This is from Shandler...
Writers are starting to wisen up, but it pays to mention this as
a caution to unsuspecting readers out there.
If you ever read that someone is a consistent "first half" or
"second half" player, turn the page. Several statistical sources
make available 5-year averages for players, and writers
occasionally use this data to draw incorrect conclusions.
For instance, you might read that, over the past 5 years, Player
X has an ERA of 4.50 before the break, but 3.25 after. The writer
might use this as support to project an improvement in
performance for that player.
This is garbage - don't believe it.
What the writer neglects to tell you is that those 5-year
averages are AGGREGATE figures. They do NOT represent CONSISTENT
trends. In other words, the player mentioned above might have
year-to-year ERAs that look like this:
1st half 2nd half
Year 1 5.20 2.05
Year 2 3.71 3.67
Year 3 3.26 4.00
Year 4 5.15 2.82
Year 5 3.97 4.01
Aggregate 4.50 3.25
Would you consider that this data set foretells a player who
improves consistently in the second half?