Posted by barkand on January 26, 2013 under Historical data | 
The U.S. equity market has been experiencing a slow melt-up for much of this month. And for the last seven days, the daily low was higher than the one before it.
We wondered if it meant anything.
The answer, in a word, is NO!
We scanned 20 years of data for SPY, the popular exchange-traded fund which tracks the S&P 500, for other situations where there was 7 (or more) consecutive higher lows. What we found was that the market is nearly as likely to go down as go up during the next 5 days. The 31 cases we found were nearly evenly split between short-term gains, short-term losses, and a neutral market. But maybe prices would betray their intentions by telling us where they were relative to the Bollinger bands. Nope. Nothing. On average, prices are between 90% – 100% of the way up from the lower band to the upper band when they reach the 7th higher low regardless of where they go from there.
One bit of value from this exercise is that, no matter what happens during the next week, the market usually continues higher over the next 3 weeks. the average gain during those 15 days comes in right around 1%. This might seem counterintuitive to some, because the market can appear to be “overbought” at times like this.
Posted by barkand on January 23, 2013 under Historical data |
Our proprietary thermostat for the stock market is red-lining now.
We mostly use this tool for identifying short-term bottoms. When our index goes below 10, there is often a rally in the market. The last time this happened was on December 27th, when the S&P 500 closed at 1418. The S&P is up 5.25% in less than a month since that signal.
Spotting peaks in the market is more problematic for most technical analysis tools, and ours is no exception. There are many times when we pay no attention to seemingly “high” readings on our index. But in the current situation, due to the behavior inside our index, we suspect the market is setting up for a short decline. It probably will not last more than a few days, but more active traders might want to lock in some profits now.
Posted by barkand on January 4, 2013 under Historical data |
The first two trading days of 2013 have given us an unusual chart pattern: a very long white candle (or bar, if you prefer bar charts) followed by a candle with a small real body. In candlestick terms, the “real body” is the difference between the high and low. In plain English, it means there was a big up day followed by a day with little movement.
We pondered what this might mean for the immediate future, and so we filtered the S&P 500 data for the last 20 years for these conditions:
- SPX up at least 2% to a 20-day high
- The next day’s open to close, in absolute value, covers less than 20% of the previous day’s gain
- Condition 1 and Condition 2 have not been seen (together) in at least 20 days
In the last 20 years, this has happened exactly 20 times. The market sometimes goes through periods of years without this situation occurring. The historical results are mixed, at best. Half the time, the market continues higher. And that logically means that the market is lower the other half of the time. However, the negative results were often quite significant, with frequent losses of more than 4% over the next 20 days.
In fact it gets worse if we relax Condition 3. If the market has a 2%+ gain to a new high and then a stagnant day, and this is happening for the second time in recent weeks, the average return over the next month is a negative 3.76%! So if we see this chart pattern again this month it will likely be a time to take some profits.
Posted by barkand on January 1, 2013 under Historical data |
Yesterday we wrote about downtrodden Hewlett-Packard being a “dog of the Dow” for 2013. Today we follow that by looking at another bit of market folklore. This is the idea that the first five trading days of January will tell you the direction of the market for the entire year.
Many have written about the value – or lack thereof – in this market aphorism. According to a quick Google search, the esteemed Mark Hulbert annually rails against this “superstition” including just last year and 2011 and 2010 and 2009 for example. You don’t need to click every link; it is nearly the same article every year, wherein he cites statistics for the Dow Jones Industrial Average going back to 1896. We decided to check it out for ourselves. What happened in 1896 or even the 1950s is of little interest to us because the market today bears little resemblance to the market then. On the other hand, using only the most recent years provides too little data to work with. So we settled on reviewing 20 years of data for the S&P 500. You can check out the chart and then we have our comments below.
Interesting, isn’t it?
* It is only 20 data points, but there is a positive correlation between the first week of the year and the entire year.
* Of the 10 best years for the S&P, 9 were positive during the first week.
* Of the 10 worst years, only 4 had a positive start while 6 had a negative return over the first five days.
Looking only at “up” and “down”, perhaps Hulbert is right. A good start is not a guarantee that the year will turn out well. But it is unusual to have an especially good year without one.
Posted by barkand on October 28, 2012 under Historical data |
Not long ago, Quantified Strategies wrote about an interesting strategy: New 5 Day Low And Closes In The Low Of The Range. We decided to poke around with this idea, but our feeble mind couldn’t manage to remember the details of this simple setup.
When we re-created the data, we bought the next day’s open. The original post called for buying the close of the 5-day low. Maybe it doesn’t make a big difference. This post is really about improving on buying the open.
If the market is making a 5-day low and closing at the bottom of the day’s range, there is probably a reasonable amount of bad feelings building up in the market. but there is no reason why there can not be even more angst the following day. It is not uncommon for the market to open even lower. But sometimes it doesn’t. With surprising frequency, the market will open somewhat higher. However, all that angst doesn’t just magically disappear. In these situations, we found that the market nearly always sells off towards or below the previous close before turning higher.
So instead of blindly buying the open in times when the market appears to be “oversold”, it might be better to use a limit order just a bit above the previous close. We tested this on SPY and QQQ, looking at the last two years of data. We placed our limit order at one quarter of the way up between the previous day’s low and high. The results below, which do not take into account any trading costs or slippage, show that the limit order is more profitable.
Posted by barkand on July 23, 2012 under Historical data, Seasonality |
Global stock markets are looking quite bleak today. European markets are getting whacked for losses more than 2% and U.S. futures are only slightly better. But there is a ray of sunshine coming through the clouds.
We checked the numbers for the last 10 years for other times when SPY, the popular ETF of the S&P 500, gapped lower by more than a half-percent on a Monday. By our count, it has happened 71 times.
A strategy of buying the open today and then selling at the close on Friday produces an average gain of 0.98%. Not bad at all. Unfortunately, it makes a big difference where the market is relative to an intermediate-term trendline. When the market is already below its 50-day moving average and then gaps lower on Monday, this strategy gives us a 1.57% return for the week. But in cases where we are above the 50-day average, as we are now, we end up with only 0.38%.
