Uncorrelated wrote: ↑Tue Feb 04, 2020 5:23 pm
Hydromod wrote: ↑Mon Feb 03, 2020 7:11 am
In my case, this is not too bad of an approximation. I can only do any leverage in a (currently) small Roth account that is isolated from entry of additional funds for rebalancing, and it is largely intended as a legacy for my kids. It makes sense to consider it as an independent entity because it won’t see any transfers. It will be allowed to grow for at least one or two decades following the strategy.
That is still a bad idea. When the time comes to distribute your inheritance, it doesn't matter from which account it comes. Money is fungible, the only thing you should care about is your total portfolio return and risk. Bucketing up your portfolio and optimizing each part individually just results in inefficiencies.
So many bogleheads are wasting their time with market timing without having the basics down. Use lifecycle investing to determine an approximate dynamic portfolio allocation for you and your kids. Optimize for max utility instead of max CAGR. Solve Metron's equation for multiple asset classes to determine the optimal portfolio allocation. Solve Merton's equation for after-tax returns and/or multiple accounts. Research factor investing. All of these things are simple compared to market timing and collectively have a much greater and much more certain impact on your performance. You'll probably want to apply all the things I mentioned above in your market timing algorithm anyway.
I'm not arguing the utility of what you are suggesting for an unconstrained optimization scenario, and I appreciate what you are proposing. But in real life, there can be constraints that modify the problem to be solved, and constraints tend to lead to a less efficient solution.
I'm a little puzzled how market timing crept into your wonderful rant though, unless you are referring to adaptive allocation.
In my specific constrained case, I have really do have two buckets to work with, a tiny Roth bucket and a large 403b bucket, and I am constrained by (i) the inability to transfer funds between them, (ii) the inability to transfer funds out of the 403b until retirement, and (iii) the inability to add significantly to the tiny one.
The main portfolio is basically a flavor of the Boglehead philosophy, given the constraints of offerings in the 403b and the investment wishes of DW. Given the moderately short investment time frame before retirement, we consider the main portfolio good enough
(as longinvest would say) to handle our retirement needs. I would like to add a cushion, though, for increased travel and inheritance.
The tiny bucket is currently too small to move the needle, but it offers the scarce resource of leverage. Until the tiny bucket grows significantly, I can't see how it much matters whether I optimize separately or together, because only the main portfolio moves the needle. My suspicion is that many of the folks trying the adventure are also constrained in the ability to take advantage of leverage (although clearly there are those that are not...).
Because I can't transfer funds for rebalancing, for the time being it seems preferable to take maximal advantage of the leverage available with the tiny bucket by treating it as an independent 100% leveraged portfolio until it gets big enough to move the needle.
I'd say it starts moving the needle once it rises to 10 to 15 percent of the total portfolio, which may not be until retirement or sometime after. That's when it will make increasing sense to treat the portfolio as a whole.
I'm thinking of this two-bucket approach as basically in line with Taleb's idea of being conservative with 90% of the portfolio and risky with 10%. Although I think that I am underdoing the philosophy a bit.