Halving a full game projection feels like arithmetic. It is actually a modeling assumption, and it is usually wrong. Minutes are not spread evenly, foul trouble and blowouts arrive late, and game script has barely started to diverge in the opening quarter.
The best reliever in baseball produces nothing on a night the manager does not call him. Skill is the easy part of a relief projection. The hard part is a decision made by someone else in the seventh inning.
Points, rebounds, and strikeouts mean the same thing everywhere. Fantasy score does not. It means whatever the platform's scoring chart says it means, which makes it the one market where a number cannot travel between apps.
Every research method you have leans on history. A debut removes it. What remains is thinner than most people admit and more useful than most people expect, provided you know which of the remaining inputs actually carries weight.
Half point lines are tidy. You clear the number or you do not. Move the same offer to a whole number and a third outcome appears, one that most research quietly deletes from its arithmetic and then wonders why its probabilities run hot.
Before the matchup, the weather, the revenge angle, or the hot streak, there is a duller question almost nobody asks first: how often does a player in this role clear this line at all? That number is the anchor, and research without it drifts.
Every projection you have ever read was a summary. Behind the single number sits a whole range of outcomes with a shape, and that shape decides whether a line clears. Learn to ask for the shape and the average stops being the interesting part.
Almost everyone who does this has a limit. Far fewer have a limit that survives the moment it is tested. The difference is not discipline, it is design: when the number was decided, how specific it is, and how much friction stands between you and changing it.
A slate can offer thousands of player props. The number that are genuinely researchable on any given day is small, and it shrinks further as the session goes on. This is the case for evaluating fewer offers, earlier, and for treating a skip as a finished piece of work rather than a wasted one.
Weather is the most confidently discussed and least carefully measured input in prop research. Some of it is real and directional. Most of the arguments built on it are far larger than the effect.
Statistics describe the job a player used to have. When the job changes, every trailing number quietly becomes a description of someone else's situation.
Every other baseball prop asks how much a hitter will do. This one asks whether a single unlikely thing happens at all, and that difference changes the entire research process.