Radical thought

[pb4uski]

Well, the experiment has begun. The inherited Roth IRA was transferred to me last week and I just bought some UPRO (55%) and TMF (45%).

Not the whole account yet as there are a couple bonds that were called that I won't get the call proceeds until tomorrow and Thursday. I'll probably wait a while before putting that money to work... or maybe not... who knows.

So far, I'm down $6.

Obviously a couple days is way too early to judge but it has been fun to watch. The HFEA portfolio is up 2.0% vs 0.6% for SPY vs 0.5% for 55/45 mix of SPY/TLT (unlevered equivalent of HFEA portfolio)... so the early returns are actually a little better than 3X.

The only odd thing is that SPY/TLT and UPRO/TMF are supposed to be uncorrelated... when one yings then the other is supposed to yang... and so far they have moved in tandom. I think the ying/yang thing is more during periods of financial stress rather than all the time.

Once/if it gets ahead a bit more then I may put in some stop loss orders just in case.

 

[Texas Proud]

Obviously a couple days is way too early to judge but it has been fun to watch. The HFEA portfolio is up 2.0% vs 0.6% for SPY vs 0.5% for 55/45 mix of SPY/TLT (unlevered equivalent of HFEA portfolio)... so the early returns are actually a little better than 3X.

The only odd thing is that SPY/TLT and UPRO/TMF are supposed to be uncorrelated... when one yings then the other is supposed to yang... and so far they have moved in tandom. I think the ying/yang thing is more during periods of financial stress rather than all the time.

Once/if it gets ahead a bit more then I may put in some stop loss orders just in case.

Do not chicken out!!! I am going for it at least a year and maybe more..


I did calculate that Vanguard cost me 10.5% return by NOT allowing me to invest when I wanted to... I will never get that money back

 

[pb4uski]

I thought about 60/40 and just decided to go with the actual HFEA portfolio....

Keep us up to date on your experiment!

An update.

55/45 HFEA portfolio put in play on Dec 18. Through Friday's close, the HFEA portfolio is up 6.4% vs 2.9% for an unlevered 55/45 mix of SPY/TLT so the HFEA is 219% of the unlevered, about as I expected.

XIRR of HFEA is 35% vs 15% for unlevered.

For the same period, just SPY would be up 8.7% with a 49.3% XIRR. I'll rebalance back to 55/45 at the end of March.

The portfolio can be a little squirrely though... there have been some days where the leverage was actually negative inception to date... IOW, the unlevered portfolio outperformed the levered portfolio.

Jury is still out, but an interesting experiment.

 

[Closet_Gamer]

An update.

55/45 HFEA portfolio put in play on Dec 18. Through Friday's close, the HFEA portfolio is up 6.4% vs 2.9% for an unlevered 55/45 mix of SPY/TLT so the HFEA is 219% of the unlevered, about as I expected.

XIRR of HFEA is 35% vs 15% for unlevered.

For the same period, just SPY would be up 8.7% with a 49.3% XIRR. I'll rebalance back to 55/45 at the end of March.

The portfolio can be a little squirrely though... there have been some days where the leverage was actually negative inception to date... IOW, the unlevered portfolio outperformed the levered portfolio.

Jury is still out, but an interesting experiment.

Thanks for the update. Very interesting!

 

[GaryInCO]

The only odd thing is that SPY/TLT and UPRO/TMF are supposed to be uncorrelated...

No, they will be nearly perfectly correlated. SPY and TLT usually have low correlation, and UPRO/TMF are usually low correl. (Note, "usually" ! NOT "always" !) But SPY & UPRO are very highly correlated, as are TLT & TMF -- and thus 55SPY/45TLT is highly correlated to 55UPRO/45TMF.

You asked how you could ratchet down the leverage as time goes on. Simplest way would be to just sell some of the UPRO/TMF. If the account is 100% UPRO/TMF, then if you sell half the position, you have essentially cut your leverage in half. Sell more or less if you want to unwind more or less. When you sell, move the funds into Tbills or similar, and you'll improve your results over the deleveraged account.

