I’d say that’s a BIT of an overstatement. If you are a retail investor, a 2x long S&P 500 ETF is much simpler than say, investing in LEAPs and rolling them over ever year or so, plus you are able to invest in much smaller “bites” with ETFs than you would be with LEAPs.

Or for example, if you are investing from a retirement account, derivatives may not be allowed, while ETFs are.

There is also the case of wanting to invest in a complex index, like a commodity index that may not be possible to invest in directly with any other single instrument.

And what about the tax implications of ETFs vs continued buying and selling of derivatives?

I think leveraged ETFs are a valuable product as long as you understand them.

I think the best way to short bonds is the way referenced in the article that felix links to. The author mentions at the end that the best way to do it, is to short the unlevered bond fund TLT using leverage. That makes sense. It gives you the similar bet as buying TBT, but acts the way you would expect.

Don’t have a margin account to give yourself leverage? If you don’t qualify for one, then you probably shouldn’t be making levered bets anyway and should definitely not be holding levered ETFs past one day.

Now you argue that holding the levered inverse ETF limits your downside, but if you wanted to limit your downside, then just take off some of your position as it goes against you. You don’t need a levered ETF to do that for you by default and expose you to path dependency, without you realizing it!

Holding an inverse levered ETF exposes you to a maximum downside of -100%. Shorting the underlying security exposes you to a downside of -inf%. (Of course, if the underlying security is a long-term government bond, nobody seriously expects the yield to go above, say, 20%, and that limits your downside in practice, if not in theory.)

My initial observation was mostly not so much that “levered ETFs should be held overnight”, as much as “levered ETFs are the only tool that will provide a, b, and c, to the widest range of investors, because most of them can’t short bonds, many of them can’t buy options, and some of them can’t trade on margin”.

It seems to me that a put is indeed a better way to make the levered bet you describe. You are still exposed to liquidity demands due to the margin requirements of daily moves, but you do not have to pay -1/2 (k-1) sigma^2 in expectations. From a cumulative P&L perspective, the leverage is applied to the payoff of the terminal distribution, not the drift (except for interest on the margin account, a very minor factor in the current environment.)

Suppose you (a) want to bet on the decline of government long-term bonds, and (b) you want to make more than a few per cent if you guess right, and (c) you want to be limited on the downside.

Shorting government bonds is out, because it fails (b) and (c), and also because very few brokerage accounts actually let you short government bonds.

Shorting non-levered ETF’s is slightly better, but it still fails (b) and (c).

Holding non-levered inverse ETF’s meets (c), but fails (b).

The only options remaining are levered inverse ETF’s and options on non-levered ETF’s. I’m not sure if anyone can make a case that levered ETF’s are consistently worse for investors than put options.