HEDGEFUNDIE's excellent adventure Part II: The next journey

[DarkMatter731]

Seems like “prohibitive” is the operative word there.

yeah, if any words ever mean anything, "prohibitive" must mean you stopped.

Be interesting to see who wants to argue with that, doesn't it? I'm not interested in really talking to a person like that.

I think the point was that no borrowing cost should be prohibitive. The equity risk premium and bond term premium are independent of borrowing costs. Just as an example, Borrowing cost in early 80s was over 15% but would be great time to start this strategy.

To be clear, by prohibitive, I meant that it seems to massively eat into the returns my model produces. I'm not running HFEA but changing interest rates from 2.5% to say, 7% in the 1990s massively changes the returns.

On a side note, does anyone know how to simulate the cost of leverage as the effective federal fund rate + <figure> over time in portfolio visualizer? By that, I mean the cost of leverage changing over time?

I could do the calculations manually in a spreadsheet then upload them but it seems like there must be an easier way to model the cost of leverage over time.

[LadyGeek]

New member SpaniardBoglehead has a question which I've moved into a new thread. See: HEDGEFUNDIE's excellent adventure [Spain]

[Hydromod]

On a side note, does anyone know how to simulate the cost of leverage as the effective federal fund rate + <figure> over time in portfolio visualizer? By that, I mean the cost of leverage changing over time?

I could do the calculations manually in a spreadsheet then upload them but it seems like there must be an easier way to model the cost of leverage over time.

I've never been able to figure out this issue. I don't know if there is something in the paid side though.

That shouldn't be too hard for them to implement. Perhaps if the issue gets raised they might add the feature.

[skierincolorado]

Seems like “prohibitive” is the operative word there.

yeah, if any words ever mean anything, "prohibitive" must mean you stopped.

Be interesting to see who wants to argue with that, doesn't it? I'm not interested in really talking to a person like that.

I think the point was that no borrowing cost should be prohibitive. The equity risk premium and bond term premium are independent of borrowing costs. Just as an example, Borrowing cost in early 80s was over 15% but would be great time to start this strategy.

To be clear, by prohibitive, I meant that it seems to massively eat into the returns my model produces. I'm not running HFEA but changing interest rates from 2.5% to say, 7% in the 1990s massively changes the returns.

On a side note, does anyone know how to simulate the cost of leverage as the effective federal fund rate + <figure> over time in portfolio visualizer? By that, I mean the cost of leverage changing over time?

I could do the calculations manually in a spreadsheet then upload them but it seems like there must be an easier way to model the cost of leverage over time.

I do a mix of cashx and a stt fund to simulate costs of leverage above the ff rate.

I think what comeinvest is getting at is that your model should probably not be affected by the cost of leverage so much. The equity risk and bond term premium are largely independent from the federal funds rate. But your model is making the highly dependent on it.

[DarkMatter731]

Seems like “prohibitive” is the operative word there.

yeah, if any words ever mean anything, "prohibitive" must mean you stopped.

Be interesting to see who wants to argue with that, doesn't it? I'm not interested in really talking to a person like that.

I think the point was that no borrowing cost should be prohibitive. The equity risk premium and bond term premium are independent of borrowing costs. Just as an example, Borrowing cost in early 80s was over 15% but would be great time to start this strategy.

To be clear, by prohibitive, I meant that it seems to massively eat into the returns my model produces. I'm not running HFEA but changing interest rates from 2.5% to say, 7% in the 1990s massively changes the returns.

On a side note, does anyone know how to simulate the cost of leverage as the effective federal fund rate + <figure> over time in portfolio visualizer? By that, I mean the cost of leverage changing over time?

I could do the calculations manually in a spreadsheet then upload them but it seems like there must be an easier way to model the cost of leverage over time.

I do a mix of cashx and a stt fund to simulate costs of leverage above the ff rate.

I think what comeinvest is getting at is that your model should probably not be affected by the cost of leverage so much. The equity risk and bond term premium are largely independent from the federal funds rate. But your model is making the highly dependent on it.

Isn't this just theoretical?

Investing in SPY at a higher interest rate would massively eat into your returns if you backtest it.

It's all well and good saying that the equity risk premium is independent from the federal funds rate in theory.

[skierincolorado]

yeah, if any words ever mean anything, "prohibitive" must mean you stopped.

Be interesting to see who wants to argue with that, doesn't it? I'm not interested in really talking to a person like that.

I think the point was that no borrowing cost should be prohibitive. The equity risk premium and bond term premium are independent of borrowing costs. Just as an example, Borrowing cost in early 80s was over 15% but would be great time to start this strategy.

