I actually emailed him after this with the following questions, he said he might get to them in a subsequent post. I agree that the bond-heavy leaning up front is a big part of what's making it wonky:
Hi there, I was discussing your latest post on retirement withdrawals with a couple people, and it seemed strange that a rebalancing method did so much more poorly than the equal withdrawals. Then we realized that since stocks have a higher return over the long run, rebalancing would almost always end up in selling more stocks, thus lowering the return.
Average expected portfolio value is one thing, but the success rates are probably the most important thing for early retirees. I thought that your results leaned towards CAPE mainly because you started with a 50/50 split which is so bond-heavy that there's no way you could sell enough bonds to make your portfolio risky enough to fail. The CAPE strategy then works because it sells more bonds than any of the other strategies, and the rest of the strategies follow in order from there. Have you run a scenario where you sell off only bonds? How does that hold up compared to the other scenarios? If you started at an 80/20 or 90/10 stock/bond split, would that make the CAPE/equal withdrawal success rate lower than rebalancing?
Lastly, what about early retirees looking at a 50-year time horizon? What strategy holds up best over a really long term retirement? This is also why I'm interested in looking at a more equity-tilted portfolio than what you have modeled. I imagine the success rates will take a nosedive when looking at longer term models.