It appears there has been a bit of a tendency for bad years to finish on a good note.
Wednesday, December 31, 2008
It appears there has been a bit of a tendency for bad years to finish on a good note.
Tuesday, December 30, 2008
“Could 2 period RSI be used to produce better dollar cost averaging returns in a 401k?Instead of doing 401K contributions on 15th and 30th of every month, what do you think of having the contributions go into a money market account, and waiting for the next RSI<20"
I ran a test from 1/1/98 to sometime last week. If someone placed $100 into their 401k twice a month for the last 11 years the total invested would be $26,400. (I used the 1st day of the month and the 1st trading day after the 14th for simplicity.)
If their returns matched the S&P 500 that $26,400 would now be worth $19,748.34.
If instead of investing the $100 on or about the 1st and 15th, the person decided to enter on pullbacks to where the 2-period RSI was below 20, the $26,400 would now be worth $19,777.66.
In other words 11 years worth of effort would have made them an extra $30.
The issue is they are only making $100 trades the entire time. $30 isn’t great but it isn’t terrible when you’re only trading with $100.
The lesson here is that trying to time the entry of your money into your 401k is a waste of time. If you are going to boost your returns you need to focus on trading the account.
Monday, December 29, 2008
Taking a broader view of this indicator I broke it down by times the 10-day exceeded the 100-day average absolute gap and times the 100-day exceeded the 10-day. When the 10-day has exceeded the 100-day (signifying the market has been subject to gap more than usual as of late) the average 1-day return for the SPY was 0.027% . When the SPY was gapping less than usual (the 10-day was less than the 100-day) the average 1-day SPY return was 0.018%. In other words the market has performed 50% better following periods when is has been more gappy than average vs. times when it’s been less gappy than average.
Tuesday, December 23, 2008
The next 1 to 5 days have been especially bullish. If you decided to buy the close 3 sessions before Christmas and then sell the 1st profitable close after entry then 18 of 21 trades would have been winners within 2 days and 20 of 21 within 5 days.
Monday, December 22, 2008
Below is a copy of Friday’s chart. The -18.66 reading is extremely low.
Thursday, December 18, 2008
Also interesting is that the worst 6 weeks was down less than 11%. This suggests a trading range may be more likely than a runaway move up or down.
To baseline the results a little bit I also looked at 50ma crosses when the S&P hadn’t spent at least 50 days below the average:
Wednesday, December 17, 2008
As I did with the 200/50 quadrant tests, I first broke the performance down by the number of points gained or lost from 1960 – present. (Click on any table or chart to enlarge.)
At first glance the numbers should look surprising. The most points gained have come following days where the market closed below both its 50 and 10-day moving average. Don’t bother thinking too hard about these numbers. They lie. The reason they are able to lie is that the results have changed dramatically from the 60’s and 70’s to the last 10 years (when the level of the S&P has been much higher). To illustrate this I will break down the performance using a set $100k/trade rather than S&P points. Below are equity curves broken down by quadrant as I did for the 200/50 a couple of weeks ago:
First let’s look at performance above both the 50 and 10-day moving average:
This chart is very similar to the same quadrant when looking at the 200/50 test. Good follow through was seen when the market was near its highs up until about 1988. From 1988-2000 the edge was weaker. Post 2000 is has been non-existent. Even the latest bull market from 2003-2007 failed to make much headway when trading above these two lines. I noted a few weeks ago that bull market saw little in the way of enthusiasm near highs and this is more proof of that.
The next chart shows performance after the S&P closes above the 50 but below the 10-day moving average.
During the 60’s and 70’s this was not a place to be buying. Since the late 80’s this quadrant has generated some nice returns. The shape here is again somewhat reminiscent of the 200/50 chart of the same quadrant.
Now quadrant 3 – Similar to 2 weeks ago this is where the market has spent the least amount of time.
Again somewhat similar to the last test. This quandrant has done fairly well historically – mostly as the market emerges from below. The last two bear markets have seen some steep losses from this area, though.
A few quick observations to take from the above:
1) Whether using the longer 200/50 moving averages or the 50/10 moving averages the results look fairly similar.
2) It’s been a good 10 years since chasing strength worked and a good 20 since it’s worked well.
