After August 18, 2015 Portfolio Workstation 1.1.3 and earlier will no longer be able to pull down FRED and Quandl data due to their use of HTTP for data requests. Portfolio Workstation 1.1.4 supports HTTPS transfers and works with both data providers. Please upgrade accordingly.
- Updated URLs for Quandl and FRED connectors to use HTTPS instead of HTTP.
- Updated ratio chart renderings in the components tab to render the ratio in addition to the components.
- Added support for FRED overlay lines on line charts. Mutual funds such as VUSTX as line charts rather than OHLC. Previously only OHLC charts supported this feature.
“Inflation” is an interesting, yet confusing topic. We interpret inflation to mean inflation of the money supply; ie. the rate at which a currency is being debased. However, “inflation” in finance pop culture has come to mean “price-inflation”, a slightly different concept. While “inflation” continues to grow the monetary base slowly and predictably, its evil twin, “price-inflation”, is the dominant factor for investors in the short run. Let’s shed some light on ways to measure price-inflation that make some sense.
Luckily, the US Treasury issues both fixed income treasury bonds (TLT etf) and price-inflation linked bonds called Treasury Inflation Protected Securities (TIP etf). TIPS are price-inflation sensitive bonds that adjust upwards in periods of price inflation, whereas normal fixed income bonds suffer in periods of inflation because the fixed size coupons a bond holder receives buy less goods or services.
The following chart is a ratio chart comparing the performance of TIPS in terms of long term Treasury Bonds: TIP/TLT. Additionally, the green line shows the amount of treasuries on the balance sheet of the Federal Reserve, with the purpose of illustrating the three QE programs and their impact on price inflation. On the hard right edge of this chart you can see that TIPS implied price inflation is as depressed as it was at the height of the 2008 financial crisis!
Next up is TIP/LQD, with the TED spread overlayed in green. The TED spread is a somewhat outdated measure of stress in the banking system. Consider this quote from wikipedia: “The TED spread is an indicator of perceived credit risk in the general economy, since T-bills are considered risk-free while LIBOR reflects the credit risk of lending to commercial banks.” These days the federal reserve has eliminated most of the inter-bank lending risk in the form of interest on excess reserves (IOER) replacing T-bills, and as expected, the investment grade corporate bond market (LQD) reflects it, as TIP/LQD hits new lows.
We would suggest that these two charts taken together imply that there is currently extreme price deflation, yet no financial stress in the banking system. This makes sense when you take into account what 0.25% interest on excess reserves (IOER) means for the big banking institutions. We will cover more on the important topic of IOER in the future.
Natural gas is one of the most fascinating commodities. It isn’t just the price that’s volatile. Natural gas itself is unstable, dangerous, difficult to transport, and expensive to store. Natural gas is uncorrelated to almost everything else, and also exhibits seasonal tendencies. If this sounds like the land of opportunity for a trader, you’re right. It’s also the siren’s call to the rocks for the unwary. As one famous example, in 2006 Brian Hunter misjudged the market, losing over $6 billion on a calendar spread gone sour and collapsed his firm, Amaranth Advisors. Their assets were taken over mostly by Citadel and J.P. Morgan. The timing of Amaranth’s losses precedes the creation of the UNG ETN by about 6 months. To what extent they are related events, we can’t say. In any case, here are charts of the natural gas futures price, and UNG since its inception:
UNG is the big daddy of natural gas ETN’s. UNG began trading on April 18, 2007. Over that time, a lot has happened with the price of natural gas, both up and down, but the first thing most people notice with UNG is that it has spent most of it’s life grinding lower. The short explanation is the contract rolling cost, known as contango. The contango in this case happens when selling out a long position in this month’s futures contract before delivery doesn’t leave you with enough money to buy next month’s contract, so you wind up with fewer of next month’s contracts. One of the big factors causing contango is that carrying a long position in futures is effectively paying “the market” to store your commodity for you, and as we already explained above, natural gas is costly to store and transport.
A popular carry trade is shorting natural gas and collecting the “roll yield.” A paper short position in natural gas futures is a synthetic version of the gas producer who stores the commodity (in the ground or in tanks) until delivery when the contract expires, and gets paid a storage premium for that time. This might sound simple, but the gas producer doesn’t know if the price will go up more than the storage premium, and the further out they try to get paid for, the more opportunity they could have forgone, since their price is locked in when they sold the futures contract.
