“These Aren’t The Gains You’re Looking For” (Leveraged ETFs)

“These aren’t the droids you’re looking for.” (Flickr/donsolo, CC BY-NC-SA)

Ah…leveraged ETFs. All that beta without all the embarrassment of trying to borrow money from your broker (or cousin) to leverage your returns. Is the regular ETF not volatile enough for you? Buy the 2x version! Still too tepid? Perhaps the 3x version is what you’re looking for.

Unless you’re new to trading though, you’ve probably heard about beta slippage. This is where leveraged ETFs, using a combination of volatility and pesky math, end up losing as compared to their 1x versions over the long haul. Leveraged ETFs aim to provide a 2x or 3x return on an intraday basis as compared to a 1x version (or the index they’re tracking). But, over time, these leveraged returns decay.

For example, let’s take a look at two gold miner ETFs, which track an index of gold miner stocks. We have the plain vanilla version, GDX, and then a triple-leveraged version, NUGT. These are managed by two different investment funds, so we’d expect there to be some day to day discrepancies. Starting at the beginning of 2011, here are the results of a buy and hold approach to each:

click to enlarge.

Clearly not a winning trade, but that’s not the point. Look at the end values of each: GDX is down 29%. If NUGT were behaving like a long-term triple-leveraged ETF might, we’d expect that percentage to be down 87% or so. Instead, it’s down a whomping 109%. That makes it a 3.75x leveraged ETF on the down side.

If we take the daily returns of NUGT and divide by three, we can get a better sense of how the two ETFs track. Closer, and the differences are overwhelmed by the shear amount of volatility in the underlying. But the differences are still there.

click to embiggen.

One might naively expect these two ETFs (with NUGT having been normalized) to end up at the same place. Instead, GDX is down 29% vs normalized NUGT is down 36%.

Clearly then, beta slippage is a real thing for buy and hold.

What about short term trading though? Do we still see the same problems, even if we’re in and out of a trade over a period of a few days?

I’ve got this nifty long/short gold miner trading system which I initially developed using index data, which you can’t trade directly. The index data goes back much further than any ETFs that follow the gold miner index, so it made sense to develop a system using as much data as I could get.

When it came time to implement the system in real life, I first had to make sure that it was profitable using actual tradable ETFs (it is!). Then I had to decide whether I would be using a leveraged ETF or not. That’s where this research comes in.

First, I tested the system using a fixed position size for each trade. Since NUGT has a back-adjusted value in 2011 of almost $4000 per share, I needed to make my position size and starting account pretty large. As a comparison, GDX was trading at around $60. The starting equity is therefore $3 million. The system traded position sizes of $300,000 on GDX, and $100,000 on NUGT (since it’s a 3x leveraged ETF). The results should be similar, right?

click to magnify.

GDX outperforms a little here and there, but then NUGT catches up. Running the system on GDX provided a slightly better return/drawdown ratio, but the difference was pretty marginal. GDX made 16.9% profit on the original account, whereas the NUGT system made 15.9%.

However if you run the system as an “all in”, compounded return system, where the entire account is traded each time, we see profits like this:

click to get really huge.

Note that NUGT starts out with 1/3 the cash that GDX does. And yet, with compounding, NUGT has returns that would make a George Soros blush, with drawdowns that would make Arnold Schwarzenegger cry like a little baby. With a crazy curve like that, it’s tough to compare.¬†Fortunately we have numbers for that.

NUGT had a CAR/MDD (compound annual return / maximum drawdown) of 1.74. This is slightly better than, but very similar to, GDX’s system at 1.68.

The bottom line though is that GDX, even starting out with 3x the cash, made a mere $10 million in profit, whereas NUGT made close to $50 million. That’s 393% profit for GDX vs 4680% profit for NUGT.

Using your entire account without a cash buffer means that your risk of ruin is very high when trading NUGT. That’s a discussion for a different time or perhaps a different blog. I am not recommending that you trade a 3x leveraged ETF with your entire account!

The takeaway is this though: just because a leveraged ETF shows beta slippage over the long term does NOT mean it will show the same slippage when trading on a short-term basis. These leveraged ETFs were designed for short-term trading. My research has led me to believe that trading the 3x ETF with an appropriate position size is a good fit for a diversified collection of swing-trade systems.

P.S. for more interesting reading on beta slippage and ETFs, check out this post on Signal Plot.



4 thoughts on ““These Aren’t The Gains You’re Looking For” (Leveraged ETFs)”

  1. Thanks again for linking to my blog. I’d be interested in hearing more about your gold miner system.

    On a side note, the SEC is considering limiting the amount of leverage available to ETFs since some retail investors don’t understand the beta slippage. I hope they don’t though because as you say, leveraged ETFs still have their place as a convenient form of leverage for professional investors.

    1. Thanks for your comment, Kevin. I’d heard about that possible SEC change I think early last year, but haven’t heard anything since. Wonder what’s up with that? I personally want to see them stick around. I mean, you as an investor have to do due diligence before putting money into something. Surely a big disclaimer on the ETF’s site ought to be enough for big boys and girls?

    1. Thanks for your comment, Tonio. No, my system definitely is not buy-and-hold…that would be the first graph I showed. The exact nature of the system is not germane to the blog post (and not something I’m prepared to reveal publicly). The point was to show that beta slippage experienced over a longer period was not necessarily a factor for shorter trading time frames.

Leave a Reply

Your email address will not be published. Required fields are marked *