Back to Community
Short a pair of leveraged ETFs with daily rebalancing

In this strategy, I short a pair of leveraged ETFs (ERX and ERY) with daily rebalancing. Theoretically, if we can maintain an equal $ value in both legs of the portfolio at every instant, we can scalp the time decay without any risk. Thus I thought that the result from relatively high rebalancing frequency (daily rebalancing in this case) could be pretty good. However, the backtest result shows that the performance is poor. I'm wondering if it is because the rebalancing frequency is not high enough. If daily rebalancing frequency is not high enough, does it mean such strategy won't work at all since higher frequency could generate critical transaction cost.

Any feedback would be much appreciated.

Clone Algorithm
109
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 57fd262d7ec3b8104c8c5397
There was a runtime error.
16 responses

Well, a big question would be whether the expense ratio of the ETF is going to outweigh the timedecay. Did you do that math?

@ Davin, thanks for the reply. I just want to repeat the results from enter link description hereand enter link description here.
They claim that if we short a pair of leveraged etfs and rebalance the position as frequent as enough(without taking transaction cost and bid-ask spread into consideration), the strategy could outperform S&P500.

@ Davin, so when you refer to expense ratio and time decay, you mean that I should compare the cost of shorting ETFs with their cumulative return right?

Here's the algo with 2x leverage short on ERX, 2x leverage short on ERY and 4x leverage long on bonds. (Might as well do something with the unused cash?) Comission set to $0.0075, $0 min trade cost and 0 slippage.

As Davin and you mentioned, fees, comissions and slippage will cost a significant amount, especially $1 min trade costs. On paper it looks profitable if very high leverage is used. Also, with 4x leverage short, there's some sudden 20% drawdowns during volatile periods, this algo will need to be prepared for tail risks, or perhaps hedge with out of money VXX calls

Clone Algorithm
54
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 57fe4fea2bd5c2105367ac79
There was a runtime error.

@Kayden,I'm wodering how the long position in bond affects strategy performance.

Also, what do you think about changing the ratio of bull and bear etfs? I think the pnl of shorting leveraged etf pairs is similar to shorting delta-hedged straddle, thus we are actually shorting volatility without worrying about market direction.

However in this paperLeveraged Performance of Shorted Leveraged ETF Pairs, it combines shorting leveraged etf pairs with longing 3M treasury bills. In this strategy, it shorts the bear triple-leveraged ETF and the bull triple-leveraged ETF in a 2:1 proportion with +20%/-20% rebalancing , and simultaneously holds Treasuries long. It also limits the total short position in the combination of UPRO and/or SPXU to be equal to 75% of the portfolio value. My concern is that if we short leveraged etf pairs with ratio rather than 1:1, it's no longer delta-neutral and it's actually betting on market direction, which is not what i want.

Thanks.

@Kayden, about out-of-money VXX call, have ever test the hedging strategy? Which term and delta you think is good?
I'm not sure if it's a good choice for hedge purpose. Based on my study, from 2010 to 2016, 1M OTM VXX call with 0.2 delta only ends in-the-money once, which happens during the market turbulence in Aug 2011.

@Lu Cao

The bonds is a seperate bonus income. I figured if we are short at 4x leverage, there's a big pile of unused cash and we might as well put it in bonds and get some risk-free money out of it.

Regarding ratio of bull/bear ETF, I skimmed the paper and it proposed a high beta strategy, but with better risk adjusted returns than the market indexes,something very different from your strategy.

I'm with keeping delta neutral, but I'm thinking if we allow +X%/-X% hedge ratio leeway it can reduce some slippage and transaction costs. It in turn increases risk, hence I was thinking about a tail risk hedge but haven't looked into that yet.

Also, any chance you know what happened in nov 2009 with that sudden gain and loss?

@Kayden,

It might have something to do with such SHY price pattern SHY from nov to Dec 2009?

@kayden I'm curious about your "big pile of unused cash". Which broker lets you spend that "cash"? It's been my experience that long or short positions require maintenance margin, and often short positions require more margin than long. So, by going short, you don't suddenly have more money to spend, you have less buying power.

@Simon It really depends if you are a retail trader or an institutional hedge fund, as the way margin is calculated changes significantly.

If you are a retail trader and go long $1mil AAPL & short $1mil SPY, then you will be margined based on a $2mil position, i.e. your broker uses a very unrealistic model of the world (just because it's cheaper and safer being a more conservative assumption) and doesn't care about correlation between AAPL & SPY.
On the institutional side though, funds that have accounts with prime brokers (MS, Goldman, Citi, etc) are being margined through the use of much more realistic models which in fact would in this case say that by doing long AAPL short SPY you have less risk compared to being long only $1mil AAPL. So they would charge less margin.

Now in terms of short-selling a stock, funds would locate a borrow from a broker (i.e. they would borrow the stock from someone and pay borrow cost on that) and then sell the position and receive cash. That cash needs to be invested at least in an overnight rate facility to mitigate the cost of borrowing the stock.

So to summarise if you are building a strategy for your personal account you are right, but if this is a strategy meant to be run by an institutional fund there is more to it.

Also i forgot to make one final point: That prime brokers are quite flexible with the use of collateral. That is if you are a fund and have an account with a prime broker and receive cash from a stock sale and go buy bonds with that cash, if and when more collateral is required you can post those bonds with the prime broker. (no need to liquidate and post cash only as margin)

Simon is right, you won't be able to use the cash. Also, the broker will limit your leverage to below 2x, as a practical matter. I would max it out at 1.8x at most. Assuming you are holding your positions over night.

I looked at this a year ago. I focused on the geared VIX ETFs, UVXY (200% long the S&P 500 VIX Short-Term Futures Index, which is ticker SPVXSP) and SVXY (100% short SPVXSP). I tried to figure out how to short the UVXY using SVXY to hedge. I looked into their underlying holdings, tried to figure out how contango and the poor trading of a geared ETF was impacting them, etc. I came to the conclusion that there was no opportunity as long as the guys managing the ETFs did their job. But I couldn't really test it. Enter Quantopian.

Using Quantopian, I backtested a simple paired strategy, short 1x UVXY and short 2x SVXY. Please see attached.

Clone Algorithm
39
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 5800da03e4897d13182be2cc
There was a runtime error.

It's a great idea. I researched it about 7 years ago only to find out that the carry costs of shorting these ETF's can be high, and locates can be hard to come by.

Yes, indeed. The leveraged ETFs (at IB) had short borrow fees of 7-15% annualized, last I checked, which handily wiped out all these gains.

Can confirm. You have to pay to borrow securities like this. Moreover, they cannot always being located for short which makes it especially tough to scale.

Agree with Taylor Smith, the costs associated with sec lending typically around the 70/30 - 80/20 mark as well as the sourcing of such securities adds to the real world fees you would be expected to pay and therefore scale