If shorting natural gas is your thing, then VelocityShares has introduced a product for you. On February 8, 2012, DGAZ debuted. This product attempts to replicate a triple leveraged short position in natural gas. As we’ll show, they do a decent job at this goal. The following chart shows DGAZ since its inception:
But how can you know what the ride might be like in this vehicle, since it is so new? Shorting natural gas looked very profitable when we looked at UNG, and if DGAZ does what they claim (it basically does), then what gives? We used Portfolio Workstation and a few other algorithms to answer this question by using UNG’s prices to create a hypothetical simulation of how DGAZ might have traded if it had instead debuted in 2007 alongside UNG. Without further ado, here is the picture:
In order to double-check our work, and also double-check how closely DGAZ tracked the movements of UNG, we created a ratio chart, with our DGAZ price simulation based on UNG in the numerator, and the real DGAZ in the denominator as a benchmark. Here is that ratio chart:
This ratio chart shows a fluctuation around the 1:1 ratio, with the implied DGAZ price (numerator) differing from the actual (denominator) by up to 12%. These differences could indicate temporary arbitrage opportunities. We note that it could differ by more than 12% in the future. The big ratio spike up and down in January and February of 2014 marked a major turning point in natural gas. We’ll be watching for this to mark price inflection points in the future.
As an interesting note, on the date of inception for the real DGAZ, February 8, 2012, the simulated DGAZ open price matches the real open price exactly at $51.70. This is purely coincidence because if we had performed this study on a different date, we would have had to use slightly different numbers for the transaction cost estimate and the nav adjustment algorithm. This phenomena also appears in the ratio chart, it starts and ends at exactly 1, even though we only targeted for it to end at exactly 1.
If you are interested in purchasing the simulated DGAZ daily ohlc data for informational and testing purposes, look for it on our Purchase page.
In yesterday’s post, What Falling Interest Rates Mean For Portfolio Investors, we asserted, “The current generation of investors are accustomed to stock panics driving money into bond funds, but if this dynamic were to flip around, and panics in low yielding long term bond funds caused a rebalancing feedback driving stock prices lower, it would be a new era of rising interest rates.” We thought it would be appropriate to back up this assertion with the data, so we used the “Start Date” and “End Date” chart controls in Portfolio Workstation to pull the returns for VFINX and VUSTX, and how correlated their close prices were for every year going back to 1987.
VFINX is the Vanguard S&P 500 mutual fund and VUSTX is the Vanguard long term treasury mutual fund. These were chosen since the inception dates are far older than the ETFs we might normally look at, SPY and TLT, and include most of the current long term bull market for interest rates.
The data in the following table does support our assertion. In every case where there is a substantial negative year for either stocks or bonds, they will show a sharply negative correlation, but most of the time they are correlated because they are both usually moving higher. This behavior has become so ingrained in investors that we feel compelled to call out that the long grind lower in rates is a key driver for this behavior.
One factor we didn’t touch on in yesterday’s post is that US Treasury interest rates are an important input for how stocks are valued. Consider why a CFO cares about the price of shares that have already been sold to the public. Most companies are carrying a large debt load that they roll over in the corporate bond market. A company’s market cap is a buffer for lenders who are looking at the chance of bankruptcy over the lifetime of their loan. This means that the market cap is sort of like a credit score for companies, where higher stock market valuations translate to lower interest rates demanded by lenders, thus lower operating costs for the company, and more profits for investors.
This feedback loop won’t always operate favorably for investors, but it has since 1987. What we were calling out to watch for was a year going forward where we see negative returns for both stocks and bonds (and a positive correlation during that time). For us, this would signal a potential secular market change from falling rates to rising rates. We don’t mean to imply such a market change is imminent, just that buying long term bond funds at 2% as part of a portfolio strategy is a completely different proposition then buying them at 10%.
* Daily close prices over the year were used for correlation
** 2015 numbers are only through 4-6-2015
The long grind lower in US Treasury interest rates continues in 2015. In this post we will have a look at where the interest rate market has been and what the implications might be going forward when incorporating a fixed income allocation to a portfolio.
