Treating a REIT valuation gap as a timing signal is a common and costly mistake
This is a loss worth studying on process rather than on money, and the process error is the interesting part. Consider an investor who builds a deployment plan around the gap between public REIT valuations and where comparable private real estate is being marked. The gap is wide by any historical measure, the argument that it closes eventually is sound, and there is credible work calling this stretch the longest such divergence since the early 2000s. The mistake is reading that and treating a wide gap as a signal about when. Take a case where 310k gets deployed into a REIT sleeve in two large tranches about six weeks apart, weighted toward industrial and data centers, on the reasoning that the discount is extreme and therefore near its end. The sleeve then falls about 13 percent over the following five months and spends most of a year below the average cost basis. It eventually recovers and ends modestly ahead, so the capital is fine. The plan is still wrong. The specific error is that a valuation gap tells you about expected return over some long horizon and says nothing about the path. Two mechanisms can close it and they work on different clocks. Public prices can rise, which happens in weeks when sentiment turns. Private marks can fall, which happens over quarters because appraisals lag and nobody wants to write down a fund. When the second mechanism is doing more of the work, the gap narrows without the position going up at all. Modeling convergence as a single event misses this entirely. Second error, smaller but it costs real money. Sizing tranches off conviction in the thesis rather than off the ability to keep buying. Two tranches six weeks apart is not a deployment schedule, it's one decision split in half. By month four there is nothing left to add with, and the best prices of the whole period are sitting right there. The better approach: deploy on a calendar over 12 to 18 months regardless of what the gap does, hold back at least a quarter of the allocation for a drawdown, and stop using the size of a mispricing as information about its duration. The thesis on convergence can still hold. The timing inference from it does not follow.
