This is about Alan failing to account for the fact that the loss rebate is part of the equation because it is applicable to a session even if it is not applied after the session. Alan is not considering the rebate as existing until it is applied.
At first I was going to argue this wasn't about language, but maybe it is. Alan is using a very narrow (but common) definition of rebate where a rebate is something that occurs only after the fact. The truth is that the math of this properly perceives the rebate as something that is active and alive regardless of whether it is actually used at the conclusion of the session. The common use of the word "rebate" here is confusing. The math is unassailable.
In that sense, what we are calling "rebates" may more properly be considered "insurance." The "insurance" is active and allows one to take risks regardless of outcome.