It seems that the big traders have the advantage. They can capture a lot of small profits with high-frequency trading, pairs trading, merger arbitrage or countless bond arbitrage trades based on the yield curve. But there are some opportunities still there for low-frequency, less intense traders.
One game still played is taking advantage of the way Goldman Sachs rolls the S&P Goldman Sachs Commodity Index (GSCI). Because it’s a public index they publish exactly when they roll from one futures delivery month to another (the fifth to ninth business day of each month). Because their volume is so high and their position is always long, they push prices down when they close out the old futures positions and up when they reenter the new contract. By anticipating their move, a trader can earn a few percentage points each year while only being in the market a few days. The process is taking advantage of congestion yield. One caveat is that the size of the position from people playing the congestion yield can grow larger than the roll itself. It can provide an edge but is no longer a layup.
There is another, more frequent event, also based on congestion yield that can be exploited: The monthly deposits and additions to funds based on automatic retirement account flows. The amount of money just from Federal and state retirement programs is estimated to be about $12 trillion (with $2.7 trillion allocated to equities), which leaves a large amount in bonds and other interest rate vehicles. Not only is that a lot of money, but the monthly flow is enough to move prices.
We know for certain that your IRA manager has an obligation to add deposits to your current holdings and divest if you are redeeming funds. Those trades need to be done sometime during the month and we expect that every firm has a strict policy of how and when this is done. Without a strict set of rules it can cause customer problems if they miss getting in before a rally or out before a drop in prices.
All customers need to be treated the same way, and using discretion to enter and exit funds is not going to satisfy anyone.
If a firm has an equal, offsetting number of deposits and redemptions they’re not going to have to buy or sell anything. They can do an internal switch. But it’s not likely that these transactions will balance out each month, and given the enormous number of firms doing the same thing, and the Law of Large Numbers, we should see a noticeable effect on price movement.
Price patterns within the month
We can uncover these patterns by accumulating the returns of the bond market and averaging them by the day of the month. For our purposes, we’ll ignore specific dates and simply reference the first trading day, the second trading day and so forth, making it easy to combine and average the results. Weekends and holidays won’t be a problem using this approach.
We can do the same with the end of the month. Working backwards, we can align the last trading day of each month, the next to last trading day, the third from last, and so on. If there is a consistent pattern, we should see it.
Let’s look first at the 10-year bond index, TNX and also the U.S. 10-year Note futures. TNX is expressed as a yield and futures are quoted as a price. When yields drop, TNX declines but futures rise. We don’t know if TNX is actively used in retirement accounts, but interest rate vehicles are actively arbitraged, so we would expect to see the same patterns in all rates of about the same maturity.
We’ll focus on futures because they are very liquid, inexpensive to trade and good for hedging. What stands out in both sides of “Notes vs. notes, (below) It’s the clustering of positive returns on the far right, the last three days of each month. It tells me that fund managers are buying bonds over those three days. Not necessarily one manager buying on all three days, but collectively a large number of managers buy during that interval.
If we look closer we see that the gains over those three days have declined significantly during the past 15 years. “Playing the bond edge,” (below) shows the monthly and annualized gains for the three days during the 10-year periods 1990-1999, 2000-2009, and the most recent five years. While this strategy would have returned 7% per year during the 1990s, it’s now down to 2.7%. That may be the result of others gaming the system, or it may be that there is less buying because investors are not choosing to invest in bonds as interest rates hit at all-time lows. We might see a flurry of activity and better returns once interest rates start moving higher.
So, where are investors putting their money if not in bonds? Naturally, the stock market. If we’re right, the decline in bond returns should be offset by an increase in equity index returns. The daily returns in “S&P tendencies” (below) confirm our premise, except that the pattern shows that buying occurs in the first three days of the month, rather than the end of the previous month. Could that be investors managing their own equity positions or simply fund managers spreading out their orders?
Note that the last day of the month tends to show a decline, so buying on close of the last day and holding for the first three days of the next month would give the payout shown in “Playing the S&P edge,” (below).
Both charts above show similar returns, but futures include the 1990s as well.
We can see that when there was a bull market in stocks, which was both in the 1990s and the past five years, returns from gaming the S&P index were more than 5%, but during 2000 to 2009, mostly a declining and uncertain decade, returns were positive but minimal.
Trading calendar patterns
Taking advantage of the beginning and end-of-month patterns is not difficult. You can use either futures or exchange-traded funds, but with futures there is the advantage of leverage—and also the risk. Remember that these patterns show the average return during many months and that some months don’t follow the pattern.
For 10-year rates, buy on the close four to six trading days before the end of the month. Exit on the close of the last day of the month.
For the S&Ps, buy on the close of the last day of the month (hopefully a lower day), and exit the trade on the close of the third trading day of the month.
This trading game is not unique. We already watch the Election Year Cycle, the Thanksgiving to Christmas period, the Sell in May and Go Away cycle, as well as the biggest of all: Seasonality. Except for seasonality, these cycles are all based on investor behavior in large numbers that seems to be as consistent as Mother Nature.
Good traders find edges and exploit tendencies. These patterns have proven to be relatively consistent and can help provide an edge when used with other technical and fundamental tools. They are not laws of nature and are not foolproof. If they worked all the time no one would lose. When traders believe they have found the Holy Grail, they tend to lose heavily. There is no Holy Grail. Hundreds of wheat locals blew up in 2006 when they assumed the Goldman roll was simply a way to print money. It was not, as an Australian drought turned the bear spread strategy on its head. Use these tendencies to find an edge but don’t count on them working all the time.