3) Pullbacks in uptrends have provided the best long-side opportunities over the last 20 years.
4) During vicious bear markets like the current one, buying below the 50-day moving average can be a difficult and dangerous endeavor. (See quadrants 3 & 4.)
The edge isn’t as prevalent initially, but over the next 2 weeks a downside tendency can be observed. Not shown above is that 70% of all instances closed below the close of the Fed day within the next 3 days. While this isn’t one of the strongest edges we’ve seen, it does suggest a pullback is the odds-on play. I also thought it was worth adding to the Fed Day Studies.
Tuesday, December 16, 2008
I got Larry Connors new book, “Short Term Trading Strategies That Work” in the mail yesterday and have read most of it. While a good portion of it has been covered by him before either in other books or on the TradingMarkets site, there are a few new ideas in there. If I can take one idea from a trading book and easily test or apply it to my own trading then I consider it worthwhile reading. This book has more than one.
There was a chapter on the RSI(2) that was quite interesting. I was pleased to see his findings were similar to Michael Stokes recent findings as well as Damian Roskil’s. Others who have published useful information on RSI(2) include Woodshedder, BHH at IBDindex, and Dogwood.
Monday, December 15, 2008
Monday, December 8, 2008
There are some issues a trader would have if they used a FTD as a market buy signal. (This is not the recommended use by IBD, but does help to determine the predictive power of FTD’s when conducting studies.) One issue is that they tend to commonly fail during difficult bear markets.
Several weeks ago in the Subscriber Letter I posted a study which looked at FTD effectiveness following the Crash of ’29. (This study only looked at the Dow.) Below is an excerpt from that Letter:
I thought it would be interesting to see how FTD’s performed following the 1929 crash. As a brief reminder, “success” for a FTD would entail either 1) The market making a new high or 2) a rally from the close of the FTD that equals at least twice the distance from the low to the FTD. Below are charts spanning the period from 1929 to in 1932.
In this chart we see several failures and one FTD that led to a rally meeting its target. While it didn’t meet the definition of success, the rally in the early part of 1930 was actually the best over the time period.
Next is ’31 – ’32:
Plenty more failures are seen here before the market finally bottoms in mid-1932. All told there were 13 failed FTD’s and one successful one before the 1932 bottom arrived.
A FTD is a positive sign when looking for a potential market bottom and subsequent rally. The current rally attempt may succeed. The market certainly seems overdue for a substantial and sustained rally. FTD and other bottoming signals have proven far less reliable over the last year. There have been a few other times where they have struggled as well. The period above is one example. Just something for the back of the mind as the current rally attempt unfolds.
Anyone who would like to purchase the FTD code for their own testing may do so here.
A note to blog readers: I will be out of action most of this week. There may not be any other posts until next week. Fear not. I shall return to the blogosphere next week with a vengeance. – Rob
A note to Gold & Silver Subscribers: I completed implementation of a new distribution system. I believe everything is working fine. If you did not receive tonight's Letter, please contact me ASAP.
Thursday, December 4, 2008
First, below is a table showing data when a two-day market rise occurs with increasing Spyx levels:
While not overwhelmingly bullish, there is a clear upside edge over the next 1-10 days based on this pattern – especially day 1. The average day over the entire time period only gained $24, so the difference is substantial.
Now let’s see what happens when only 1 of the days occurs with a rising Spyx:
Very choppy action with a slight negative overall tilt.
What if the Spyx sinks both days while the market rises?
Results here are quite a bit worse on average. It’s still close to a 50/50 proposition but risks clearly outweigh rewards.
I did look at how the 1st (bullish) test has performed over the especially choppy and bearish period from 6/1/07 to now. Over this time there have only been 4 instances and results were split. As challenging as the current market is I’m not prepared to view the results in a clearly bullish light. I do think it’s better than seeing either of the alternatives though.
Lastly, I took a look at running similar tests using straight volume instead of the Quantifiable Edges Volume Spyx to see if it provided the same edges. First I present the “bullish scenario” with two up days on rising volume:
What if we look at the bearish Spyx scenario using volume as a substitute?
Rather than clear downside edge what we have here is again less pronounced. The suggestion is chop rather than true downside.
