Let's talk leverage!

Manage
1

One of the advantages of investing in physical real estate (not REITs) is the leverage available. While investing in real estate is not nearly as easy as it’s made out to be IMHO (that’s a entirely different post for another forum), the leverage available can really amplify returns.

Needless to say leverage is a double edged sword for stocks. I’ve dabbled with leverage earlier in my quant investing journey. I had developed some strategies which backtested nicely, and I borrowed to invest in them. As it turns out one went quite well and popped early, and then fizzled, while the other fizzled from the get go. In hindsight these strategies were highly optimized (only 5 stock ports) and not very robust, and not well suited for leverage over the long term.

I’ve spent much more time recently on creating more robust strategies, and have them running in real time/OOS with good results so far. I’m reluctant to apply leverage based on my previous experience, but I’m curious what everyone’s experience with leverage is, i.e. are some types of strategies more suited for leverage, any specific techniques, etc?

Thanks,
Ryan

2

Hello Ryan,

I have posted a few times that a simple QLD, TLT strategy since 2009 would be a dream. Unfortunately 2008 you would lose 75%. Hard to swallow. My current leveraged strategy is up 50% in 4 months. I have a very small amount of capital in it. It is always Hedged with Bonds and Gold with different weightings every month. Simulation results show a 24% drawdown but I know it will be much worse. Can I survive the next big correction will have to wait and see. I also use leverage ETFS as part of a book 10% max allocation. The rest of the book is very low volatility with uncorrelated assets. This way I get the best of both worlds. I think you can use leverage but you better know the risks and not more than 15% of the total book as a general rule. Imagine the opposite -50% in 4 months and that’s why leverage is so deadly.

Cheers,
MV

3

You don’t end up with nothing if your real estate falls too much. You just might be underwater on your debt to assets. That doesn’t really happen in the stock markets. Leveraged ETFs can go to zero and margin calls can leave you with nothing. Plus real estate is much more likely to pay returns even if they are reduced. Stocks often drop dividends in a market downturn because company cash flows drop. Which brings up another point. Anyone who proclaims that dividend stocks are good during downturns assumes they will keep paying out. Not necessarily. Even with low yield treasuries will continue to provide income. Dividend stocks may look like bonds, but they don’t function like bonds in times of distress. A dividend stock can just as easily stop paying dividends.

4

I’ve recently started investing in leveraged ETFs such as TQQQ. It’s part of a lifecycle investing strategy, where you use leverage when young, and a deleveraging glidepath as retirement approaches.

I prefer leveraged ETFs over a margin account on individual stocks. No margin calls, and the LETF can’t go negative. And if the LETF goes to zero, the unleveraged index is probably going to zero too.

As for mortgages, I feel that people ascribe an unwarranted “safety” just because you can see and touch a house but not a stock. If mortgages are “safe” why doesn’t everyone leverage to infinity and make infinite money? The bank can call your mortagage and ruin you, this happens to reckless investors.

5

I have invested in rental property, private lending MBS, and leveraged stocks. I have made money and lost money in all areas. The key to any leverage is to use a risk adjusted amount of it. People leverage their home because historically populations increase and supply is always needed to meet demand, therefore there’s usually a good resale market and risk is low. The lower the risk, the more you can leverage. In stocks we think of the risk being higher but it’s a sliding scale. The risk of owning TSLA right now is high IMO but the risk of owning WEC is low by comparison. I think the price gain potential may be higher with TSLA but where is the leverage trade off you can apply to WEC? In other words, how much leverage can you put on WEC before risk:reward exceeds that of TSLA? At least that’s how I like to think about using leverage. There’s no hard stop % to put on it because it’s security dependent and how you foresee the reward potential. The less risk the easier to project and that’s another reason investors love real estate, They may purchase in a historical 5% market that has not seen more than a 3%DD over the past 20yrs so they can project that 20yrs down the road they should have 170% return. That’s a good place to use leverage but again even in real estate there is a point of max levered risk and that’s why a bank won’t let you buy a house for a $1.

Just remember that you can’t play the game unless you have money so not losing money should always be your primary focus.

6

One of the best ways I have found to discuss leverage and different Assets is to understand the following:

  1. Risk and Volatility
  2. Uncorrelated assets
  3. Global asset Diversification across all asset classes

There are many papers on this going way back. They fall into many categories but all have the same principles. Here is a good paper on it.

https://inside.bwater.com/publications/engineering_targeted_returns_plus_risks

Other Articles to search for are:

Tactical Asset allocation
Risk Parity models

Recently I have started listening to https://investresolve.com/blog/. These interviews with different Asset managers explain the concepts from many different angles. Obviously they are very biased towards this type of investing but it has worked for the last 20 years extremely well. The reason is Stocks and bonds are highly uncorrelated for the last 20 years. That was not the case over the last 100 years but for now it is.

Cheers,
MV

7

Mark,

Thank you for the link!

I had been doing the optimal beta portfolio without leverage with some of my money. I had been selecting assets with similar expected returns and constricting an optimal beta portfolio from those assets.

