exact condition. A cleared over 24.5 in two of ten games: 20%. B did it in five of ten: 50%.
| Measure | Player A | Player B |
|---|---|---|
| Average points | 25 | 25 |
| Median points | 21.5 | 25 |
| Over 24.5 | 2 of 10 | 5 of 10 |
The median is useful context, too—not a magic replacement for the average. It does not tell you the full range of outcomes or tonight's chance. Looking at the game-by-game results tells you more than swapping one headline number for another.
Keep the huge game in the record
It is tempting to erase A's 52-point night because it makes the average inconvenient. Don't. If the score is correct, it happened. The player may have a real ceiling, even if he does not reach it often.
NIST's guidance on outliers distinguishes an error from a genuine unusual observation. An extreme result should prompt a closer look, not automatic deletion.
The useful question is why it happened. Was it extra playing time? A different role? An unusually efficient shooting night? You can compare results from a clearly defined role, but explain the filter and keep the unfiltered picture available. Removing only the games that hurt your argument builds a flattering sample.
Change the target, change the hit rate
Now lower A's hypothetical market to 20+ points. Seven of his ten games qualify. Raise it to 30+ points, and only two do.
The average stays 25. The historical hit rate changes from 70% to 20% because you asked a different question. The sportsbook price can change with the target, too. A lower bar does not automatically mean better value.
That is why a 20+ history cannot support an over 24.5 ticket without being counted again. For more on changing targets, see our alternate-lines guide.
The price still gets a vote
Even B's five hits in ten games do not tell us that tonight is a 50–50 event. Ten games are a small sample. Minutes, teammates, opponent and role may differ from the games you counted.
At hypothetical -110 odds, an ordinary cash bet with no pushes needs to win about 52.4% of the time to break even. A historical 50% does not establish that chance. Neither does a 25-point average.
If you want the difference between a promising number and a properly priced bet, read our plain-English EV guide. And if a small winning run is doing most of the convincing, our sample-size explainer is a useful next stop.
Ask a better question before betting
Instead of “Is his average above the line?”, try: “How do the actual results sit around this exact target, and is this still the same role?”
Check the average for overall production. Look at the median and game log for the shape of the results. Count the exact target, keep the sample size beside the percentage, then consider the price and current situation.
Sometimes that process leaves you with a bet. Sometimes it leaves you with a pass. Either is more useful than treating a big average as an automatic green light.
That is the kind of context we want to put together with Trends, coming soon to BetSlip Live. Past results are research signals, not promised winning chances. For supported slips placed elsewhere, BetSlip Live can help you import, review and follow live progress. Coverage varies by event and market; the app does not accept wagers.