Rebuilt the whole sleeve around sector weights instead of yield, and it worked
Two years ago my REIT holdings were a yield screen. I owned whatever paid over 6 percent, which in practice meant a lot of older retail, a couple of net lease names with thin coverage, and two office REITs I told myself were mispriced. Blended yield 6.9 percent, total return over the period I held them was slightly negative before dividends.
So late in 2024 I tore it down and rebuilt on sector logic. Roughly: 30 percent industrial, 20 percent data centers, 15 percent senior housing, 15 percent self storage, 10 percent apartments, 10 percent left in a broad fund because I wanted something I wasn't picking. Blended yield dropped to 3.8 percent, which felt like a real loss at the time because I'd been managing to that number for years.
Total return since the rebuild is up meaningfully, and more importantly the underlying operating numbers moved the way I expected them to. Senior housing occupancy has been climbing on demographics that were visible five years out. Industrial rents kept marking up on renewal even where headline leasing slowed. Self storage was the one I nearly cut and I'm glad I didn't, because the supply pipeline there fell off a cliff and pricing power came back.
The part that nearly broke it was the six months in the middle where nothing worked. Generalist money was all in AI names, REIT valuations went nowhere, and I was sitting on a yield 3 points lower than before with no capital appreciation to show for it. My spouse asked a very reasonable question about why we'd taken a pay cut to own the same asset class. I did not have a good answer for two quarters.
What I'd keep: writing down the reason for each sector weight before buying, one paragraph, dated. When the middle six months happened I could go back and check whether the reason had changed or just the price. It hadn't changed.
What I'd do differently: I sold the office positions all at once at the bottom of a bad week. Should have staged it.