I had not thought of using leverage to get additional assets with similar expected returns and enable even further diversification.

Much appreciated.

-Jim

8

I was very averse to using leverage until a few months ago, when a good friend pointed out to me that withdrawing from my retirement accounts to pay tuition costs, home improvement, and (in the future) post-retirement living expenses would be far more costly than getting a huge cash-out refi and using the leverage for investing, as long as you plan to pay the loan out early. Obviously, you’d face major sequence-of-returns risk, but I did the numbers.

Here’s the scenario I ran. I got a $320,000 loan at 3.75% for 30 years (cash-out refi) and I plan to pay it back as soon as possible after four years but in any case within ten. Of that $320,000, I pay $205,000 over the next four years for various things–my previous mortgage, tuition payments, home improvements, expenses. After that, I continue to withdraw $15,000 a year for living expenses.

The alternative is to deduct $152,000 from my retirement accounts over the next four years and not take out a loan, instead just finishing paying off my mortgage.

In both scenarios, I would shift $14,000 per year from my cash account to my retirement accounts.

I then ran 32 different scenarios. In each one, I assumed I was invested in all accounts in a global index fund (I used the Wilshire 5000). Each scenario had a different starting date, from 1978 to 2009. In each scenario, I would end the mortgage as soon as my savings in my cash account exceeded the principal owed, but in no case would I keep the mortgage more than ten years, even if it meant depleting my cash account entirely and withdrawing money from my retirement account in order to pay the principal owed. In addition, I would be paying capital gains taxes in the leveraged cash account and no taxes on gains in the retirement accounts.

If I had started in August 1983, I would have lost more money using leverage than I would have using my retirement accounts. But in every other starting year, I would be ahead using leverage. In six out of the 32 scenarios I’d be able to pay off the principal within four to six years, while in the rest I’d keep the loan for nine or ten years. The advantage of taking out the loan over not doing so averages out to about $230,000. So I figure I have a 96% chance of doing better with leverage than I do without–if I stick to this strategy carefully.

Everyone’s circumstance is different. But cash-out refi’s have such low interest rates right now, that it’s worthwhile looking into.

Of course, it’s also possible that I made a big mistake somewhere in my calculations. But I’ve gone over it a number of times and can’t see where I might have gone wrong.

The reason August 1983 is such a bad time to start is that the ten-year forward CAGR of the W5000 at that point is less than 5%. There are a few other times when the ten-year forward CAGR is that low–1999, 2000, 2006–but not in August. If I had run 32*12 scenarios starting in every month of every year, I might have gotten a few more failures, but I still think there’s at least a 90% chance of being ahead at the end.

A big reason why this works is that you’re removing some sequence-of-returns risk by paying back the principal at the earliest possible time. In other words, if the stock market does well, you get out early; if it doesn’t, you stay in until it does.

9

When Yuval mentions it that way I am indirectly leveraged. I continue to pay my minimum mortgage payment and instead invest any surplus I have. To me it seems like the smartest choice is to focus leverage on lower risk assets. Portfolio margin works that way anyway. If you buy higher beta investments your allowable portfolio leverage is lower.

I wonder how well portfolio margin holds up in a system risk event though when asset correlations go to unity though as often happens in crash scenarios. One needs to really do diligence on managing tail risks with margin as a crash can wipe you out. Don’t assume assets that aren’t currently correlated will maintain that during a crash scenario. When liquidity leaves that spills into other more liquid assets. History does not provide a definitive either because the risks and exposures are different every time.

10

Well said.
I’d be very leery of using leverage. It is easy to make assumptions about the past and how wonderfully profitable it would have been to use leverage. But a big and quick Black Swan event can wipe you out at any time and as Schm pointed out, that is totally unpredictable. And Black Swans WILL come. They always have. The only question is WHEN.

13

Duckruck,

If one were to try to objectively calculate the amount of leverage to use one might think about using the Kelly-criterion.

You correctly point out that the usual way of calculating optimal-Kelly does not take into account the fat tails.

Fat tails are often addressed by using a t-distribution because, of course, a t-distribution has fat-tails which is logical enough.

In fact, you can control how fat the tails are with a t-distribution with nu. nu = infinity is the same as a normal distribution. Anything greater than 10 is pretty much like a normal distribution.

So a t-distribution has 3 parameters. The mean, sigma and nu.

I would suggest using a t-distribution with sigma = the standard deviation and nu = 3. Mean = mean of your data, of course.

Use this for Kelly and for your Baysian priors. Or wherever you encounter the problem of fat tails.

BTW, there is also the problem of skew sometimes. I use bootstrapping to address that.

One link to support this: A Prospect-Theory Approach to the Kelly Criterion for Fat-Tail Portfolios: The Case of the Student t-Distribution

I am not claiming that would answer every question on how much leverage to use but it would be one method to take into account the fat-tail problem objectively, ahead of time.

Obviously, looking at how your portfolio would have done during worst-case periods would be another good method. A kind of stress test. I think may be what you are suggesting. That is also a good idea.

Jim