Thursday, December 04, 2008

Trading levered ETFs is only slightly less foolish

than buying lottery tickets.

Here is some commentary I received from a fellow trader earlier this week:
Both double-long AND double-short ETFs behave as short gamma, i.e. they must buy more to re-hedge the more that the market rises and sell more as the market falls.

Q. Why do the levered / inverse ETFs not exactly track the underlying index? Why is XXX only up 5% if YYY is up 3% and it's meant to give double the upside?
A widely held misconception about these funds is that they will offer twice the return of the underlying index, which means that if the S&P 500 returns about 10% a year, then the SSO should return 20%.

But that’s not true, because these funds only double the daily return, and there’s a big difference between doubling the daily return and doubling the annual return.

An example: Let’s say that one day the market goes up 10%, and the next day it falls 10%. The two-day loss for the index is 1%, but the loss for the leveraged fund is 4%. Here’s why:

Index: (1 + 10% ) x (1 – 10%) = 1.1 x 0.9 = 0.99, 1% loss
2x Fund: (1 + 20%) x (1 – 20%) = 1.2 x 0.8 = 0.96, 4% loss
Over a two day period the fund’s losses are 4x the amount of the index, not 2x. This example actually comes from an actual fund prospectus, and is a clear indication that investors in 2x funds should not expect their investment to provide double the return for any period longer than one day.

They are also trying to hit a moving target to some extent, as they need to re-invest at the close each day without knowing what the close actually is until after the trading day, which leads to slippage.

Q. These levered / inverse ETFs sound fantastic - why would anyone trade anything else?
They've certainly proved to be extremely popular so far, and while they certainly can enhance ones "firepower" for a short-term directional bet there are several potential drawbacks an investor should be aware of - in addition to the fact that they will likely not give 2x (or -2x) the returns of an index over any length of time, as discussed above.

Firstly, they can be an expensive way to get levered exposure to an asset compared to traditional methods, e.g. by trading swaps, options, futures or even something as simple as trading on margin. Total expense ratios are typically around 90-95bps per annum, making them amongst the most expensive ETFs in their own right, but more importantly far more expensive that a more traditional levered trading vehicle such as a swap. This is for a couple of reasons: firstly that management fee is largely retained by the fund, not used to pay for the leverage. Any trading costs the fund incurs to employing its trading strategy simply eats into the returns of the fund, so an investor is essentially paying another layer of fees over and above what they would do if trying to replicate the strategy themselves.

Secondly, liquidity is often worse than in comparable "vanilla" ETFs, which means higher transaction costs may well outweigh whatever commission dollars an investor saves by having to trade less shares for a given notional amount of exposure. Again there are several reasons for this, but one of the major ones is that it's harder for market-makers to hedge the ETF since they cannot hedge precisely using related futures and options (the delta of the fund varies intra-day and thus the market-maker cannot know exactly what notional he needs to hedge - see the next section for more on this) which hampers liquidity, but also that the funds are simply more volatile and the depth of the market is typically lower than with a more traditional ETF which, ceteris paribas, results in wider bid-offer spreads and more market impact.

Q. Does the growth of the levered funds have anything to do with the rise in end of day volatility?
It's certainly part of the story. Let's take the example of a double-levered ETF: the fund must attempt to rebalance their net delta to 2x (or -2x) the notional of their fund, as of the close of each day, in order to attempt to return twice the following day's return. Along with the fact that the bulk of these funds use delta-1 instruments to get leverage (primarily swaps, but also stock portfolios in margin accounts, ETFs, futures - most don't trade that much options so far as we know), this means that the brokers handling this business do the bulk of their trading into (or at) the close each day in listed equities or futures, and the size of the trades they must do is a function of the magnitude in the move in the market that day (and the size of the fund, which can be considerable e.g. SDS (double-short SPX) has $6.5bn AUM).

Most interestingly however, both double-long AND double-short ETFs behave as if they are short gamma, i.e. they must buy more to re-hedge the more that the market rises and sell more as the market falls. It's our contention that a good portion of the violent late day moves we have been seeing are more due to the recent explosion in AUM of the levered funds and the fact that they must rebalance each day (in addition to the effects of dealer hedging of variance swaps or index option gamma), rather than the more commonly-cited reasons in the media such as mutual fund redemptions or attempts by un-named groups to "manipulate" the market close.

No comments:

Post a Comment