Moving Average Backtest Page Added To Site, Have A Look

Apr 13, 2010 in Backtest | moving average

New application added. This simple application generates a report for any ETF in our database based on the user-defined rules regarding a moving average cross.



Moving Average Backtest Page

Comments and feedback appreciated.



Hedge Funds and ETF's

Apr 12, 2010 in Hedge Funds | Volatility

In the world of hedge funds, it is standard practice to list a ‘volatility target’ within a presentation to potential investors. You never see this listed in any mutual fund or investment advisor type of presentation. Nevertheless, whether it’s in the presentation or not, it’s a topic that is important to post about as often as possible.

First, volatility is a good way to think about ‘drawdown potential.’ High volatility means that the high to low intermediate moves are likely to be large – and since you can never be quite sure what the future will bring, you should generally avoid the highest volatility ETFs unless you feel especially confident in a high return expectation. The relationship between your general return expectation and the underlying volatility of the ETF is an important one.

What hedge funds state in their presentations is a ‘volatility range’ to expect – that is, what the hedge fund manager believes their strategy equates to in a bottom-line percentage, always stated as an annual figure. Many presentations then try to target a ratio of returns relative to that volatility figure. Two such examples are listed here:


Target: <15% Volatility with >15% Return



Target: 8-12% Volatility with 1.5x+ Return




Both of the above examples from actual hedge-fund marketing books are stated within the same structure: namely, the Sharpe Ratio. Return and volatility of return are both used as quantitative targets.

How does this apply to ETF’s? Each ETF has its own same characteristics. Viewing total return and annualized volatility for each ETF is a nice breakdown of the major components of the Sharpe Ratio. Moreover, you can COMBINE ETF's into portfolios that suit your own risk tolerance.




One major issue for all investors is that volatility is not static. Large changes in market volatility complicate the discussion of absolute targets of volatility. But what we can observe from actual experience is that RELATIVE volatility across different types of ETFs have been quite consistent. Below is an example of a few different types of ETFs. You can see that while the LEVEL has swung around significantly, if we were to rank each of these ETF’s – you would see that each ETF has maintained its exact ranking for every period for the last few years. Emerging markets have maintained high relative volatility vs the S&P 500, which has been more volatile than the defensive Consumer Staples Sector SPDR which in turn has been more volatile than the aggregate bond market.



What this says is that the risk of drawdown among these different ETF’s is skewed. We cannot precisely say what kind of risk there is --- but we can think that if we entered a long position in bonds and mis-timed the entry, the punishment for being wrong would not be as great as if we did the same thing in Emerging Markets.

It is professional to think in terms of volatility and risk-adjusted returns. But you do not need to be a ‘long-short’ hedge fund manager to maintain an efficient (risk-adjusted) portfolio. Overall portfolio volatility can be diluted through exposure to shorter-term fixed income. Indeed, rotating between high returning ETF segments (high relative strength) and low-volatility investments is a strategy that generally leads to the same place hedge funds are ultimately targeting: a high Sharpe Ratio.

Note: The Sharpe Ratio measures reward per unit of risk. It is calculated as the annualized average daily excess return divided by the standard deviation of daily excess return.




Global Asset Class Rotation

Mar 30, 2010 in Backtest | Relative Strength | Screener

This is to highlight some specific thoughts on the important topic of global money flows and its implications for investors.   ETF’s have caused a major shift as they allow cheap ways to access new markets.  It is our belief that this innovation will cause the discussion to increasingly become ‘which MARKETS should I own?’ – and less about ‘which stocks should I own?’

Below are some performances of a few of the major ETFs over the past three years.  I am starting at a very high level here and then working towards my ultimate point of coming up with a process for interpreting global money flows.  (Note that all returns posted here make the proper dividend and distribution adjustments as total return is absolutely essential in any professional discussion of performance)

The point of this slide is to simply highlight that even if trading long-only, a well-executed high-level rotational strategy could have produced strong, consistent returns.

The next image highlights how a combination of 80% bonds and 20% emerging markets has performed over the past 3 years.  Most investors think myopically about returns.   The discussion of returns, without the context of risk, is meaningless to professional investors.  

