AllocateSmartly (Ochotona)

A place to talk about speculative investing ideas for the optional Variable Portfolio

Moderator: Global Moderator

Post Reply
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

AllocateSmartly (Ochotona)

Post by ochotona » Sat Apr 08, 2023 5:28 pm

🚧 split off Moonshots thread \ DS 🚧

Make this your moonshot. I have 10% of my portfolio going to this. Look at how low the drawdown is, backtested 50 years. Look at the Sharpe / Sortino ratios.

https://allocatesmartly.com/bold-asset-allocation/

Warning - seriously not tax efficient, lots of trading per year, so only use it in an IRA / 401k / HSA, etc.

The blog exposes the rules in public. You want to code it? Go right ahead! The original SSRN paper is also published. I pay because I'm not a good coder and I have another day job!
User avatar
dualstow
Executive Member
Executive Member
Posts: 14226
Joined: Wed Oct 27, 2010 10:18 am
Location: synagogue of Satan
Contact:

Re: Moonshots

Post by dualstow » Sun Apr 09, 2023 6:45 am

ochotona wrote:
Sat Apr 08, 2023 5:28 pm
Make this your moonshot.

I looked that over twice — still waking up — but it doesn’t look like a moonshot. It looks like a strategy that gets in and out of TIPS, S&P 500, etc. Those assets that they finally list at the bottom.
A moonshot is a crazy stock that will probably lose money but might end up growing 20x because of success (or silly memesters).
RIP Marcello Gandini
D1984
Executive Member
Executive Member
Posts: 730
Joined: Tue Aug 16, 2011 7:23 pm

Re: Moonshots

Post by D1984 » Sun Apr 09, 2023 7:54 am

ochotona wrote:
Sat Apr 08, 2023 5:28 pm
Make this your moonshot. I have 10% of my portfolio going to this. Look at how low the drawdown is, backtested 50 years. Look at the Sharpe / Sortino ratios.

https://allocatesmartly.com/bold-asset-allocation/

Warning - seriously not tax efficient, lots of trading per year, so only use it in an IRA / 401k / HSA, etc.

The blog exposes the rules in public. You want to code it? Go right ahead! The original SSRN paper is also published. I pay because I'm not a good coder and I have another day job!
Three quick questions: One, is there any reason almost all of their strategies--including this one--seem to only go back to 1970 (i.e. only be backtested back to that date), or 1972, or 1973? If they really want (for instance) to show how such a strategy might do in a rising rate environment from a truly low interest rate base they should backtest it back to 1964 or 1965 (or even better, back to 1955 or 1956)? Starting in the early 70s means starting when rates were already at 5 or 6 or 7% which means that rates are much higher than they are now (and higher still than they were in, say, late 2021). I presume they are pulling their underlying pre-1980s/pre-1990s asset data from S&P, Compustat, Refinitiv, CRSP/WRDS, Ibbotson Associates/SBBI/Morningstar, Bloomberg and/or BloomBarc, Global Financial Data, and the like; I know a bit about the data availability from said providers and it should at least get them back to the 1920s if they so desired....going back to 1964 or 1955 should be a cinch (plus since apparently they only have backtested to 1970 or so it would provide valuable out-of-sample data to show that their models weren't simply overfitted).

Two, unless I am reading their description wrong, the "aggressive" version of this strategy (the one with the near 20% CAGR back to 1970 with only a 14.5% maxDD) will--provided the momentum risk indicators are in risk-on mode--go "all-in" on just one asset class, right?

Three, how are they getting Barclay's aggregate bond index data back to 1970? The actual "agg" index (i.e. the Barclays' total aggregate bond index) only goes back to 12-31-1975; the BloomBarc (Bloomberg Barclay's) backtested version of said index only goes back to 12-31-1972, and even all of the data providers I mentioned above do not (at least to the best of my knowledge) provide any version of this before that because that would entail having detailed data on intermediate-term (IT) corporate bond asset class returns (there are corporate long bond indexes back to the mid-1920s and IT govt bond indexes back to 1870 or so but AFAIK there are literally no IT corporate bond indexes with monthly or daily granularity for any time before 1972 or 1973) which unfortunately none of them do.
User avatar
Xan
Administrator
Administrator
Posts: 4392
Joined: Tue Mar 13, 2012 1:51 pm

Re: Moonshots

Post by Xan » Sun Apr 09, 2023 9:01 pm

D1984 wrote:
Sun Apr 09, 2023 7:54 am
Three quick questions: One, is there any reason almost all of their strategies--including this one--seem to only go back to 1970 (i.e. only be backtested back to that date), or 1972, or 1973?

