I'm primarily a boglehead investor, but I ran into an ntsx discussion and it took me down a rabbit hole of financial theorycrafting. I haven't seen any real discussion or research around this which makes me think I'm missing something obvious. This would be inside an ira.
Proposed strategy: swap from a pure VT allocation to ntsx/ntsi approximate global mix (or use VT with TYA or similar) only in the following macroeconomic conditions:
- treasury yield curve inverts (with a delay to skip blips)
- 2% fed interest rate or higher
- 3.5% CPI or less
The idea is to maintain high equity exposure, while adding the leveraged bonds to absorb part of the initial crash. From what I can tell this has happened roughly 4 times historically and would have paid off in 3, with only a small insurance premium for the fourth.
Exit condition is interest rate or cpi going outside the initial bounds. This caps gains, but I'm fine with that - it's intended to be a conservative hedge.
The obvious downside is if the fed consistently manages a soft landing like 2022-2024, in which case I paid for crash insurance that I never used. It seems to me that this is still a win though, since I'm mostly concerned about mitigation of SORR, not maximizing the portfolio entirely.
Thoughts?
Edit - general responses:
Low n: can't argue against this, but I feel that the inverted treasury curve canary is also low n and that's fairly accepted. Unfortunately I don't know that there's a great way to test this, although I'd want to see if I can find a sophisticated monte carlo option (or code one) that handles full regime logic without just effectively replication of the current small sample.
Overfitting: I don't believe this is tightly fit. Historically it works anywhere in the 1% - 4% interest range and roughly from 2.5% - 4% interest. You can also delay how long you wait after the inversion significantly and it still functions. 2%/3.5%/1 month are the values where I like the risk vs return. I also didn't optimize those numbers, you get better results if you ride the crash down further. Again however, I do agree this is not convincing due to the small n.
Treasury yield curve inaccurate: I'm more concerned about this now since it seems likely that 2022-2024 will permanently distort investor expectations and reactions, so I'm less convinced it makes sense to rely on the older historical data.