willthrill81 wrote: ↑Mon Jan 24, 2022 9:41 pm
martincmartin wrote: ↑Mon Jan 24, 2022 9:31 pm
willthrill81 wrote: ↑Mon Jan 24, 2022 9:01 pm
For instance, let's say that you have a 30/70 AA during the first decade.
You've dinner a great job of pointing out why an equity allocation that low wouldn't work. The bond tent starts at 40% or 50% equities, rising to 100% after 10 or 15 years. So that's an average AA around 70/30 or 67/33.
So you think that 30/70 is too low for the strategy to work at all but 40/60 is enough for it to work fine?
You mentioned 30/70 as the AA for the first decade, I thought you meant that was the average, not the starting point. Sorry I misunderstood.
I give up. I spent a lot of time modeling the above in Excel already, and it seems clear that math won't change your mind.
Please don't give up. I wrote my comment as I was falling asleep last night, in an effort to keep the conversation going. In retrospect, I should have waited until I could give it the mental effort it deserves.
Can you share the Excel spreadsheet? For example, if you save it in Dropbox, you can get a "share" link which you can post here using the URL tag. Or you can import your Excel spreadsheet to google docs and share that link. I'd like to understand your modeling better. One of my fears with bond tent, and any purely historical analysis, is that there's some hidden assumption that hasn't happened in the data, but could. Perhaps your analysis is pointing out just such an assumption.
Now that I've had a good nights sleep AND my morning caffeine, here's the analysis I'd like to do:
I'd like to compare your sequence of returns, with both fixed AA and bond tent, with the returns for the two decades swapped. As you've said, with the returns the way you've modeled them, fixed AA produces a "good" result, and bond tent a "bad" result. But if we swap them, so that the bad equities are in the first decade, would a fixed AA produce a result worse than the "bad" result, and bond tent better? The bond tent only claims to raise the worst case result over all (historical) sequence of returns. It certainly will make things worse under some scenarios.
Sorry again for trying to respond when I couldn't give your thoughts the attention they deserve.