I would be very suspicious of any backtest that went back past the early 1970s, as the nature of money was different before that time. A bond was a different thing, cash was a different thing.
D1984
Executive Member
Executive Member
Posts: 730
Joined: Tue Aug 16, 2011 7:23 pm

Re: Moonshots

Post by D1984 » Mon Apr 10, 2023 6:43 am

Xan wrote:
Sun Apr 09, 2023 9:01 pm
D1984 wrote:
Sun Apr 09, 2023 7:54 am
Three quick questions: One, is there any reason almost all of their strategies--including this one--seem to only go back to 1970 (i.e. only be backtested back to that date), or 1972, or 1973?

I would be very suspicious of any backtest that went back past the early 1970s, as the nature of money was different before that time. A bond was a different thing, cash was a different thing.
I'm not so sure of that.

The "classical" gold standard at $20.67 an ounce was effectively killed (at the latest) in 1933; from that point on one could say we were on a quasi-gold standard but there was technically nothing stopping the government (had it wanted to) at valuing gold at some other figure than the $35 an ounce it chose (and tried to maintain) from 1934 to early 1971. At the very least it was clear by the early or mid-1960s that the link between gold and the dollar was not going to be forever sacrosanct; we had to form the "gold pool" to intervene and try and prevent prices from deviating too much for the $35 peg and by (IIRC) 1966 or 1967 Switzerland and France said screw this, refused to participate, and started to redeem more and more of their dollar holdings for actual gold.

Arguably the overall "deflationary" gold standard environment died even earlier i.e. in the ashes of WWI. If you look at inflation averages from 1914 or 1915 to the early 1930s (I used https://www.measuringworth.com/calculators/inflation/ ) you will notice two things:

One, average overall inflation from 1914 to 1931 or 1932 came in at around around 2% or a bit under/over...this is actually roughly comparable to average inflation from 2005 to 2020.

Two, prices (whether at the 1920-21 recession bottom or the 1932-33 Depression bottom) never actually got anywhere close to as low as they were in 1914 or 1915. This was the first time this had ever happened in the U.S. Previously (at least back to the early 1790s when the U.S. first defined the dollar in terms of gold) the pattern had always been generally deflationary over the long term; prices would rise during wars, national crises, supply shocks, and economic bubbles/booms and then would fall afterwards to eventually end up about the same (and actually slightly lower when all was said and done). The post WWI environment was the first ever time this didn't happen; it again failed to happen after WWII (there was actually sharp--sometimes double-digit in fact--inflation in 1946 and 1947, mid-single digit inflation over much of 1948, followed by barely any deflation at all over 1949).

Also, note that even from, say, 1900 to 1913 the overall pattern was not really deflation (whether of the gentle type of maybe half a percent a year on average as productivity growth caused prices to fall in real terms or the sharper type seen during falls that followed the rises during wars and the like) but rather low single digit inflation in the 1.3% range on average for the period as a whole (in contrast with average deflation of around 0.40% a year from 1800 to 1899).

Finally, by the mid-1910s (and certainly at least by the early to mid-1920s) corporate bonds and safe-short term instruments (quasi-cash like money market debt, banker's acceptances, high-grade commercial paper, and short-term Treasury securities) were available that were similar to equivalent instruments today.

The bottom line? I can see the sense of not wanting to use data from the 1800s in any kind of backtest; once could even argue that not using data from the 1900s/1910/1920s or even 1930s was justifiable....but to treat bonds in , say, 1955 or 1962 or 1967 as not being a rough equivalent of bonds today is just IMO a stretch too far.
Last edited by D1984 on Mon Apr 10, 2023 8:00 am, edited 1 time in total.
User avatar
ochotona
Executive Member
Executive Member
Posts: 3353
Joined: Fri Jan 30, 2015 5:54 am

Re: AllocateSmartly (Ochotona)

Post by ochotona » Mon Apr 10, 2023 10:49 am

Go to allocatesmartly and send them an email... I don't have the answers to those questions.
Post Reply