That includes shortstop. That’s why 1B starts at -12.5 runs per 162. Shortstop gets +7.5 default. That’s a 20 run differential for league average players between positions, and then it’s further adjusted against their peers by RAA. So that’s about 2 WAR of defensive value difference baked into those positions. Not only was 2007 Pujols not worth -12.5 runs, he was worth +13.1 runs according to Fangraphs! That means the defense he provided at 1B in 2007 was worth more than the defense provided by league average 2B, 3B, and SS combined. It’s just that 2007 was the only once-in-a-billion year because he couldn’t replicate it again for some reason.
The point I’m making is that the data is flawed. A normal baseline positional adjustment for a qualified 1B is always gonna be in the -10 to -12 range depending on how many innings they took at 1B. You can see it below in the last column (remember 2001-2003 don’t since count he played 3B/OF). The “fielding” column is the adjustment on top of positional that adjusts him above (or below) average fielding for the position.
Posts 1.7 RAA in 2005 then an otherworldly 24.5 RAA two years later and then quickly reverts back. It’s obviously not based on ability or any true measure of performance. A deeper dive into these numbers shows that most of the value is coming from range, so then what is the conclusion exactly? Albert Pujols had all time elite unicorn range at 1B in 2007 but not in 2005, and not in 2009 or any year thereafter? It’s logically impossible and indicates some kind of flaw in the data or coding (things you’re supposed to check before publishing data and claiming that they mean something).