Today’s post will demonstrate the estimated and hypothetical performance of several popular 2x and 3x leveraged ETFs during the course of the 2004 to 2007 market. To make it a bit easier to visualize their performance, the daily returns from the historical period have been added to the last day of 2009. (The first days of 2010 have been ignored). In other words, if leveraged ETF XXX is at 10.0 at the end of 2009, and the simulated performance for the year of 2004 was 10%, the chart will show XXX at 11.0 at the end of 2010. If history were to repeat itself and we have a bull market similar to the 2004-2007 period, perhaps some of these leveraged ETFs might have similar performance characteristics as these charts.

S&P500

SSO

SDS

UYG

SKF

FAS

FAZ

TNA

TZA

ERX

ERY

EDC

EDZ

Analysis
Generally speaking, the bull leveraged ETFs clearly succeed when volatility is low and the underlying index achieves significant gains. Even though ERX tracks a relatively volatile index (compared to the S&P 500), the index’s gains were so significant that ERX would have likely experienced significant gains during the last bull market. Alternately, ERY suffers from both significant decay and also decline due to being the inverse of a bullish index. Although these charts put leveraged bull ETFs in a good light, these results are only hypothetical, and future posts on this topic will show the losses (or gains) leveraged ETFs can experience in other scenarios.

If you like these simulations, you can see similar simulations (as well as more data) using QLeverageSim, which is a free utility for helping traders and investors understand the characteristics of leveraged ETFs.

Disclaimer: The results described in this post are purely hypothetical and are not an indication or guarantee of any index’s or leveraged ETF’s future performance. The leveraged ETF simulations are using data from ETFs that track similar but not the same index (such as using XLF data for FAS’s simulation). The resulting simulation data is thus inherently faulty since the leveraged ETFs track different indexes. The simulation results also do not account for fees and other costs.

ADDED: Decay indicator with configurable history
ADDED: Leverage Factor chart
ADDED: An icon for the software
ADDED: Index RSI estimation chart
ADDED: Drawdown chart (for leveraged ETF and estimation of index’s drawdown)
ADDED: Ability to ‘append’ the visible data to the end of a symbol’s data cache (explained below)
ADDED: Checkboxes to show/hide chart series (future versions will remember settings)
ADDED: EEM, UPRO, SPXU to the symbol list
CHANGED: Data is now downloaded from Yahoo Finance
CHANGED: Adjusted Close data is now used (now FAS/FAZ finally work)
FIXED: Series Leverage % label was sometimes wrong.

Click here for the download page. (Requires .NET 3.5 SP1)

Decay Indicator

This feature deserves a blog entry of its own (and it will get one eventually). Rather than show a chart of total decay for the visible data, there is now a configurable indicator that shows the amount of decay over a particular history. For example, a chart of FAS with a 20 day decay indicator shows how FAS was decaying at about 15-20% per month in January through April, but since June it has been decaying at about 1-2% per month. The decay has reduced significantly due to the lack of volatility relative to earlier this year.

A decay measurement can be conceptualized as an approximation of “the amount of loss a leveraged ETF would experience if the underlying index is flat.” In the FAS chart, the ETF has been having significant gains since the index has been gaining significantly. However, if the index were to be flat over a 20 day period, FAS would have experienced roughly a 2% drop. The decay indicator is purely historical and is not intended to be used as a future indicator of performance. Decay is effectively another representation of historical daily volatility.

Leverage Factor Chart

It been covered on numerous blogs and articles how compounding in leveraged ETFs can work to a trader’s advantage during upward trends with low volatility. What ends up happening during these strong uptrends is that compounding can make a leveraged ETF outperform its daily leverage target. Using FAS as another example, since the March bottom there has been enough upward movement in the index such that FAS has gained at a multiple of roughly 4.42x of the underlying index. QLeverageSim’s Leverage chart shows the progress of the leveraged ETF’s leverage factor. It also works for simulations using SPY, IWM, XLF, etc. Try those and see the incredible leverage factors during bull markets.

Drawdown Chart

To make it a bit easier to visualize pullbacks and potential entry points, a chart of the drawdown % of both the leveraged ETF and the index (estimation) can be displayed. Here is a chart that shows the pullbacks in TNA and its underlying Russell 2000 index.

