How to read a projection

The median, the distribution, the 80% range, volatility, and what the Model % line is actually claiming.

6 min read · Updated 2026-09-17

Every projection on the board is the summary of thousands of simulated games, not a single guess. Reading it well means reading the spread, not only the middle.

The projection

The headline number is the centre of the simulated outcomes for that player and market. Half the simulations landed above it and half below. On a market with a small integer support, such as home runs or stolen bases, the centre is often zero even when the model likes the over, which is why the probability matters more than the point estimate.

The distribution and the 80% range

The research page shows the typical simulated range under the label 80% of simulated outcomes. That band is the fifth to ninety fifth percentile region the simulation produced, so one outcome in five is expected to land outside it. A wide band is not a fault in the model. It is the market telling you the stat is noisy.

Example: Illustrative only

A receiving yards projection of 58.4 with a typical range of 22 to 101 says the model expects a middling game and would not be surprised by either a quiet one or a big one. A posted line of 59.5 sits almost exactly in the middle of that, so the probability will be close to a coin flip whatever the projection looks like.

Volatility

Volatility describes how wide the distribution is relative to the level of the market. High variance rows carry a caution flag, because a large gap between projection and line means much less on a stat that swings wildly than on a stable one.

Model % and its basis

Model % is the probability the simulation gives the exact side the app sells at the exact posted line. It is decision conditional: a leg that pushes or voids is excluded from the denominator, so the number answers the question you are actually asking.

Under the number sits a basis line naming the model version that produced it and what that market's graded record says about probabilities like it. Two probabilities that look the same are not the same claim when one comes from a recalibrated market with thousands of graded rows and the other from a market with no record yet.

What a projection is not

It is not a forecast of what will happen, and it is not evidence that the price is wrong. Comparing it with a price is the next guide: model versus market.

Related

Slateline is an educational research and analysis product. It does not place bets or accept wagers. Projections are estimates, not guarantees. If gambling stops being fun, call 1 800 GAMBLER.