To be clear, by prohibitive, I meant that it seems to massively eat into the returns my model produces. I'm not running HFEA but changing interest rates from 2.5% to say, 7% in the 1990s massively changes the returns.

On a side note, does anyone know how to simulate the cost of leverage as the effective federal fund rate + <figure> over time in portfolio visualizer? By that, I mean the cost of leverage changing over time?

I could do the calculations manually in a spreadsheet then upload them but it seems like there must be an easier way to model the cost of leverage over time.

I do a mix of cashx and a stt fund to simulate costs of leverage above the ff rate.

I think what comeinvest is getting at is that your model should probably not be affected by the cost of leverage so much. The equity risk and bond term premium are largely independent from the federal funds rate. But your model is making the highly dependent on it.

Isn't this just theoretical?

Investing in SPY at a higher interest rate would massively eat into your returns if you backtest it.

It's all well and good saying that the equity risk premium is independent from the federal funds rate in theory.

You can't simply add in higher borrowing costs. If interest rates were higher in any given historical year, almost assuredly stock prices would be lower and you would be buying the same earnings flow and expected growth for a cheaper price with higher future returns because you bought the same thing for a lower price.

It's not entirely theoretical... future returns have correlated reasonably with interest rates in actuality. There is of course variance and the correlation is not perfect.

[comeinvest]

I think the point was that no borrowing cost should be prohibitive. The equity risk premium and bond term premium are independent of borrowing costs. Just as an example, Borrowing cost in early 80s was over 15% but would be great time to start this strategy.

To be clear, by prohibitive, I meant that it seems to massively eat into the returns my model produces. I'm not running HFEA but changing interest rates from 2.5% to say, 7% in the 1990s massively changes the returns.

On a side note, does anyone know how to simulate the cost of leverage as the effective federal fund rate + <figure> over time in portfolio visualizer? By that, I mean the cost of leverage changing over time?

I could do the calculations manually in a spreadsheet then upload them but it seems like there must be an easier way to model the cost of leverage over time.

I do a mix of cashx and a stt fund to simulate costs of leverage above the ff rate.

I think what comeinvest is getting at is that your model should probably not be affected by the cost of leverage so much. The equity risk and bond term premium are largely independent from the federal funds rate. But your model is making the highly dependent on it.

Isn't this just theoretical?

Investing in SPY at a higher interest rate would massively eat into your returns if you backtest it.

It's all well and good saying that the equity risk premium is independent from the federal funds rate in theory.

You can't simply add in higher borrowing costs. If interest rates were higher in any given historical year, almost assuredly stock prices would be lower and you would be buying the same earnings flow and expected growth for a cheaper price with higher future returns because you bought the same thing for a lower price.

It's not entirely theoretical... future returns have correlated reasonably with interest rates in actuality. There is of course variance and the correlation is not perfect.

Correct. Empirical evidence as well as micro-economical theory make it plausible that investors are willing to pay a certain rate for risk-free delay of consumption, and another rate ("equity risk premium") for the risk of either the possibility of a terminal loss of money, or the risk or inconvenience (uncertainty) of temporary drawdowns. It seems plausible that the ERP would be stacked on top of the risk-free rate. I think that's how people smarter than I typically model the markets. Of course I'm sure this model is not perfect, but a starting point. If true, I'd rather invest, especially invest with leverage, when interest rates are high, now low. Because if we believe in mean reversion, we will get a handsome return boost when interest rates fall. Of course we cannot time the market. Otherwise we wouldn't have started (m)HFEA in 2021. Also, mean reversion considerations might be already priced by the market into long-term rates and equities valuations. The best is probably to avoid trying to micro-manage or reverse-engineer the market, but instead just stay your course. One exception though, I would probably sell my treasuries if rates were to go down to close to zero again in the future. But even that is arguable, and valid counterarguments have been made earlier in this thread. Proponents of bailing out of bonds when rates are very low, argue that short-term and mid-term rates might be artificially depressed by the government, and not reflect investors' required return premium.

[comeinvest]

I do a mix of cashx and a stt fund to simulate costs of leverage above the ff rate.

Can you please remind me how exactly you get the cost of leverage from CASHX and a STT fund in PV? You want to simulate T-bill rate + 0.35% for the cost of leverage of treasury futures, and T-bill rate + 0.5% for /ES futures, right?

[cos]

I do a mix of cashx and a stt fund to simulate costs of leverage above the ff rate.

Can you please remind me how exactly you get the cost of leverage from CASHX and a STT fund in PV? You want to simulate T-bill rate + 0.35% for the cost of leverage of treasury futures, and T-bill rate + 0.5% for /ES futures, right?