 

[pb4uski]

No, they will be nearly perfectly correlated. SPY and TLT usually have low correlation, and UPRO/TMF are usually low correl. (Note, "usually" ! NOT "always" !) But SPY & UPRO are very highly correlated, as are TLT & TMF -- and thus 55SPY/45TLT is highly correlated to 55UPRO/45TMF.

You asked how you could ratchet down the leverage as time goes on. Simplest way would be to just sell some of the UPRO/TMF. If the account is 100% UPRO/TMF, then if you sell half the position, you have essentially cut your leverage in half. Sell more or less if you want to unwind more or less. When you sell, move the funds into Tbills or similar, and you'll improve your results over the deleveraged account.

I wrote that wrong... I meant the SPY/UPRO are uncorrelated with TLT/TMF...IOW what should happen is when SPY/UPRO yings TLT/TMF should yang.

 

[Texas Proud]

Bought mine in Dec, but just checked and up 8% on mine... and I lost over 10% more gain due to Vanguard NOT letting me buy these so had to move the account over to Fidelity..

 

[HarveyS]

I also bought in late December. Thanks to PB for sharing his radical thought. Fun to participate in this.

 

[Texas Proud]

I might start my own thread but for now I will do it here...


So, bought Dec 19th.. was gong to buy sooner but Vanguard did not let me so moved my small account to Schwab... lost 10% while waiting for the move..


At 3/31 UPRO is up 31% and TMF is down 15% for a total return for the qtr of 13%... UPRO got to almost 70%... going to rebalance on Mon... had a small amount of cash and got two divis...


S&P up almost 12% and DOW just over 12%, so not a big delta...

 

[pb4uski]

I bought about the same time... 12/23/23. We must have used different allocations... I was 55 UPRO/45 TMF so 9.7% gain since inception vs 4.2% for unlevered 55 SPY/45 TLT. My calculations don't include dividends.

Grandpa, you're a genius! (money earmarked for my grandson).

I'm also interested in the leveraged risk parity portfolio that is 20% UPRO, 13% TYD and 67% DMBF (iMGP DBi Managed Futures Strategy ETF). While data for DMBF is only available since Jun 2019, that portfolio is much less volatile than the HFEA portfolio.

Thanks for reminding me to rebalance!

I think I'll set up a HFEA portfolio and a leveraged risk parity portfolio as a horse race for a while.

 

[Texas Proud]

Ooops... I moved it to Fidelity, not Schwab.... not a big deal on results, just wanted to correct the record...


My comparison was to the last 3 months of the indexes as listed by Fidelity.. where can I get a blended rate on the two funds? My results were a lot better when compared to that blend...


Just rebalanced back to 60/40.. surprised how big it was... about 10% of portfolio...

 

[Texas Proud]

I might start my own thread but for now I will do it here...


So, bought Dec 19th.. was gong to buy sooner but Vanguard did not let me so moved my small account to Schwab... lost 10% while waiting for the move..


At 3/31 UPRO is up 31% and TMF is down 15% for a total return for the qtr of 13%... UPRO got to almost 70%... going to rebalance on Mon... had a small amount of cash and got two divis...


S&P up almost 12% and DOW just over 12%, so not a big delta...

Wow, just looked and down below the starting point... looks like it will be a bad quarter and only a couple of weeks in..

 

[pb4uski]

Since Dec 18, 2023:

	AA	Return
HFEA Portfolio
UPRO	54.7%	9.2%
TMF	45.3%	-29.0%
	100.0%	-8.1%
Unlevered
SPY	54.7%	4.9%
TLT	45.3%	-9.4%
	100.0%	-1.6%
Leveraged risk parity portfolio
UPRO	20.0%	9.2%
DBMF	67.0%	10.1%
TYD	13.0%	-14.7%
	100.0%	6.7%

 

[OldShooter]

Wow, just looked and down below the starting point... looks like it will be a bad quarter and only a couple of weeks in..