Portfolio Workstation has a data connector for FRED that allows us to overlay data published by the St. Louis Federal Reserve Economic Data (FRED). We will use this feature to overlay interest rates on mutual fund performance charts to help with this study. Adding the overly is as simple as checking the desired box, as pictured here:
The following image shows VUSTX, the Vanguard Long Term Treasury Inv Fund with the 10 year yield overlayed. This mutual fund has been around since the mid 1980’s, so it makes a great illustration of the gains that could have been had by investing in treasuries over a long period of time. The green line is the VUSTX performance line, with the split and dividend adjusted price on the right side. The blue line is the 10 year US Treasury interest rate as published by St. Louis Federal Reserve Economic Data (FRED).
This picture does a good job illustrating the long bull market in treasuries, although it doesn’t quite go all the way back to the beginning when long rates peaked at about 15% during the Volcker era. Interestingly, the largest drawdown for VUSTX holders since inception was 18.4%, and the compound annual growth rate (CAGR) was 8.3%. If that CAGR seems higher than where it should be given the average level of interest rates during that period, you’re thinking clearly. The average interest rate was approximately 5.5%.
Under the surface, there’s an awful lot going on here. One factor is that VUSTX carries treasuries at a variety of maturities out of necessity. Maintaining a target maturity duration as a lender is a aiming for a moving target. Over the course of this study, there had to have been a 100% turnover in the holdings of VUSTX, otherwise what used to be a long term fund would now be a short term fund, or more likely sitting on the cash from paid off loans. Because of pro-active rebalancing to maintain it’s long term charter, VUSTX very likely had an average maturity longer than 10 years over the course of this study, and by implication a slightly higher average interest rate than 5.5%.
Another important factor is that bonds purchased in a higher yield environment have a greater market value in the lower yield environment, which is what we see playing out over and over as rates grind lower. An exaggerated and oversimplified example should convey this concept so that we can move past it. Pretend you purchased for $100,000 a fixed income note yielding 10%, and the market for those notes moves to 5% yield the very next day. Your note is still kicking out the same fixed income coupons as before, so how is it a 5% note now? Because the market is bidding your note at $200,000 now ($10,000 of annual interest is 5% of $200,000). You can sell the note into the market and realize a huge capital gain, and by doing so you will have pulled future income to the present.
One of the purposes of this post is to call it out that there is a limit to this quality of the interest rate bull market. As rates go lower, there is less future profits to pull forward to the present. A zero yield bond is a loan with no risk premium. It is literally an IOU that only pays back principle, and as long as nothing goes wrong with the borrower. If bond yields get to zero, there is no profit incentive to be in the lending business.
Going back to the 8.3% CAGR that our model bond fund has, we can see that the lower nominal rates go, the more dependent on the “pull profits forward” effect it’s performance becomes, and at the same time, there is less future profits available for pulling forward because there is less profit from interest at lower yields.
This brings us to the 60% stocks, 40% bonds portfolio. This common portfolio allocation strategy relies quite a bit on bonds, and projecting the past into the future without regard to the impact of the changing nominal yield enviroment would be a setup for disappointment. Having said all that, let’s look at a long view of the past for the 60 / 40 portfolio using Vanguard’s S&P 500 fund VFINX at 60% and VUSTX at 40% and the 10 year yield overlay.
One final observation here is the rebalancing effect on stocks from bond funds in a falling yield environment. With so many investors maintaining fixed allocations between stocks and bonds, great performance in bonds has been a driver for higher stock prices. It is unlikely that the future will resemble the past in the bond market in the coming decades. The current generation of investors are accustomed to stock panics driving money into bond funds, but if this dynamic were to flip around, and panics in low yielding long term bond funds caused a rebalancing feedback driving stock prices lower, it would be a new era of rising interest rates.
In a recent post, Simulating an Ideal 130 – 30 Portfolio we looked at what a skilled stock picker working within the Nasdaq 100 might be able to achieve. We used Portfolio Workstation to model a hypothetical portfolio for this stock picker and compared it against a benchmark strategy of buy and hold QQQ. The main idea was that the stock picker would allocate 90% to long picks, negative 30% to short picks, and 40% to treasuries, meeting the criteria for both the 130 – 30 hedging structure, and the 60 / 40 net allocation to stocks and bonds.
