It's very unusual that an individual performs significantly (three sigma) above or below the team performance, but in the case of that happening, they could be withheld from the main pool or split a side pot.
But not without first either getting the assistance they need to catch up with the team, or teaching the team in their superior methods.
In terms of how concentrated the problem is, the overall pool covers 1000+ people, but we use the same general formula to break it down to cover groups of 15-50. That’s enough that if you have one outlier low, you’re probably okay (including just explaining why you propose granting raises over the pool because of this). But if you strategically “hire low”, your whole group suffers. (Which is another reason to not bias your offers low-leading is hard enough, so don’t start out to do it on hard mode.)
To the “if everyone did a great job” concern. If everyone did a great job, the business does better and the share-based comp becomes more valuable automatically; we would also likely make the raise pool and SBC pool larger because the company was out-performing because everyone did a great job.
My last company, only 2 people in the department could get the highest rating, other people were passed over, including myself. I warned my manager about this, but not surprisingly we lost quite a few good people because the market is good for devs and its really easy to get a solid increase in salary. Especially if you get passed over because corporates are going to corporate.
Obviously it depends on company policy, but they're not going to give raises from money they don't have.
Is that actually how it works? If everyone in a group does a good job, or if everyone does a great job, they'll get the same raises either way? Furthermore, if management is convinced that one person did a bad job, then that increases the amount available to the rest? The incentives are fascinating. How many people are in the group to which the raise pool applies?