3 years is my minimum evaluation period for an investment idea. I put a $100k into an account by itself and don't look seriously for 3 years. At that, 3 years is probably too short. 5 or even 10 years is a better evaluation period for an idea.

 

[Texas Proud]

3 years is my minimum evaluation period for an investment idea. I put a $100k into an account by itself and don't look seriously for 3 years. At that, 3 years is probably too short. 5 or even 10 years is a better evaluation period for an idea.

OH, I am not getting rid of the test... I was just surprised in how fast it dropped...


Also, I am supposed to re-balance every 3 months so cannot ignore..

 

[pb4uski]

3 years is my minimum evaluation period for an investment idea. I put a $100k into an account by itself and don't look seriously for 3 years. At that, 3 years is probably too short. 5 or even 10 years is a better evaluation period for an idea.

I agree but wonder if in that case you would view 3, 5 or 10 year rolling returns based on backtesting as relevant information to assess an investment idea?

 

[OldShooter]

I agree but wonder if in that case you would view 3, 5 or 10 year rolling returns based on backtesting as relevant information to assess an investment idea?

Absolutely not. The easiest thing in the world is to backtest until one finds a winner. Just use Will Rogers' rule: "If it don't go up, don't buy it."

That said, I am an investor, not a trader. A half century of data has shown that the market is near random over short periods of a few years. This means that there will be many schemes that work until they don't -- their luck runs out. So I no longer do this kind of stuff though I am tempted from time to time.

 

[pb4uski]

So your only way to test an investment idea is to just put $100k in play and wait 3 years and see what happens?

 

[OldShooter]

So your only way to test an investment idea is to just put $100k in play and wait 3 years and see what happens?

Pretty much. Sometimes I have made a decision in two but even three years is closer to a speculation interval than it is to serious investing. I have one $100K experiment with index funds that I started in 2015 but that is more of a benchmark than an investing idea. That indexing is a winning strategy is very old news.

 

[pb4uski]

That seems very limiting. I have no problem with defining a strategy and then using backtesting to assess it since there is no other sensible option.

I agree that could be a slippery slope if it degrades into multiple trials to adjust an investment idea so that it backtests well, but that isn't what I was referring to.

 

[GaryInCO]

Absolutely not. The easiest thing in the world is to backtest until one finds a winner. Just use Will Rogers' rule: "If it don't go up, don't buy it."

That's not backtesting, at least not proper backtesting. Properly-done backtesting has significant predictive power. Will Rogers' rule doesn't.

In response #13 I presented a 37-year backtest. I didn't build it by trying 100 things until I found one that works -- I just looked at the historical performance of a well-known strategy. You can find 60/40 backtests everywhere, but I added some value by generating synthetic 3x data to approximate the HFEA performance over a longer timeframe. That gives you a longer sample period to get a feel for how the strategy behaved in past market conditions.

At the time I hadn't heard of HFEA, but based on that test and others, I put a significant chunk of $$ into UPRO and TMF -- and UGL (2x GLD), in a 40/40/20 proportion. Unfortunately, as per my usual, I pulled the trigger at almost the worst possible moment in the last century, or at least since 1937. I opened the position in August 2022, and as a result it instantly went into drawdown. Trusting in the historical performance, I gritted my teeth and hung on. Based on my testing I chose to do annual rebalancing, though in 2023 I called an audible and skipped the rebalance. I could see TMF had more "down" in it, so I chose not to rebalance to a larger TMF position. That turned out to be a good call, as TMF is down 30% and UPRO is up 20% since then. (Exiting TMF entirely would have been a better call, but oh well...)

Result: in October '22, 2 months after opening the position, it was down nearly 40%. Thank God I didn't open it at the start of '22, or it would have dropped 65%! Then it went mostly sideways until last November. Last month it finally clawed its way back into profit! With the recent drop it's down about 3%.

The 37-year backtest may not be as good a prediction as it used to be, because things have changed -- no more declining interest rates, etc. But I think a backtest like that is the best prediction we have. It's not a data-mined curve-fit any more than standard 60/40 is. It's just a test to see how an HFEA-like portfolio would have performed in the last 40 years.