A library for people who want to understand player props properly: how projections are built, what a posted line actually says, how probability and hit rates differ, and how to judge a model honestly. Educational research and analysis only.
Most people research player props backwards. They start from a name they like and look for evidence. This guide starts from the line and works outward: what the number claims, what a projection can add, and how to know whether your process is actually any good.
A CS2 map is really two games stitched together, and a team plays both of them. Underneath that sits a spending problem that decides how well armed anybody is when the shooting starts, which is why identical players produce wildly different nights on identical maps.
Two professional players post very different kill numbers, and the obvious reading is that one is better. In League of Legends that reading is usually wrong before it starts, because the largest single determinant of a player's statistical shape is not talent. It is which position on the map the team assigned them.
There is a category of research factor that feels like insider knowledge and behaves like decoration. Altitude, flight distance, time zone shifts, the third road game in five nights. Each has a plausible story attached. Only some of them survive contact with the size of the effect they are supposed to produce.
You check one projection source and it likes the over. You check a second and it likes the under, on the same player, the same stat, the same night. The instinct is to decide which one is broken. Usually neither is, and the disagreement is telling you something more useful than either number was.
Dead ball duty is one of the few genuinely concentrated roles in soccer: two or three players inherit a stream of chances that has nothing to do with open play. It moves several prop families at once, and it changes without announcement.
Whistles feel like hidden information, which is exactly why official tendencies get oversold. The effects are real and measurable at the team level, and much smaller by the time they reach any single player's line.
Most logs record what happened and nothing about why you decided it. That version teaches almost nothing, because outcomes in a high variance activity are mostly noise. The useful version captures the decision while you are still uncertain about it.
Before you know anything about two fighters, you know something about the division they compete in. Heavier weight classes tend to end fights earlier. Lighter ones tend to reach the judges more often. That structural tendency is the starting distribution every volume prop in the division inherits, and the specific matchup adjusts it rather than replacing it.
A tennis bracket is not decoration. Seeding decides who meets whom and when, which means the distribution of match competitiveness is engineered before the first ball is struck. Early rounds concentrate mismatches. Later rounds concentrate close matches. Volume props inherit that pattern directly, and a games line read without its round position is a line read without half its context.
Most prop research assumes the event arrives in the shape the schedule promised. Some days it does not. A doubleheader splits one date into two games, weather truncates a contest before the ninth, a suspended game finishes three days later, and a line built for a full game has to be settled against something else entirely. This is the unglamorous half of prop research, and it prevents losses no projection improvement ever could.
Form is the most persuasive story in prop research and the least reliable. The useful question is never whether a player is hot, it is whether something actually changed, because the streak is priced long before the mechanism is.
Injury news is the highest value and most abused input in prop research. A status tag is a prompt to investigate, not an answer, and the biggest researchable effect is usually not on the injured player at all.