2/9/2010 Big Gap after 20-day Low Close

This morning ES opened about 10 points higher than yesterday's RTH close. Historically, there is an edge on the long side when market gaps up after weakness. I use the following criteria to find days similar to this morning:

1. Gap bigger than 0.75%.
2. The prior day being lowest close of 20-days.

Since 2002, there have been 30 such instances and 63.33% closed higher for the day with an average of 0.46%. The gap is filled 60% of the time. My generalization is that market tends to re-test the low and fill the gap early in the day and went on close higher later in the day.

When I further add a filter of the previous day's closing price higher than SMA200. The number of instances drops to 2 and both closed positive for the day.

2/4 Unfilled Large Gap with High Volume making 20-day Low

Whew....what a day. ES was already down six point before the Jobs number was released and the worse than expect job number caused ES to trended down pretty much all day long.

Looking at daily SPY historical data from 2002, there are 27 instances that gap down more than 0.7% and not filled on the same day. The next day has an average of -0.04%. However, the average is rather deceiving, 7 out of the next day close higher for more than 1% and 4 out 27 closed down more than -1%. Out of the 4 instances, 10/6/2008 closed down more than -6% and 11/19/2008 closed down more than -5%. The chance of the next day develops into a trend day is definitely higher than normal, and I would be very cautious the next day.

Interestingly, this kind of behavior seems happen more frequent in sever market drop more than anything other period. 12 out of 27 instances took place after June 2008.

2/3/2010 Consolidation After Two Higher Closes

The most salient feature of today's market is that SPY consolidated after two days of higher closes. I did a search on the historical data of SPY since 2000 to see the historical tendency of the day after. By using the rules of 2 higher highs, followed by a small negative drop, and above SMA200, I found 18 instances. 60% closed higher from the open of the next day and with an average of 0.11%. There is no obvious edge on either side, but I will align myself with the long side and will be looking for entering long position at favorable location.

With the market trading 6 point below yesterday's cash close forming a potential gap at 6:32 AM, I looked into the historical data and found that when market gaps down, the historically tendency does even better, an average of 0.23% vs 0.3% when market opens above yesterday.

50-Day High to 50-Day Low and Inside Day

After seeing "From A 50-Day High To A 50-Day Low In 8 Days" post on "Quantifiable Edges", I decided to run my own testing seeing what happens after SPY move from 50-day high to 50-day low in less than 8 days.

Since 2002, I found 4 such instances on 3/10/2004, 5/16/2006, 2/26/2007, and 7/25/2007. Regardless of number of days out, decent size drop occurred after all 4 instances with and average of -1.76%, -1.63%, -2.30%, -1.42% and -1.22% for 1, 3, 5, 10, and 20 days out.

Interesting inside days seems to occur soon after such quick drop, which also took place on 2/1/2010 after a move from 50-day high to 50-day low within 8 days. This finding somewhat give support to my hypothesis that the market is in the consolidation phase. The weakness may continue, but I do not expect severe drop to follow unless there is some sort of catalyst.

A Day of Consolidation

Today (2/1/2010) ends up being another inside day. This is the second inside day within two weeks with the last one being 1/25/2010. Historically when SPY is trade below 50-day moving average but above 200-day moving average, it seems that frequent inside days signals a some sort of bottoming process and high possibility of market retracing the losses within 20 days.

I sound very uncertain because, inside day does not seem to help much on timing the rally.All I can be certain is that imminent farther sever drop does not seem likely under current condition according to historical data.

Another phenomenon I observed is that when the inside day is made on volume lighter than 20-day average, the chance of next day close higher is 60% vs. 33%, and average return of 0.15% vs. -0.07% (n=16 vs. n=15). This could be purely random, but interesting nonetheless. Today's volume happens to less than 20-day average.

So how does this information help intra-day trading? I am more incline to go long at lower extremes rather than fading breakouts, and I will tend to take profit a little bit early when fading breakouts.

Stats on Market Profile Normal Day

According to the definition in Jim Dalton's book "Mind Over Market", the main character of Normal Day is its wide Initial Balance not upset during the day and it is more of an exception rather than normal.

After this morning's somewhat slow trading, I decided to spend some time doing some counting on Normal Day, and here is what I have found:

Since 9/7/2005, there have only been 20 trading days that has the characteristics of Normal Day outside of holiday. It is indeed a exception rather than the norm to have market traded within the IB outiside of holiday.

Another interesting observation I found is that Normal Day is generally caused by lack of participation rather than strong directional conviction. Outside if IB, the volume of each 30-minute period is -36% less than 20-day average of the same 30-minute period. The two 30-minute periods that make up IB has 7% more volume than 20-day average.

From the daily data, I doubt Normal Day has any much significance. I can't find any strategy to trade Normal Day to produce profit better than a coin flip.

Consecutive 20-day Lowest Close

Yesterday close lower than yesterday and extend the streak of 20-day lowest close to two and the streak of consecutive 20-day lowest low to 4.