I'd just use a STT fund and call it a day. Sure, it's not quite correlated with the T-bill rate, and it's even less correlated with the T-bill rate plus a fixed spread, but it provides a nice, fairly conservative estimate of debt financing costs. Keep in mind that backtests aren't significantly helpful, anyway, at least not in the statistical sense. So long as your strategy is theoretically sound, you should be good to go.

To answer your question more directly, the average cost of leverage over time implied by 50/50 CASHX/VFISX is ~5% annually, but again, it won't correlate all that well with the T-bill rate plus a spread, and it might create some issues resulting from term exposure.

[willthrill81]

The negative correlation between stocks and LTT that so many were counting on completely failed this year, and that resulted in a -55% nominal drawdown (worse after inflation) in only six months. Imagine how bad it would have been if unleveraged stocks were down -50%.

The strategy might work going forward, but it takes guts of steel to have a meaningful amount of money devoted to it.

[Hydromod]

The negative correlation between stocks and LTT that so many were counting on completely failed this year, and that resulted in a -55% nominal drawdown (worse after inflation) in only six months. Imagine how bad it would have been if unleveraged stocks were down -50%.

The strategy might work going forward, but it takes guts of steel to have a meaningful amount of money devoted to it.

Actually, daily correlations were between -46 and +22% this year on a moving basis. It's the monthly correlations that have steadily climbed since the start of COVID. We haven't seen monthly correlations so high since the beginning of 1998.

Note that monthly correlations went positive near the end of 1988 and stayed positive for almost 10 years. In that same period, daily correlations were also generally positive. A UPRO/TMF portfolio would have done quite well in that period, likely at least as well as the ten-year period ending at the end of 2019. In and of itself, positive correlations aren't necessarily a death knell, they mainly tend to increase portfolio volatility. It's when all assets are trending down at the same time that there are problems.

Really, the flight to safety effect is the thing folks were relying on.

This is why I'm looking into momentum methods. Allowing something like DBC or UUP to enter the mix when both stocks and treasuries are freefalling may have helped a lot this year. But these are better as tactical allocations, because of their drag as long term holds, and swapping out could be pretty costly in taxable.

[willthrill81]

Really, the flight to safety effect is the thing folks were relying on.

Exactly, and that hasn't happened at all during the current downturn in stocks.

This is why I'm looking into momentum methods.

I've long thought that a target volatility approach would be better than a static AA with these leveraged strategies. Otherwise, the whole strategy is apt to blow up if/when stocks and bonds do a swan dive simultaneously. It's hard to recover from a 70% or bigger drawdown.

[Hydromod]

Really, the flight to safety effect is the thing folks were relying on.

Exactly, and that hasn't happened at all during the current downturn in stocks.

This is why I'm looking into momentum methods.

I've long thought that a target volatility approach would be better than a static AA with these leveraged strategies. Otherwise, the whole strategy is apt to blow up if/when stocks and bonds do a swan dive simultaneously. It's hard to recover from a 70% or bigger drawdown.

I agree. I like my risk-budget minimum variance approach a little better though.

I'm working out momentum ideas to add some additional protection during events like 2022. The idea is to keep the strategy for allocating assets given an asset set, but develop the asset set by dropping the worse-performing assets. Things like DBC and UUP only float up as options in years like 2022. Kind of like going to cash, but with a little return. Not great in taxable, though.

[yolointopants]

The combination of falling bond prices and falling equities made 2022 unique, and there's always the "fighting the last battle" risk in trying to adjust to deal with the past.

I don't know how to backtest it, but with the available data and using similar interest rate hedging funds (not hedge funds but actual etf funds that hedge) it seems like a non trivial piece of the 2022 market meltdown could have been avoided using something like PFIX.

I use mostly PSLDX in my tax advantaged account, but this would apply to the usual UPRO/TMF combo and instead of mixing PFIX or an interest rate hedge at 90/10 or even 80/20 to reduce volatilty at the expense of overall returns, but improving risk adjusted returns.

Is anyone aware of a way to backtest this strategy? The other option, and this has been mentioned but with TYD, is using TYA in lieu of TMF. At some point you need to poo or get off the pot because no strategy is going to be perfect and tinkering usually lands you with worse results not better.

The backtest data is short, so YMMV, but going 100% PSLDX to 80/20 PSLDX/PFIX delivered comparable CAGR with lower downs, lower volatility, and higher sharpe. Let me acknowledge, I looked at a variety of rising rate hedge ETFs and all were young, all performed well, but this scenario of 2022 was literally the worst time for long stock / long treasury and the perfect time for rising rate ETFs, so there's clearly a bias there.