Note how despite a massive volatility spike in the overall marketplace, the standard deviation of daily returns for our 80/20% portfolio was quite low.   This assumes no rebalancing.  Value could have easily been added over this return with a relatively simple re-balancing rule.

This brings us to the Sharpe Ratio. The important aspect the Sharpe Ratio framework brings is to factor in ‘drawdown potential.’   If you own securities with higher relative volatility, the high to low moves will be larger and cumulative negative returns become increasingly difficult to overcome.  By thinking in terms of a sharpe ratio rather than returns-only, you will inherently adjust for the ever-present possibility of drawdown.  You cannot fully know the risk that awaits -- but you can at least think in terms of reward/risk.

To make this practical and to keep this from becoming a dissertation, I will just simply show one way to de-compose the Sharpe Ratio into a multi-factor model that allocates toward the segments of the market that show strong combinations of high relative strength and lower relative volatility.  This model uses two timeframes to calculate return and one part volatility.   It is considered a statistical model, a subset of the APT framework – and can more easily be thought of as ‘risk-adjusted relative strength.’

Below is an image of the model as it stood on June 30, 2007.  I can 'roll back' the model to any historical date through the calendar control in the top right corner.  Global asset classes are color-coded for ease of understanding what is showing strong relative strength.  International equities and select US equity segments were leading the market on a risk-adjusted basis during the middle of 2007.  Though not shown in the image, you can go to the site and view what was lagging (hint: housing and financial segments).

Fast-forward 1-year through setting the calendar control to June 30, 2008.  At this point, the screen is dominated with fixed-income ETFs. One way to interpret this is that global money is flowing INTO bonds.  These kinds of shifts are likely not whipsaws as you might find with the US-based Sector SPDRs --- which can have violent rotations even within a secular bull move.

This was one example of a global asset flow shift.  I could run this ETF Screener 50 times, run the results through our Backtest ETFs page and then post all of those images here.  But to truly understand this, it is better to interact with the data yourself.   It takes time to understand complex financial relationships and it is nearly impossible to see this in just one-dimensional snapshots in time through some images.   Change the model inputs, vary the timeframes.  This tool is meant to enable users to interact with each other at a fundamentally higher (risk-adjusted) level of conversation.


A few ETFreplay details

Mar 26, 2010


A nice feature of ETF's is that the ETF providers issue holdings on a daily basis -- so you can always access up-to-date information of how the ETF is weighted and its sector allocation or a number of other useful pieces of information.    This image of an internet ETF shows the holdings of FDN --- with other useful info just a click away on the First Trust website.



For fixed-income ETF's, we find its helpful to add the approximate duration of the ETF (in parentheses after the name) so when looking at a number of ETF's in a list, you can quickly understand the ETFs approximate sensitivity to interest rates.  Again, other ETF statistics (all updated each night) are just a few clicks away.


Putting it together -- ETFreplay screener helps point to Market Leadership

Mar 23, 2010 in Screener is a platform built in part for finding market leadership ideas to consider.    We believe that there is a lot of 'market generated information' to be interpreted by the relative price performance of well-defined trading baskets (indexed ETFs).  Global asset flows into and out of regions and sectors are important --- and these flows can be understood by seeing how market segments are performing relative to each other.

A quantitative process can help give ideas as to what is going on in the marketplace.   This can be achieved with relatively simple models.   Simple is better than complex.   The more filters you build into your backtesting, the more UNLIKELY it is to work in the future.    
I ran the ETFreplay screener ( ) for 1 year ago today.  You can do this by setting the calendar control in the top right corner to the final date for which you would like to capture.    I set the screener parameters to pick-up short term relative strength.   The idea would be that the strongest ETF's out of the bear market might very well be the new market leadership.   Using ETFreplays backtesting tab -- I entered the top 5 ETF's into the text boxes at equal 20% weights (make sure it sums to 100%).   I then ran the subsequent 12-month performance.   
Investing is not a purely quantitative process.   Thought and insights and forecasts are crucial.   But a quantitative process can help guide you in the right direction and looking in the right places for ideas.  Even if you don't invest or trade ETFs,  it will certainly be helpful for any investor to understand global asset flows into and out of certain market segments (int'l stocks vs us stocks vs commodities vs fixed-income etc)...
Below are the results of the above idea:


Commodity Comments

Mar 23, 2010 in Commodities


Gold vs Stocks (Total Return) during the Bear Market.



