A projection is the middle, not the answer
The projection tells you where outcomes cluster. The line asks how much of the outcome sits on one side of it. Those are different questions.
A projection of 6.2 strikeouts does not mean "about 6". It is the centre of a spread that includes 3-strikeout nights and 10-strikeout nights. The over/under probability is the share of that spread above the line.
Shape matters as much as the centre
Two players with the same projection can have very different over probabilities:
- Skew. Yardage and points distributions have long right tails. The mean is dragged above the median, so the over can be less likely than the projection suggests.
- Spread. A volatile role (committee back, part-time forward) widens the distribution. Wide distributions push probabilities toward 50% on lines near the mean and raise the chance of extreme results.
- Discreteness. Count stats move in whole numbers. "Over 6.5" and "over 6" are very different when the result is often exactly 6 or 7.
Reading model outputs this way
When a PropBetEdge tool shows a projection next to a line, the useful follow-up is: what probability does the model put on each side? Probabilities are directly comparable with a price; projections are not.
Every distribution is a model assumption. Poisson, normal and negative binomial shapes can put noticeably different probability on the same line. When a tool does not state its distribution, treat projection-to-probability conversions as approximate.