In this post, we will isolate the impact of the short stock picks by substituting the 90% allocation to long stock picks with a 90% allocation to QQQ in the 130 – 30 portfolio from before. As a reminder, the short picks used in this portfolio were selected with the benefit of hindsight by assuming that a skilled stock picker would be able to pick short positions that would have returns similar to the top of the fourth quartile of performance over the period under study, January 1, 2007 to present. We sorted the Nasdaq 100 and chose the stocks at positions 76 – 80. Reality would be quite different, since nobody would probably sit in the same short positions for 8 years, and the Nasdaq 100 isn’t the best index to be picking shorts in. By looking past these oversimplifications, we can develop an intuition for how different portfolio building strategies might play out in the future if carefully implemented.
Here is the new 130 – 30 portfolio, with the only stock picks appearing on the short side:
These results are considerably worse than when our stock picker was also skillfully choosing long positions. It makes sense though, because this portfolio implies that any stock picks would be average, achieving index-like returns.
This time, our benchmark will be a 60 / 40 portfolio composed of QQQ and TLT. Here is the chart and stats for the benchmark:
Your eyes do not deceive you! It is almost an exact match. One takeaway is that maintaining short positions over long periods of time is difficult. The short position picks used here are from the bottom quartile for returns over a period of about 8 years. Over these 8 years, the worst stocks have already dropped out of the index and wouldn’t have been identified by our sort (a hindsight bias). A successful short stock picker can’t just sit short in mediocre stocks and outperform, they will need to have at least some short positions in stocks that are falling out of indexes, and without getting caught in short squeezes or takeovers.
At the wikipedia page for the Nasdaq 100, we counted 72 companies that were replaced in the Nasdaq 100 since 2008, for various reasons, but not all of them bad. This high turnover implies a fertile ground for stock pickers.
In conclusion, we find it encouraging that given the narrow parameters of this study, it seems clear that a highly skilled shorting expert could produce much better results than the compelling hypothetical scenario we’ve shown here, and with no effort given to picking long positions if using index ETF’s as a substitute.
Hedge funds and stock pickers love the 130 – 30 portfolio structure. Oliver Stone’s Wallstreet sequel “Money Never Sleeps” even features it in the big comeback scene. We get a glimpse of Gordon Gekko’s computer screen as he parlays his hidden $100 million to up over $1 billion in assets under management. “My guys are the best,” he proudly touts, and we see the classic hedge fund asset allocation on his screen: 130% long, and 30% short, netting to 100% long over $1 billion AUM.
We used Portfolio Workstation to simulate how the 130 – 30 portfolio might look like for a skilled stock picker working with the Nasdaq 100 (pictured upper right). In order to do this, we made the assumption that whomever picked the stocks in the portfolio was capable of picking long positions from the top of the second quartile of returns, and short positions from the top of the bottom quartile over the period under study. The time period is January 1, 2007 to present, as shown in the Chart Controls windows in the included screenshots. To be very clear, we are exploiting the benefit of hind-sight to identify what those stocks are for this study with the intent to illustrate what a skilled stock picker might have been able to realistically achieve.
Sorting the Nasdaq 100 constituents by performance for the period under study and selecting a small sample of stock picks as outlined leads to the following:
Long positions: SBAC, CELG, SIAL, CHPK, DISCA
Short positions: ADBE, PCAR, YHOO, PAYX, MSFT
The 130 – 30 portfolio allocations from these picks is illustrated in the following screenshot. The positions are fixed fractional allocations and are maintained by rebalancing daily at both the open and the close with a cost estimate of 0.25% slippage per trade. Note that the long stock picks add up to 90% allocation, and there is a 40% allocation to the treasury bond fund TLT. The short position allocations add up to negative 30%. This sums to 100% as follows: 90 + 40 – 30. Even though we are doing lots of stock picking on both the long and the short side, we are still preserving the basic tenet 60% stocks (net) and 40% bonds.
Let’s have a look at how the hypothetical portfolio stacked up to our benchmark, QQQ over the same period. All of these stats shown are computed based on daily time series, so only compare them to each other. For example, comparing these sharpe ratios to sharpe ratios calculated elsewhere based on monthly returns won’t work.
We find the most interesting difference to be the Sortino Ratio, which more than doubled the benchmark. The Total Return was almost twice as good as the benchmark. The 2008 financial crisis was the largest drawdown for both, but the hedged portfolio fared much better with only a 27.5% drawdown, vs 54.5% for the benchmark. Looking at the ratio chart with our portfolio in the numerator and QQQ in the denominator, it becomes clear that the portfolio gained most of it’s ground over the benchmark during the financial crisis (pictured next).
