In order to preserve the integrity of our data (and our Program's subsequent reliability tracking), we do not adjust our Program's predictions' output for any reason whatsoever, not even for known casualties. For example, if two apparently key players in a team are injured at the same time, we do not compensate for that in any way, even though it is obvious that the affected team's performance may suffer. If we are aware that such a situation exists, then we ourselves simply strike any such match involving that team off our betting list until such time as the situation reverts to normal, or unless we can see, through the subsequent reliability tracking, that the team's performance hadn't been affected in any way by those players not participating.
We believe that the moment we try to subjectively compensate for specific occurrences in isolation, we will immediately be guilty of compromising the pure statistical output data of the Program. Effectively, this would also negate the very purpose for which the Program was founded: that of predicting what happens "on average" across the course of a season, ignoring unexpected happenings from time to time. Sticking to that concept - but dropping a match from our betting when we are aware that something fundamental has occurred that may well impact on a team's performance - has made us money year after year, and we don't intend to change tactics now.