I was on vacation last week with limited access to the internet. While subscribers still got the nightly Subscriber Letter, I was unable to produce blogs. I’m back now and will have some edges to share this week. This is one study I found interesting from Sunday night’s Subscriber Letter.

(click to enlarge)

Pretty much everything I’m looking at short-term is suggesting a pullback. Longer-term that is not the case.

I did a study looking at situations where historical volatility has contracted rapidly like it has in the last 3 days. For the study I looked at the 3-day historical volatility and compared it to the 10-day historical volatility 3 days ago. In other words I divided Friday’s 3-day historical volatility by Tuesday’s 10-day historical volatility. A result below 1 would indicate the last 3 days have been less volatile than the previous 10. A number above 1 would indicate a recent uptick in volatility. For the SPX, Friday’s 3-day over Tuesday’s 10-day came in at a very low 0.23.

I then looked to see what the 3-day historical volatility has typically been 3 days later. What does a sharp contraction over the last 3 days indicate you might expect over the next 3 days? What I found is that when the 3-day over the offset 10-day dropped to 0.25 or lower the next 3 days were 5.5 times as volatile as the recent 3 days. This is based on 1,111 trading days since 1960 – or about 9% of all trading days. Going back to just 1999 gave similar results, as the 3-day historical volatility increased by 5.4 times.

This suggests we may see some increased movement over the next few days. A breakout of the recent 3-day range could be very sharp.

Thursday’s action was extreme in that the gap down was large and the SPY opened at its high and closed at its low. Often reactions this strong and persistent are overreactions. Overwhelming negativity such as this has led to quick, sharp, bounces in the past. Below is a study that exemplifies this:

The Stock Traders Almanac noted that the Dow Has been up 16 of the last 19 years on the 1st trading day in July. It’s been a while since I last looked at 1st day of the month tendencies.

From 1960 until the late 80’s there was no decided edge on the 1st trading day of the month. In the late 80’s this changed and the 1st day of the month became an outperformer. The most often cited theory as to why is that defined contribution retirement plans like 401k’s became more popular. This meant there was a lot of money going into funds right at the beginning of the month, and funds managers were putting it to work.

Below is a performance report showing all the 1st trading days of the month from 1/1/1987 through June, 2009.

The average performance on the 1st day of the month is over 0.2%, which is substantially stronger than the remaining days.

The 1st day of month strength has not been immune to bear markets, though. Below is an equity curve. As you can see, there were sizable dips in the strategy in 2002 and 2008/09.

Lastly I broke out the performance by month. You’ll note July has been the most reliable month over the last 22 years. It also ranks 3rd based on average profits. August has had the worst performance.

Friday night the Quantifinder identified the below study. I haven’t updated it in over a year so I thought it would be interesting to do so.

Results are much the same as they were a year ago. Low QQQQ volume in an uptrend has often signalled at least a short-term drying up of buying interest. Historically this has created negative expectations over the next several days and weeks.

Much of the net negative thrust has often been found in the 1st two days following the setup.

Originally published in the blog on 9/11/2008, the Quantifinder noted the below breadth-based study for Tuesday’s close. I re-ran it in last night’s Subscriber Letter and have updated it below. (Note: Up Volume % is tracked in the members chart area each night. It is equal to (Up volume / (Up volume + down volume).

click table to enlarge

The fact that there was a pullback within three days every time is a fairly compelling stat. There have been 2 upmoves occur in the 6-day holding period since I published the first set of results last September. They both had pullbacks first though, allowing either opportunity for profit or at least a chance to move your stop to breakeven. Below I’ve listed all the trades along with their 6-day exit:

Note even the 2 most recent losers saw some sizable pullbacks. The 3/6/09 instance which came very close to the bottom was the worst of the bunch from a short perspective but still moved south by over 1.5% before launching higher.

One relationship I’ve been watching lately is the S&P vs. the US dollar. To illustrate the relationship I’ve created a chart below. In red is the closing prices of the S&P 500. The blue is closing prices of UDN, which is the US Dollar Bearish Fund ETF.

As you can see they’ve pretty much moved in concert over the last year-plus. Since UDN is a bearish fund this means that the S&P has actually moved opposite the dollar. A weak dollar has been cause for celebration and a strong dollar has closely preceded most of the drops in the S&P.

But can the performance in the dollar provide a trading edge in the S&P? Below is the equity curve of an incredibly simple system. If UDN outperforms SPX on the day, go long SPX at the close. If UDN underperforms, go short SPX at the close. Basically you’re either 100% long or 100% short depending on how the dollar (inverse) performed that day relative to the SPX. For purposes of the test, no commissions or slippage are included.

That’s more than an 82% non-compounded gain over the last 18 months. This edge won’t last forever. Still, this should demonstrate that the correlation between the dollar and the S&P is important. Moves in the dollar have a definite impact on the S&P. S&P traders would be well served to monitor the dollar’s performance closely and on an ongoing basis.

Also notable about Wednesday’s action was that while the market has declined for at least 3 days in a row, the rate of decline has lessened both of the last 2 days. Back in September I found this pattern to have bullish consequences when the market was trading below its 200-day moving average. I decided to test it out when above the 200ma as well. I found that prior to 1987 there was no bullish influence based on the pattern. Since 1987 one has appeared. Here is the test using the SPY. First the base case without the slowing rate of decline requirement.

(click table to enlarge)

Closing down AT LEAST (rather than exactly) 3 days in a row has been bullish over the period. Now let’s add the slowing rate of decline requirement:

(click table to enlarge)

Results here are quite a bit stronger than the base case.

I noted in a tweet last night that big moves down are more short-term bullish when they aren’t occurring from a high. This was a bit of an understatement. They’re often short-term bearish. Yesterday’s drop moved the S&P 500 from a 10-day closing high to a 10-day closing low. Since 1960 there have been only 10 other times this has happened. In last night’s Subscriber Letter I show the results of those 10 instances.

To get a larger sample size I also reduced the requirements from a 10-day high and low to a 7-day high and low. Those results are below. (Click table to enlarge.)

These results suggest more downside, especially over the next 2-4 days.

For more discussion on big drops from highs you may want to review this former study.

P.S. I just noticed that Dr. Brett also looked at the 10-day high to 10-day low move this morning. See what he has to say here.