Oct 16, 2017

What would happen if I allocated 100% to my alt-risk strategy (in terms of ruin risk)

Here was an interesting little exercise with results I guess I wasn't expecting.  None of this is very real so I don't know if I can attach much meaning to it.  The question I asked myself was what would happen to my ruin estimate if I allocated 100% to what little I know about my private Alt strategy's too-short return profile.

First, without getting too far into the nitty gritty, my alt strategy is something like 30/50/20 credit risk, fixed income momentum, and volatility risk via short options -- though that internal structure changes and drifts a little more than I might want.  The goal was to get a moderate return and low vol profile that is at least as efficient or more so than a plausible multi-asset allocation when looking backwards at history. The purpose was to nudge my overall efficiency up and left a little bit. This I have mostly achieved since the beginning of 2012 shortly after I started this thing. I'd love a little higher return but it is what it is. The idea of allocating 100% to my Alt came to me just as a fake hypothetical in order to explore some things I've been working on related to ruin estimation processes.

For comparison I'll use the Vanguard 60/40 mutual fund since it might represent a reasonable baseline allocation like my own might be in practice.  This is a convenient fiction for posts like this so I'll go with it.


When contemplating the idea of switching my entire allocation to my Alt strategy I have to entertain the following dissonant thoughts:

  • Would I actually do that? Not a chance. Too un-proven. Too much work (maybe) and I am relatively tax-inefficient.
  • Do I believe my own attempts at annualizing a monthly return series? No, not really. A linear annualization creates extreme results at the horizon that are just not plausible to me. A rolling 12 month approach is neat but unrealistic and counts middle months too much...though in the end for this post that is what I used.  In addition, I have not taken this thing through a full cycle.  I have no idea what a 2008 or 1987 type environment would do to me.  That and my momentum portion is more one-sided than not since I do not have a lot of short positions over time. 
  • Are my expectations about my return profile and my ability to extract risk premia and anomalies stable? No, not really.  
But this is just a retirement game so we'll keep playing.  

Here are the first several moments of the annualized data: 


Strategy      r        sd      skew 
VSMGX 0.0808 0.1138 -0.8669
Alt 0.0722 0.0427 -0.2918

And this is what the density looks like after we have ding-ed the returns with expectations for inflation, fees, and taxes by something like -.04. That's a rough way of doing it but again, I'll go with it for now just to make the point.




Now what I'll do is adapt the ruin estimation process to use, separately, these two kinda faked-up return distributions and see what happens.  The other assumptions in the ruin estimate are age (60), spend (.04), max age(120), gompertz life mode and dispersion (90/8.5), years run (200), mini-sim iterations (1000).  When I do that for each return distrbution I get this where the portfolio longevity for the Alt strategy is in red and for Vanguard it is in green:


Strategy     r    sd    skew    ruin est
VSMGX 0.0408 0.1138 -0.8669 0.16676
Alt 0.0322 0.0427 -0.2918 0.05093

And this is where it got interesting with results I wasn't expecting.  I knew that even with the lower return the low vol would work in my favor in ruin estimation and in fact it does drop it down a bit.  On the other hand, if I were to live for another 60 years or beyond, my strategy would kind of suck when we look at it formally (and assuming I programmed the math right).  I hadn't thought about what might happen over extreme timeframes.  But those extremes are more or less irrelevant (ex some medical breakthrough), and this is another reason why, to me, retirement finance that looks at intermediate and shortish portfolio timeframes, rather than infinite or one-period versions, can be important.  Am I going to allocate more to myself? I suppose I could for a while but I really don't want to be actively trading when I am 85 and I'd also have to trust that I could pull it off for the next 25 years. So no, probably not 100%.







No comments:

Post a Comment