Don't expect historical correlations to remain constant during a crisis.



Closer Look at the Rydex Small-Cap Pure Value ETF

Mar 22, 2010

A strong performing ETF this year is the Rydex Small-Cap Pure Value ETF.  Rydex came out with ETFs that track the 'S&P Pure Indexes' a few years ago.   What are these?

Index providers create growth and value indices out of their core index.  For example, Russell takes the Russell 2000 and divides it into 2 groups: growth and value.   They have developed rules for how to do this, such as price to book value and 'expected analyst earnings growth.'   For some companies -- it is obvious which index it qualifies for:  such as a company with a high price/book and high expected growth rate will obviously be a member of the growth index.   But there are many companies that sit more in the middle, companies where Russells construction rules are in a bit a conflict.   Russell in this case takes the weighting and divides it into part growth and part value,  such that if a given company has a 0.40% weight in the Russell 2000, it might have a 0.31% weight in the Russell 2000 growth and the remaining 0.09% goes into the Russell value index.

The S&P 'Pure Indexes' don't do this -- for these indexes, S&P uses only the companies that qualify 100% as value or growth.   This is not magic, it is just a different method of index construction backed by a major index provider (S&P).  ETF analysts should understand who the index provider is for any given ETF and their reputation when reviewing ETFs.  In this case, S&P is a major index provider and not a provider created out of thin air simply in an attempt to lay 3rd party credibility to financial products (ETFs).

Back to the point, what is the net effect of this different construction methodology?  

The Rydex funds become more concentrated with less securities in the index.  This concentration makes them more volatile than others.   You can't say its better or worse, its just a different method.  I ran the Rydex Small Cap Pure Value ETF in the backtest tab of ETFreplay ( ) just to look at the characteristics. 

Given the very large shift in overall market volatilties we have seen, I settled upon looking at volatility since June 30, 2009.  Why?   Because this is when the financial sector (XLF/KBE) seemed to normalize -- overall market volatility has come in a great deal since then and if you run charts back before that date, the net volatility number will be quite skewed.




ETF Segment Review: Total Return vs Price-Only Return

Mar 18, 2010 in Total Return
The history of the stock market is that dividends make up a material component of equity index returns.  The one exception being the decade of the 1990s, where there was massive P/E expansion that dwarfed the impact of dividends.

But there is a more fundamental point, as long as you are going to invest in something like an ETF, which tracks an underlying index, you might as well use accurate data of the underlying index, which is always TOTAL return. The 'price-only' ETF return found on financial charting websites like yahoo and google can be quite different than the underlying index TOTAL return.

Emerging Market Bond Index:

Pharmaceutical Stock Index:

Preferred Stock Index:

Real-Estate Investment Trust Index:

Investment-Grade Bond Index:

Utility Stock Index:


Total Returns (distribution adjusted) vs Price Returns

Mar 09, 2010 in Total Return

On the anniversary of the stock market low, these are trailing 12 month examples of how price charts that don't adjust for dividends & distributions are simply inaccurate for professional analysis. The total return for SPDR Junk Bond ETF (JNK) is over 20 percentage points better than its price return -- due to dividends. The Pharmaceuticals HOLDR (PPH) had 1) dividends and 2) a major payout distribution in 2009. If you are not watching total return, you are not getting the right impression in a chart.



50% Stocks 50% Bonds 10-Year Total Return (including dividends)

Mar 06, 2010

A portfolio that actually sought to reduce risk. Return estimates are an individuals choice -- but it should be pointed out that the long-term return/risk (sharpe ratio) of the S&P 500 has never been attractive. You can do much better, if nothing else by simply lowering the standard deviation through diversification.


Follow ETFreplay