When do projections for rate stats start to make big jumps during the year?

[CottonWisconsin]

[Michael@HQ]

It's a gradual process, not a semi-radical change. From this link:

Once the season starts, the process changes. The performance baselines begin to integrate each player's historical trends and his stats-to-date. These two elements are weighted by varying levels during the course of the season. For instance, early in the season, history might account for 90% of the baseline and current stats only 10%. As each player accumulates AB or IP, that weight shifts away from history. The rate at which this shift occurs varies by player, based on a variety of factors such as a player's age, injury history and his base performance indicators.

[Markus_ls]

In my experience HQ is pretty conservative on changing projections. This is appropriate because they are trying to limit themselves to what DATA suggests and therefore small data (like the most recent partial year) is downplayed. But it's often valuable to speculate (at your own risk!) on larger shifts that they won't catch up to for awhile.

The HQ data is my sanity check. The speculation is my "high stakes" gambling urge that sometimes pays off if you do it carefully.

Example: I saw the Sox go to Okajima a couple of times in last weekends BOS vs NYY series. To me that was a significant show of confidence for a somewhat fragile bullpen and based on Saito and Otsuka's 2006 was enough for me to go after Okajima (in Holds league). I can make that giant leap of faith (and potentially be wrong). It might be irresponsible for HQ to make that kind of leap on such an incidental occurance like a couple of games. Even for things like a full 6 weeks of stats it might not be prudent when 5 years of data say something different (in other cases).