There certainly are. Despite huge effort around the behavioural model we have to acknowledge we’re dealing with humans and volatile humans at that. By actively using the models we have been able to use many more results to calibrate which helps get the averages right but, we acknowledge that they could still be wrong.
- Not enough players playing their first matches at the start of the pool so that, even though the pool appears a little high or low, we can’t set its starting level. We call these pools 'derivative’ and do our best to adjust them dynamically using the level pump.
- Not enough players playing multi-pool. In some cases, a pool of players play pretty much in isolation so they are impossible to calibrate automatically. Or there’s just one player who pops in for a bit of a knock and makes everyone look good.
- Duff or unexpected results, if not averaged out by large numbers can have an adverse effect. We only adjust if we feel there are enough but, even so, there are cases where too many results are just not representative.