NiceUnparticularMan wrote: Mon Sep 15, 2025 1:18 pm
rkhusky wrote: Mon Sep 15, 2025 9:43 am
That doesn’t follow. Risk premiums can be positive or negative as defined in the Fama/French factor models.Risk premiums as I would define them can’t be negative as long as investors are risk averse. What you are talking about is instead observed performance in some historic period relative to the relevant very-low-risk asset.
It’s better to come up with a different term, then to redefine widely used existing terms.
NiceUnparticularMan wrote: Mon Sep 15, 2025 1:18 pm
And not all risks are compensated risksYes, we discussed this already. You have to believe stock market risk of ANY kind is compensated before it makes sense to agree to exposure yourself to it. This again is true in single-factor as well as multi-factor models
And again, it is true that in some period, market portfolios can underperform the relevant very-low-risk asset. So in those periods, observed returns to market can be negative, the same as for any factor.
Again, so much of this conversation is basically taking the form of you pointing out something true about other factor risks, and then me pointing out the same thing is true of market risk.
And then I will respond that the market factor is substantially different then other factors and provide a plot like the one below that shows TSM and an 80/20 mix of TSM/SCV compared to short term Treasuries, indicating that the market factor is an order of magnitude more important than any other factor. And that even the market premium itself is uncertain, making all other factor premiums so uncertain that one should not put much stock in them and realize that it’s just a gamble that they will have any benefit to your future financial outcomes.
NiceUnparticularMan wrote: Mon Sep 15, 2025 1:18 pm
It’s all about return because even risk is tied to returnWell, ultimately the point of savings is to give up control of some money at Time 1, and then be able to get back control of money at a later Time 2.
It is true it would be nice if for every $1 you gave up at Time 1, you knew you would have more $$ at Time 2.
However, when you don’t know what $$ you will get at Time 2, it gets complicated choosing what you want to do.
Usually this starts with some sort of expected return assumption or calculation, which is the probability-weighted average of all the possible results at Time 2. But because the actual results may be different from the expected results, there is risk.
Then including because of reasons like the diminishing marginal utility of money, interactions with other parts of your financial life, and so on, sometimes having more $$ than the expected return at Time 2 would be less helpful than having less $$ than the expected return at Time 2 would be harmful. So you are willing to give up some of your maximum expected return to reduce this risk.
Then when you are looking at the entire possible lifecycle of an investment–which may extend to multiple generations of natural persons, charities, and so on–there may actually be substantial uncertainty about when you would want to get your final results anyway, which adds a LOT more complexity.
So “optimizing” across all this is extremely tricky. You have to balance all sorts of different risks and figure out what overall plan using all your financial resources–including things like your own human capital–makes the most sense for you.
In that context, it typically does not make sense to look at just one specific moving part of your financial plan and say, ‘Well, whatever has the highest expected return is best for me.” You instead need to understand how that one moving part fits into the overall plan, and it is possible–indeed, I would say inevitable–that the highest expected return choice will not be the best choice for that particular part of your plan.
But then you realize that there is so much uncertainty about the future that slice and dicing and fine-tuning with these different stock factors is no more likely to bear fruit than just rolling a pair of dice and that you can’t reject the null hypothesis that factor investing has no effect on your future quality of life, so you might as well just go back to adjusting your stock/bond ratio and leave it at that. And even the US/Int’l split is unlikely to have much bearing, so you might as well just use market weight. And even a TDF is likely to be good enough in the long run. That the only benefit of all this factor stuff is that it gives you something to do, which might keep you from getting bored in retirement, which I suppose might be an increase in your quality of life.
NiceUnparticularMan wrote: Mon Sep 15, 2025 1:18 pm
You are either trying to get the highest return for a given amount of risk or you are trying to get the same return for the least risk. Those are the points on the efficient frontier.OK, this may explain a lot.
In a single-factor model, the efficient frontier is actually a line, not two points. 100% the very-low-risk asset and 100% the efficient risky portfolio are points on that line, but all the points in between can also be efficient. Moreover, assuming low-cost leverage, points beyond 100% of the efficiency risky portfolio can also be efficient.
In a two-factor model, 100% the very-low-risk asset, 100% of risk factor A, and 100% of risk factor B are all efficient as well. And then these define not a line but instead a surface, and everywhere on that surface is also efficient. And this efficient surface it can go beyond those points as well with leverage.
And then what used to be a line between 100% of the very-low-risk asset and 100% of risk factor A and 0% of risk factor B is still a line across this surface, but there are many other points not on that line that are still on the efficient surface, the ones where risk factor B is not fixed at 0%.
Then in a three-factor model, the efficient surface becomes an efficient three-dimensional region in space. The old efficient line remains a line through this region, but the other efficient points are now spread out around this line in two different dimensions.
Then in a five-factor model . . . well, describing things in that many dimensions gets tricky. But the principles are the same. The efficient region is N-dimensional in nature.
And rational investors want to be on the efficient frontier and expect to have to take higher risk to get higher return.OK, armed with the insights above, you should now understand the rational multi-factor investor wants to be in the efficient N-dimensional space. Somewhere.
So, rational investors who take more risk than the market portfolio do so because they want to outperform the market portfolio.
And armed with those insights above, you should now understand why this is not right as a blanket statement.
If you assume a fixed point on the old efficient line, say 50% risky assets and 50% very low risk asset, it is true if you compare two risky portfolios, one which is 100% exposed to market and 0% to the other dimensions of risk, and one which is 100% market and has some positive exposure to one of the other dimensions of risk, then you are going to assume a higher expected return for your 50/50 portfolio. And more risk.
However . . . and I know this is sometimes a struggle for some Bogleheads–you do NOT have to assume your 50/50 very-low-risk/risky ratio is fixed.
Like, maybe in your risky portfolio you load up on 100% market plus some other dimensions of risk, but then you back down your total allocation to your risky portfolio to 40% from 50%.
Now your 40/60 factor portfolio might not actually have a higher expected return than the hypothetical 50/50 market portfolio.
In terms of risk–this 40/60 portfolio will have less risk on the market dimension, but more risk on the other dimensions.
Is this a good trade off? That depends on the entire financial context of this portfolio.
But the point is you don’t have to be trying to “beat” the market portfolio in the way I think some Bogleheads who mentally fix this ratio are thinking. You can be trying to manage your risks so you have more of some risk exposures and less of other risk exposures.
That said, if you had a 100% market stock portfolio as literally your only financial element, and were never planning to contribute or withdraw just let it grow, and then you shifted to a stock portfolio with the same market risk AND other risk exposures as well, sure, under those specific assumptions, that would be all about accepting higher risks for higher expected returns in this sort of model.
Even then, though, you could compare this to something like a leveraged 120% market portfolio instead of an unleveraged market plus other risks portfolio.
OK, so if you maximize your leverage to buy stocks, and again this is literally your only financial element, with no plans to contribute or withdraw, then factor investing becomes all about accepting higher risks for higher expected returns.
And that could still be efficient as that word is defined.
But it also doesn’t describe a plausible real world situation.
So real world factor investors may not in fact be trying to beat a max-leveraged market portfolio, they may in fact have something considerably more modest in mind, including more some form of swapping around the risk mix as opposed to simply increasing it.
And all that could be happening within an N-dimensional efficient space.
See above about the likely uselessness of getting too wrapped up in the details and fine-tuning of optimizing in N-dimensional spaces, because at that point the efficient frontier is just a notional concept and the future is so fuzzy that you might as well aggregate all the risks into one measure of risk and go back to two dimensional space.