Several times lately I have substituted Volume Spyx levels for actual volume levels in testing patterns and found the edge to be more pronounced. The S&P 500 volume Spyx is updated each night on the home page. It can be viewed for free. Gold level subscribers are able to download historical data on both S&P 500 and Nasdaq Volume Spyx Levels each night for research or further evaluation.
Wednesday, December 3, 2008
Since I showed that table a month ago we have had SIX more 5% drops. Five of the six saw the same pattern with some kind of bounce in the next few days. Below I’ve updated the table which lists all 5% drop days and whether they closed above the 5% drop day close in the next week.
Tuesday, December 2, 2008
Rather than view a table I’ve put together 4 equity curves based on owning the S&P when it was in a certain quadrant. I’ve also changed the test to buy $100k worth of the cash index rather than 1 share. This effectively gives us percentage returns. It also changes the end results slightly so you’ll need to be aware of the differences.
First let’s look at owning the S&P when it’s above the 200 and 50 day moving averages. Recall from yesterday that since 1960 the market has spent 53% of its time in this quadrant:
Over time this market position has been a steady winner. It has sat out bad drawdowns and has had some nice upside participation. I find it interesting that it peaked in 2000, though. The most recent bull market of 2003-2007 failed to see much headway at all when the market was above these two lines. I would classify it as a weak bull. It was extremely rotational and saw little in the way of buying enthusiasm when the market began hitting new highs. While I observed these characteristics at the time, this is the first time I’ve viewed it in this manner.
Now the 2nd quadrant – above the 200ma and below the 50ma. Recall from yesterday that since 1960 the market has only spent 15% of its time in this quadrant:
Here we see that this was not a bullish quadrant to be in until the bull market of the 80’s began. During the 60’s stocks spun their wheels in this area and during the 70’s they lost money. The total gains are not as large here as in Quadrant 1 when looking at percentage gains. Yesterday they were larger when looking at point gains. That’s due to the fact that this quadrant did its losing early on when the points were low and its winning more recently when the points were high. Thus the point gains were a bit exaggerated.
The exaggeration is not a bad thing. In fact, using points rather than %’s basically front weights more recent developments. Think of it as an Exponential Moving Average vs. a Simple Moving Average. The gains since 1980 have been much greater per day in here than in Quadrant 1. The total percent gained is almost the same over that time and the market has spent over 3 times the amount of time in quadrant 1 vs. quadrant 2. (55% vs. 16% since 1980) Again, Quadrant 2 outperforms Quadrant 1 although its not as obvious just from the graph. Short-term pullbacks in long-term uptrends have made for good buying opportunities.
Now Quadrant 3 – below 200ma and above 50ma. The market has spent its least amount of time here – only 9% of days since 1960:
Now let’s look at Quadrant 4 – below the 200ma and below the 50ma in which the market has spent 23% of its time.
A few spikes up when the market was emerging from a bear market or steep correction. Other than that it was not a place you wanted to bet on the long side. When the market enters this quadrant from above it’s probably best to adjust your strategy and trade in bear market mode.
A big take-away is that the market evolves over time. Sometimes this is missed when looking only at tables. Sometimes equity curves are able to capture this well. By understanding how the market has reacted in the past, especially the more recent past, we can better adapt our trading to take advantage of its movements.
Dr. Brett Steenbarger of Traderfeed also discusses market evolution and he had an especially interesting post today about how reaction to bearish momentum days has changed over the last several years.
Monday, December 1, 2008
A few comments:
1) The biggest long side edge appears to be when the market is trading above its 200 but below its 50-day moving average. At this point the market is in a long-term uptrend but is not extended to the upside over the short-term.
2) There’s a clear distinction between long-term uptrend and long-term downtrend results. The simple rule of looking for long trades above the 200ma and short trades below the 200ma appears to provide an edge.
3) Going long when the market moved into quandrant 1 (> 200 and > 50) and exiting when it left this quadrant would have only been a profitable trade 29.8% of the time. The edge comes from the fact that some very big moves were caught.
4) Going short when the market moved into quandrant 4 (< 200 and < 50) and exiting when it left this quadrant would have only been a profitable trade 22% of the time. Again the edge comes from some very big moves – including the current move.
In the next few days I’ll show a similar test using shorter-term moving averages.
