But then you risk losing a lot of money if the market trends. Rebalancing daily you would have made money every day. And I did.

But the trade isn't as attractive with less volatile markets, more efficient rebalancing and higher borrowing costs.

how did you make money every day? There are days where being short both sides in equal $ amounts does not yield a profit if looking at closing prices. How were you timing the market such that you managed this every day?

Okay, slight exaggeration. But the profit graph was straight up with only very tiny declines along the way, which were essentially just noise.

In any case, that's not the point. Leveraged ETFs don't suck because they have mathematical deficiencies. The fact that they decline in choppy markets is a feature, not a bug. They suck because they have to rebalance huge portfolios daily and they're trading against an intelligent market that knows in advance that they have to rebalance.

The consensus I have heard from the ETF providers and those shorting them is that the math is the main reason behind the short and its definitely the main driver when not rebalancing daily (which is also more profitable over time). I know its the main reason I have done it. The rebalancing effect is very small compared to that.

That's why the poorly informed retail investor shorts leveraged ETFs. It's not why quant funds with lots of money who drive up borrowing costs are shorting them. Though it's possible it's why hedge funds that exist solely to gamble with other people's money without adding any alpha short them.

The math element of leveraged ETFs has always seemed extremely flawed to me, even though 99% of online resources will spout that logic (usually without even included the trending caveat). From what I've been able to determine what stinkypete says is correct although I'm interested to hear other sides.

Shorting pairs of leveraged ETFs without rebalancing is nothing more than a bet on mean reversion. It has worked wonderfully in the short time these ETFs have existed.

If we ignore the money these ETFs lose to rebalancing costs, it's nothing more than a martingale bet. Your bet keeps getting bigger and bigger as markets trend in one direction or the other, and you make money when they mean revert. But you'll get destroyed as soon as there's a steep sustained trend, because your position size will keep getting bigger as you lose more and more money.

There's a million ways to bet on mean reversion, and this isn't a particularly good one.

You are rarely going to be right arguing math against sp. In this case everyone else is basically clowning around making **** up and he is right.

I'm only talking about quant funds and hedge funds. There are multiple funds that use this as part of their long term portfolio. The math works as a mean reversion trade as you explain. The growing position size in a trend is a benefit not a negative. Rebalancing out of the larger positions costs you money over time.

Can you provide evidence for this claim?

No investor with any level of sophistication is doing this as a mean reversion trade. There's much better ways to make that bet than shorting an expensive to borrow ETF long term hoping the market moves so you'll increase your effective position size by an arbitrary amount.

Like distribute their marketing books? No I can't do that. What kind of request is this?

Not all of these etfs are expensive to borrow either. I've been short plenty of them at under 2%.

Talk directly to the etf providers if you want data on the size and duration of trades that some firms have on these. Some funds share this info with them. I've gotten numerous examples from direxion.

I'm not saying you can't make money shorting these things. I'm just saying your explanation for why they're garbage derivatives is wrong. And that all the stuff on the web that suggests they're essentially guaranteed to lose money because of the "math" you presented over time is garbage.

Explain how they make money over a long time horizon. Not a single double short strategy with a manageable borrow for etfs that has been around for 7+ years is profitable last time I updated my data like 3q of last year.

How would you have done if you shorted $1000 worth of FAS and FAZ at any time 5+ years ago and held that position until today without rebalancing? Ignore borrowing costs if you want.

Wait, I'm confused. Are you saying that shorting both sides of the pair and holding is a horrible trade now? Because that's what I've been saying all along.

I'll point out that I was wrong about this. I hadn't looked at recent data because I don't trade these things anymore. It hasn't worked well lately, in a trending market, as predicted by "the math", for the reasons I explained above.

will have to look tomorrow.

I misspoke. I should have said "NOT profitable" considering I started the sentence with a negative.

Not all sectors have been trending. Baskets work better than pair trades, but pair trades work fine given enough time. No markets trend for decades at a time.

So why is this now a bad strategy but used to be good? Because markets have been going up for 8 years?

Thanks for the advice. Interesting stuff. I also found this interesting:
http://www.signalplot.com/the-defini...everaged-etfs/

The double short with daily rebalancing worked well when markets were crazy volatile and borrowing was cheap.

The volatility helps because when the underlying moves a lot, the fund has to make a huge trade at the end of the day to get back to the same leverage.

So if a 3x leveraged fund is trading at $100 and the underlying goes up 5%, they're now at $115.

Before the move they had $300 worth of exposure, and now they have $315 worth of exposure, but they need to get up to $345. So they need to buy $30 worth of the underlying.

What if the underlying moves 10%? Fund goes from $100 to $130, exposure goes from $300 to $330 but we need to be at $390, so we need to buy $60 of the underlying.

As you might imagine, when you have a billion+ dollar fund that has to make a trade equaling 30%+ of its assets under management at the end of the day, it might not be particularly cheap to make that trade. I don't know exactly what kind of deals they had to make these obviously huge block trades but it's pretty easy to see that trading costs/slippage would have been huge. So in 2009 pairs like FAS/FAZ were losing huge amounts of their assets to trading costs/slippage, because it's outrageously stupid and expensive to have such high turnover on a daily basis.

I obviously wasn't the only one that figured this out, so borrowing costs were eventually driven up to a point where this trade isn't blindly profitable anymore, but I'm sure there's plenty of quant funds out there (hire me plz) with much more sophisticated versions of essentially the same trade still making money on it.


As for the short equal dollar amounts and hold trade, it was always stupid because it's a very very high risk trade relative to the amount invested with no more alpha than the low risk daily rebalanced trade.

The decay (these days, in the few volatile ones I looked at, NUGT+DUST, UWTI+DWTI) is much more from the math than the slippage. The pair-short-and-hold trade sucks because these things can and do 10x themselves out of the blue and shorting that for any decent percentage can bury you.

Like I've explained, that's true in choppy markets when they mean revert. When they trend, that "decay" is actually positive, so it's wrong to call it decay.

The decay from slippage is real, so in the absence of borrowing costs you can't lose money on a double short daily rebalanced trade unless there's a big enough move to wipe out the funds entirely (or there's some other anomaly that causes them to track incorrectly).

It's expected decay. Even with no expected mean reversion, the things decay eventually by some analogue to the central limit theorem. The trades suck because trends are volatile enough to wipe you out on even single-digit percentage stakes, not because a pair short is -EV in a vacuum without slippage. ANY short you can't let go of is effectively signing up for a martingale though, so that's not a special critique for leveraged ETFs.

If I understand what you're saying, that's not strictly mathematically true. It depends on your assumptions about volatility vs. expected return. But I'd be interested to see you expand on this.

True, but with the leveraged ETF double short your martingale bet grows exponentially with price moves rather than linearly in a regular unleveraged short.

Actually I'm pretty sure its strictly mathematically false. No mean-reversion would mean 100% autocorrelation which would be amazing for these products.

Edit: I get that he didn't mean it to the extreme but it seems like he's just waving around the CLT