Forgive me for not putting stock into stats that are deemed only "accurate" based on a 3 yr sample. Yet we want to use a single season sample and factor it into actual value for a player? In what universe does that make sense?
Not to say that defensive stats are completely useless. Give a large enough sample you surely can tell to a certain degree how good/bad a person performed defensively. But there's way you can put an exact value on that defense and there is too much variance and personal bias/subjectivity to call defensive stats accurate. Certainly not over a single season (or 1 month, in this case).
You put stock in pitching and offensive numbers when they can have a huge luck factor tied to them in BABIP. That's why you take what actually happened on the field. Sometimes bad hitters have good years and vice versa. It doesn't mean we don't take what actually happened that year and see how valuable they were.