Friday, November 28, 2008
Wednesday, November 26, 2008
Let’s look at the how the market set up as of Tuesday’s close as an example. SPY closed higher 3 days in a row. Tuesday was the narrowest range in the last 5 days. When trading under the 200-day moving average, this combination can signal the market is likely to pull back. See the test below:
The propensity to pull back is most pronounced over the first 4 days.
Now let’s break down the above results a little differently. First let’s look at times the SPY set up as above but was NOT coming off a 50-day low:
Results here are decidedly more negative than in the original study. Two days out for example there are only 18% winners.
Now let’s look at the first setup again, but this time we only want to see those trades that were coming of a 50-day low:
There is no longer a bearish edge to the setup. In fact, there appears to be a bit of a bullish one. Now 2 days out there are 82% winners. (Although instances are low and that’s not really the point.) The point is that it is much more dangerous to short a market coming off of fresh lows. This is especially true when the lows occurred on extremely oversold conditions.
While previous attempted rallies did quickly roll over the last 2 months, that’s not always the case. Examples of oversold bounces off lows that would have been especially dangerous to short include August 2007, October 2002 and September 2001.
Tuesday, November 25, 2008
Below is a table taken from the S&P 500 Spyx Volume 1 document (Volume 2 should be out shortly). The document is 6 pages long and details market statistics related to Spyx levels. The document is available in the Quantifiable Edges Charts page. To access the charts pages you must be a subscriber, but trial memberships are available with just a name and email address. (Click here to sign up.) The table below looks at the period 1/1/1994 - 6/30/2008.
While volume Spyx can be used to help establish a trading bias on its own, Spyx tend to show larger edges when used in conjunction with price movement or other indicators. Below is a study from last night’s Subscriber Letter which demonstrates how I might use Volume Spyx in my analysis.
Low levels of Volume Spyx have typically led to market underperformance. This is especially true when they occur on an up day. There have only been 2 instances where the S&P 500 Volume Spyx has come in under 40 and the market has gained 5%. Therefore I loosened the parameters to a 3% gain.
Still the number of instances is low, but a low Volume Spyx combined with a price spike higher has been a bad combination. Below are the 7 instances and their 2-day returns:
I will be pointing out unusual Spyx activity and what it may suggest as instances arise. I’ve also decided that at least through the end of the year I will post the chart of the S&P 500 Volume Spyx on the Quantifiable Edges home page for all to see.
Monday, November 24, 2008
The above results actually go through the Monday following Thanksgiving. Positive tendencies may have existed in the distant past, but over the 21-year period I looked at there doesn’t appear to be any edge – bullish or bearish.
The market rallied hard on Friday after hitting historically extreme conditions. There are reasons to be optimistic about some follow-through this week. Thanksgiving doesn’t appear to be one of those reasons, though.
Friday, November 21, 2008
Also notable is that the CBI popped up to 13 today.
The top 3 breadth stats come from Worden Bros. TC2000.
P.S. Traders may want to take a look at the charts following the previous two extreme posts.
Wednesday, November 19, 2008
Frequently I’ll see studies based on different indicators conflict with each other. It’s normal. One tool I use to help me sort through the studies is the Quantifiable Edges Aggregator. It helps to provide a quantified snapshot of all I’m seeing and aids me in setting my market bias.
Having studies based on one indicator conflict with each other is not normal. In this case it’s NYSE Up Issues %. It muddies my interpretation of the given indicator and dilutes its value. It’s very rare but in such cases I simply zero out the studies. When the edge is not clear, I step aside and wait until there is a clear edge.
Just a quick reminder to those in the Boston area that I'll be giving a presentation on short-term market edges tonight.
Tuesday, November 18, 2008
Monday, November 17, 2008
As I discussed in Thursday’s blog, the average true range percent in the Dow over the last 30 trading days has been over 6%. That’s not to say that Thursday’s reversal bar isn’t a positive one. Things to look for generally include a strong move higher, strong volume, and strong breadth. Thursday qualified in all areas. Against the current backdrop I've been looking for a bit more confirmation. Tests have been mixed.
Friday, November 14, 2008
Listed below is the one-day performance following all 5% up days for the S&P 500.
A couple of things to note: First, 8 of the last 9 have closed lower. Second, the lower closes have generally been tame – both in relation to the 5% up day and compared to the instances that continued to rise the next day.