Let's first see what happen after 2 consecutive days making 20-day lowest close. To find situation similar to today, I specified the following additional condition:

1. Both volume higher than 20-day average.
2. Volatility higher than 20-day average.
3. Above 200-day SMA.

Since 2002, there have been only seven such instances and all have occurred between 2004 and 2007. Within the next five days, all instances moved higher than the open of the next day. Even though the size of sample is really small, I would be very cautios with any farther short postion in next few days.

Outside Day on 20-Day Low

The historical tendency played out nicely today. SPY made 20-day low on expanded volatility forming a outside day. I did a quick check on the historical stats on 20-day low made by outside day.

Since 2002, there have been 19 such instances. 15 instances close higher with an average of 0.70% and 4 instance close lower for average -1.34%.

My interpretation is that market has high tendency to consolidate, but if it does not, weakness ensues.

Stats on Fed Day

Dr. Steenbarger's blog is one of my daily must reads. In the article "How Do Federal Reserve Announcements Affect The Markets?", Dr. Steenbarger wrote that the volatility of the FED day has been exaggerated. The article was written near the end of 2006, I decided to do a quick check to see if the quality has changed.

Since 2004, there have been 48 FOMC announcement. 38 out of 48 has volume higher than the 20-day average. In average, SPY volume on FED day is 12% higher than the 20-day average.

29 out of 48 FED days have high low range greater than 20-day average. The SPY range is roughly 0.33 higher which is slightly more than two S&P points.

The quality Dr. Steenbarger described still holds true three years after writing the article. The most interesting thing I found in my research is that 29 out of 48 FED days close positive for the day. Five days later only 8 out of 29 days remain higher than the next open of the day following the FED day. Conversely, 18 closed lower on FED day and 7 out of 18 remain lower than the open of the next day. The sample is pretty low, but my interpretation is that the market tends to over extend itself on the FED day and also on the open of the following day reversion steeper than usual.

If the scenario plays out, I probably would want to align myself accordingly.

Inside Day followed by Outside Day

What's peculiar about last few days is steep drop followed by an inside day then an outside day in a midst of longer term rally. I did a simple test to see what would happen in days after such condition by specify the following condition:

1. Inside day followed by outside day.
2. Price is lower than five days ago.
3. Higher than 200 day simple moving average.

There have been 19 such instances since 2002. The result is overwhelmingly bearish in the next day but three and more days out, market becomes quiet positive. It would be interesting to see how this pattern plays out with the positive bias going into the FED day.

20 Day Low vs. Close Higher/Lower for the Day

When I first started learning swing trading, one of the fist thing learned is to trade breakout and follow the trend. I learned specifically to short SPY when it makes 20-day low and go long when it makes 20-day high As anyone who has done any trading, nothing was this easy. The market does produce big moves more often than normal distribution, but the likelihood of reversion is also higher than normal distribution.

I next ventured into the world of technical analysis, and learned that I should not be following trend. I should fade trends most of the time. Instead of selling 20-day low and buying 20-day high, I should buy when the market makes 20-day low with reversal bar.

In this post I compared the average of 1, 3, 5, 10, and 20 day out after SPY made 20-day low and breaking the group into days that closed higher and days that closed lower. Since 2002, there are 18 instances that SPY closed higher for the day when it makes 20-day low and 178 instances that SPY closed lower for the day on the day of 20-day low.

The chart shows two important points:

1. In average, SPY performs better when it closed higher for the day when making 20-day low.
2. Regardless how the day closes, SPY has positive expectancy.

However, hiding behind the positive expectancy is that when SPY close higher on the day of 20-day low, 5 and 10 days out about 60% of time SPY actually drops below the low.

Three Consecutive Lower Lows

It's been a while since my last post, I have taken time off blogging to work on compiling historical data and better ways to look for the edge.

Last Friday (1/22/2009) marked the third consecutive days SPY closed lower. Before today's opening, I did a quick check on the historical tendency on the day after three consecutive close. Without any filter, there are 108 instances since 2002 where SPY closed lower three consecutive days. Out 0f 108 instances, 60% of the following day closed higher, but three days out SPY has negative expectancy if going long. My interpretation is that market has high tendency to retrace after three consecutive lower closes, but the bounce is USUALLY not a big rally and market tends to chops and consolidate immediately after three days of consecutive lower close. After 10 days, market tends to retrace all the drops caused by the three days. However, when things goes wrong the floor just drops off. The most recent incidents are 12/19/2008 and 2/2/2009. SPY moved around -9% and -15% respectively 20 days out.

Adding a filter of the last drop being more severe than -1.5%, the picture changes completely. The next day closed lower 56% of the instances with average of -0.36%. The historically edge lies with the short side producing an average of -0.36% and -0.19% for longs. Market also tends to bounce back very strong after such severe drop 5 days out with peak of 2.93% 19 days out.