The next image is the correlation matrix for all of the constituents of the portfolio, sorted starting with the least correlated to the entire portfolio (YHOO 0.588) towards the most correlated (SBAC 0.951). We note that the 5 least correlated are all the short picks. TLT almost appears to separate the longs and the shorts, but CELG got in the way of that clear divide.
For completeness, the charts and stats for QQQ and TLT during the period under study are included below.
- Suppressed excessive chart renderings on application startup.
- Changed label on the summary report from ‘Maximum Drawdown’ to ‘Largest Drawdown’.
- Renamed the default workbook from ‘Default’ to ‘Default Workbook’ so it is more clear that it’s a workbook.
- Reorganized the Workbooks tab:
- Moved the ‘Import’ and ‘Export’ buttons from the middle to the top of the tab and gave them icons.
- Moved ‘Move Up’ and ‘Move Down’ buttons to the right-click context menu on the Workbooks tab.
- Moved delete button to the right-click context menu on the Workbooks tab.
- Dow 30 workbook includes the recent removal of T and addition of AAPL.
- “About” tab on Mac OS X version now works correctly.
“Portfolio margin” is where brokerages will agree to evaluate the risk in a customer’s account by factoring together the offsetting directional risks of long and short positions. Under portfolio margin, by using spread trades a customer can take on far larger size and use more margin than they otherwise could if each of the various positions in a client’s account were evaluated for risk in a vacuum. Well constructed spread trades can increase consistency and decrease risk. This is the reason we want to develop an intuition for spread trade mechanics.
Let’s take a look at a very simple example: long QQQ and short SPY. The first chart shows that QQQ has outperformed SPY in the recent past, and you can see a spread opening up between the performance line for each. The second image shows the ratio chart for QQQ/SPY. The ratio chart is calculated by taking the QQQ share price and dividing by the SPY price for each point on the chart. We can see the ratio has risen from an average of around 0.35 to 0.5, a 43% increase.
At first blush, it would seem that we could capture this entire 43% move shown in the ratio chart simply by buying QQQ and shorting SPY. Unfortunately, it doesn’t work out that way. Let’s take a look at what actually will happen. The following chart is a rendering of a hypothetical portfolio that keeps QQQ at 100% of the value of the account, and SPY short at -100% of the value of the account. No transaction cost estimates are used.
This hypothetical portfolio returned only 36%, not the 43% we were hoping for. The good news is the largest drawdown was only 17% – one benefit we were hoping to see. QQQ and SPY each had a drawdown of 55% over the same time period, although they delivered triple digit returns.
The reason for the difference deserves some explanation. Consider this simple illustration of a hypothetical +10% day in both SPY and QQQ.
Starting account balance: $1,000
QQQ holding: $1,000 -> $1,100
SPY holding: -$1,000 -> -$1,100
At close, sell $100 QQQ, buy $100 SPY
QQQ holding: $1,000
SPY holding: $1,000
Ending account balance: $1,000
Our hypothetical account had a flat day, but the volatility forced some trades to happen to rebalance the account so that we could maintain the 100% long QQQ and -100% SPY positions. If you don’t make these rebalance trades periodically, you could face margin calls, or else not profit as much as you should for the position you intended. The chart above takes these rebalance trades every day at the open and again at the close. Now let’s introduce an estimate for transaction costs. Say we pay 0.25% of the value of every dollar amount traded. The following chart shows this new, much more realistic estimate of what we could expect.
Now we’re all the way down to 16% return, and this doesn’t even include the cost to borrow the SPY shares we are shorting, or the margin interest we would have to pay to the broker! Both of these fees will vary day to day, and are different across brokers, so we won’t dive into those things here, but there is something going on here worth calling out. The big reason for all of the trading expenses is the short position. Consider when it moves favorably. The value of our borrowed shares drops and we need to short more to maintain the -100% allocation. It is worse when it moves unfavorably! The value of the borrowed shares increases and we have to reduce the size of the position to maintain the -100% allocation (otherwise potentially face a margin call).
Finally, let’s render a hypothetical portfolio with the same 0.25% transaction cost estimate, but let’s concede to only rebalance on the first trading day of each week. As long as no margin call happens, the return pumps back up to 30%, with a largest drawdown of 18%. Not a very good return for ten years time and all the work to do the rebalancing, but now we have an intuitive understanding of the mechanics of a simple spread trade.