Data Appending

This feature could be somewhat controversial and potentially dangerous, but I like the capabilities that it provides. Basically the feature allows the performance data of the currently visible chart to be appended to the end of the current symbol’s data. As an example, consider a trader who wants to simulate what might happen to SPY if we were to experience another drop in the market similar to the last bear drop in 2002. To do this, a user would display the region they want to copy for appending and then click Edit-Append Visible Data.

Once the data is appended, their SPY chart will contain the same performance characteristics of the 2002 drop at the end of their most recent data.

Another potential use of ‘appending’ is to approximate the kind of market activity that would be required to result in RSI’s reaching a particular value. A less sophisticated trader looking to sell when RSI reaches 70 might try appending/resetting data to see what kind of future data would be required to reach their target value. This technique is certainly not recommended, but since all indicators and values are updated accordingly with the appended data, the feature does allow for traders to ‘play’ with market data to see how past performance applied to the future would affect the resulting values of leveraged ETFs and indicators. As a warning, since RSI is dependent on both time and strength, there is no guarantee that a particular ETF value in the future will result in a particular RSI value. It depends on the ETF’s path to that point.

FAQ

What does QLeverageSim do?
QLeverageSim is a basic (free) utility that can simulate leveraged ETFs from historical data of 1x ETFs. It also displays other information and indicators such as amount of decay, drawdown, effective leverage, gains, and RSI.

Does QLeverageSim make trade suggestions or place trades?
No. QLeverageSim is for simulations and analysis of historical data only.

What is the purpose of QLeverageSim?
QLeverageSim was designed for the sole purpose of helping traders understand the characteristics of leveraged ETFs in a variety of different market scenarios such as trending up, trending down, sideways, volatile, calm, bull and bear.

Who should use QLeverageSim?
Traders and investors who want to understand the performance characteristics of swing trading or long term investing in leveraged ETFs.

How accurate are the simulations and calculations?
Nearly all calculations could be done in excel and give the same results. However, no simulation or calculation includes transaction fees, costs, or taxes.

How much does QLeverageSim cost?
It is free.

Who wrote the QLeverageSim software?
Kevin Kerr (me) over the course of several weekends.

What is needed to run QLeverageSim software?
A windows PC with .NET Framework 3.5 SP1 installed.

Where does QLeverageSim get its data?
finance.yahoo.com

Is there a user manual or guide that explains the various indicators and charts?
A user manual will be written eventually, but currently one does not exist. The decay calculations are best described in the Measuring Leveraged ETF Decay article. Any questions or comments can be sent to my first name (Kevin) at quantumfading.com.

Recently I decided to try to stress test one of my own personal strategies (it is a 100% mechanical mean reversion strategy). I wanted to see how this strategy might perform in a variety of market conditions, such as a random but similar market crash of 2008. Ultimately, I wanted control over randomizing historical data such that I can perform tests against infinite sets of data such as:

1.  Crashes similar to 2008
2.  Crashes 5% worse than 2008
3.  Crashes 10% worse than 2008 (etc.)
4.  Bull markets
5.  Bear markets
6.  Sideways markets
7.  Volatile and nonvolatile

Unfortunately, historical ETF data only goes back about a decade or so. Hence the data is somewhat limited for finding a variety of such scenarios. In some discussions with a friend (PhD in electronics), it was brought up that I should take a look at performing randomization in the frequency domain rather than the time domain. Since we are both more familiar with digital signal processing rather than statistics, it was only natural for us. I will explain the details of the technique in the next post covering this research.

Charting the Randomized Data

I was not expecting much from the results, but a few tests and tweaks gave me data that was exactly what I was looking for. Here are some demonstrations. (green is original data, red is a random data set)

Crash similar/worse than 2008

Crash much worse than 2008

Strong bull market (followed by a severe crash)

Volatile market

Performing frequency modifications gives control over how the data is manipulated. By adding randomness to the low frequencies, steep bear and bull markets can be created. By adding randomness to high frequencies, volatile markets can be created. It is much like how an equalizer for a stereo system controls the amount of bass (bull/bear markets) and treble (volatility).

Purpose of Frequency Domain Randomization

The reason I am interested in this research is not to optimize a strategy against a variety of random historical data sets, but rather to identify risks. For example, my personal strategy has a weakness of fully scaling in too early during strongly trending markets. By changing some parameters to my strategy, I was able to not only maintain the original performance on the original data set, but also significantly reduce drawdowns during significant trends in the random data sets (like a 2008 bear market that drops by 75%).

I will provide more details of the research as it develops.  If the technique proves useful, I can add it to QLeverageSim or just create a standalone utility for generating the random data.