Thursday, November 13, 2008
In the chart below I show the Dow Jones Industrial Average from 1986 – present. The yellow line represents the 30-day average true range on a percentage basis. The average true range over the last 30 days closed at an astonishing 6.02% on Wednesday.
Theoretical Dow Jones Index
A method of calculating a Dow Jones index (most often the DJIA) that assumes all index components hit their high or low at the same time during the day.
In other words, the "theoretical Dow" uses the daily highs for all 30 Dow components to calculate the index high, and the lows to calculate the index low. In January of 1992, Dow Jones started using the "actual" method, which calculates the index at 10-second intervals throughout the day. Before this point, the theoretical calculation was the only way to compute the high and low of the index. This method assumes that all stocks hit their high or low at the same time. Because this rarely happens, the theoretical high will almost always be higher than the actual, and the theoretical low will almost always be lower than the actual.
Taking the pre – ’92 exaggerated range into account, it appears almost certain that the recent market has displayed a higher Average True Range % than 1987. Going back further in history, the only other period where ranges approached these levels were…the 1930’s.
You may notice from the charts that the spikes in volatility tend to occur near market bottoms. An ATR spike can alert you that a bottom may be coming. Like many indicators that use moving averages, though, the 30-day ATR% is a lagging indicator. It typically peaks after the bottom has been made. Look for this measure to turn down again once a bottom (at least a temporary bottom) is in place.
Tuesday, November 11, 2008
1) Notice how prior peaks in the size of the average gap have coincided with bottoms in the market.
2) The fact that we are nearly twice the level of any prior history once again speaks to the unique and extreme nature of the current environment. This can’t be stressed enough.
3) Prior studies I published on gaps that used static assumptions for gap sizes may need to be reworked for the conclusions to be meaningful in the current environment.
Monday, November 10, 2008
Market Rewind – Jeff Pietsch does an excellent job of posting running market commentary, links, and trading thoughts. I especially enjoyed his “Studies, Systems, & Methods Vault” which was posted on Saturday.
Blog For Trading Success – Ray Barros is a trading instructor and his blog is full of tips and trading techniques as well as market analysis.
Condor Options – Plenty of insightful information on options strategies.
Masteroftheuniverse’s Weblog – A combination of trading commentary and anecdotes. Good news for those who enjoy his writing - it looks like he has a book coming out, too.
More to come...but not today.
Much more on this study was published in last night's Subscriber Letter. If you haven't yet trialed it then simply email QuantEdges@HannaCapital.com with your name and email address an I'll send you last night's Letter along with 3 upcoming ones.
Friday, November 7, 2008
Since my Dow history file goes back to 1920 I decided to look at that. There have been 4 times that the Dow dropped 5% or more 2 days in a row. They were all between 1929 and 1933.
Of course the Dow didn’t drop 5% on Thursday. It did drop over 4%. So I loosened the parameters to 4%. Results below:
I can’t even count how many tests I’ve run over the last month whose results came up “1987” or “between 1929 and 1940”.
Thursday, November 6, 2008
The meeting will take place at MIT. It is free and open to the public, although the organizers request that you RSVP. More information may be found using the link below:
I hope to get to meet some of you there.
Of the 13 instances, 10 closed higher the following day. If you give the trade 4 days to work then 12 of 13 closed higher than the entry at some point. The lone loser was October 16, 1987. That date is notable because it was the Friday prior to Black Monday.
Wednesday, November 5, 2008
Tuesday, November 4, 2008
First I wanted to see if there was general excitement about the new guy. In other words, there is not an incumbent victory…
Nothing terribly exciting here. Pretty much 50/50 over the next week.
Next I broke it down by party and ran the stats out a bit further.
Since 1920 this is how the Dow has performed after a Republican has won the white house. (I excluded 2000 since no one knew if a Republican or Democrat won for a long time after the election.)
Still not much better than 50/50 until you get out a couple of months.
Now lets see what has transpired after a Democrat wins:
A big Day 1 is apparent here, which could bode well for Wednesday if Obama wins. Beyond that – not much noteworthy. You’ll note that 40 days out the Democrat and Republican return is about the same.
No huge edges here. I’m likely done with this study for at least 4 years.