APBRmetrics

[MVP]

I realized that comparing a player to a player with similar characteristics to predict an outcome (how long they'll be effective for example) seems to be an accepted method in this field.

However, is it really effective to do this? Wouldnt one have to compare a player to a whole sample of players with similar characteristics instead of comparing them to just one player?

Or is it more effective to compare them just to the player that is "closest" to them for every considered characteristic(variable)? If so is there any evidence that it's effective? really just curious!