May Seasonality on the S&P 500

1,408 views

With all the Sell in May talk, I thought it might be helpful to see how bad May has actually been, seasonally speaking.

May Seasonality

Click on the chart to enlarge. Calculations start at the close of the last trading day of April and end on the close of the last trading day of May.

$SPX data goes back to 1928. $SPY data goes back to 1993. The two sets of data demonstrate how May seasonality has changed recently, if recently is over the last 20 years.

May Statistics for $SPY ($SPX)

  • Average Monthly Profit/Loss = 0.70% (-0.22%)
  • Winning Months= 60.00% (54.12%)
  • Worst May = 2010 loss of -7.9% (1940 loss of -24.1%)
  • Best May = 2001 gain of +6.2% (1990 gain +9.2%)

The data show that May has historically been weak, but over the last 20 years, May has been able to grind out gains.

I want to note that the gap up shown in the $SPY results on the first day in May caught my attention. I thought that there was likely an error somewhere in my code. I have not been able to find any error, and going through charts by hand, it does seem that the first day of May has tended to see a rather large gap up, at least over the last 20 years.

Profit Distributions, Using $SPX Data

May profit Distribution

Equity Curve:

May Seasonality Equity Curve

The equity curve shows that May tended to be weak from 1928 to around 1985. Since then, the month has tended to close higher. This shows that one has to be careful blindly following market mantras.

If you are interested in this monthly seasonality type study, Market Sci has recently penned a couple of excellent posts on this subject:

Calendar Month Seasonality Debunked?

“Sell in May” Debunked?

Improving Sell in May: Use the 200 Day

1,537 views

The previous article, Long Term Investors Crazy Not to Sell in May?, we examined a basic Sell in May and Go Away strategy. This update will add an additional factor to the strategy, which is to use the 200 day simple moving average to keep the strategy out of the S&P when it is in bear market territory.

The Rules:

  • Long $SPY from the close of the last trading day of October through the last trading day of April IF $SPY is above its 200 day simple moving average.
  • Long VBLTX (Vanguard Long Term Bond Index Fund) from the last trading day of April to the last trading day of October.
  • First day of test is 6.20.1996. Last day is 3.28.2013. First day is first day of history for VBLTX.
  • No commissions or slippage included.

Note: I spent some time trying to find data for a bond index which would allow me to go back farther than 1996. I did not have any luck. If anyone knows of a suitable government bond / credit index (or mutual fund tracking such an index) with accessible data going back farther than 1996, please let me know.

The Results:

Sell in May with 200 Day Avg

The top pane is $SPY. The middle pane is the equity curve for the strategy. The lower pane shows the drawdowns in percentage terms. The blue portions of the equity curve represents time in $SPY while the green represents time in VBLTX. Compared to the basic strategy, adding the 200 day moving average requirement keeps the system in VBLTX for significant amounts of time.

Statistics:

Sell in May with 200MA Stats

Because of the way I coded his, all the trade information pertains only to the $SPY trades. The returns from VBLTX are built into the overall returns but are not considered to be trades.

If I subtract out the added return from VBLTX, the annualized return from just holding $SPY is 10.69%.

$SPY buy and hold over the same period is 5.26% with a maximum drawdown of -56.47%. So even if you do not rotate into a vehicle to provide a return on your cash, you still double buy and hold just by selling in May.

Historical Profit Table:

Sell in May with 200MA Profit Table

The strategy has performed unbelievably well, any way you slice it.

However, I have my doubts going forward. They are best encapsulated in a comment left by Kill The Banks, in the previous post:

I think there needs to be some evaluation of bonds as an appreciating asset class going forward. There was some blogoland chatter a few months ago regarding the “max theoretical gain” left in bonds (forget if it was the 10yr or the 30yr) and that number being around 17% total net or so (IIRC). Sentiment seems to be that given the current interest rate situation the only direction rates can really go is up, which translates into “bonds down”.

Me, I’d be wary of the forward usefulness of any methodology incorporating bond longs that has good backtest results.

With that comment in mind, what would we rotate into, if not for bonds? Is there another asset class that should be considered during the May – October seasonal period?

Long Term Investors Crazy Not to Sell in May?

1,044 views

I was reading a post from EconomPic called Checking in On the World’s Greatest Rotation Strategy. This strategy is essentially just Sell in May and Go Away, but has one buy the Long government / credit bond index rather than just sitting in cash from May – October.

Oddly enough, I had never tested this simple strategy before, and the results EconomPic posted were good enough that I had to take a look under the hood to see exactly what was making this system run.

The Rules are simple:

  • Long $SPY from the close of the last trading day of October through the last trading day of April.
  • Long VBLTX (Vanguard Long Term Bond Index Fund) from the last trading day of April to the last trading day of October.
  • First day of test is 6.20.1996. Last day is 3.29.2013. First day is first day of history for VBLTX.
  • No commissions or slippage included.

The Results:

Sell in May Basic

The top pane is $SPY. The middle pane is the equity curve for the strategy. The lower pane shows the drawdowns in percentage terms. The blue portions of the equity curve represents time in $SPY while the green represents time in VBLTX.

Statistics:

Sell in May Basic Stats

Because of the way I coded his, all the trade information pertains only to the $SPY trades. The returns from VBLTX are built into the overall returns but are not considered to be trades. Perhaps in future tests I’ll break out both $SPY and $VBLTX as separate trades.

If I subtract out the added return from VBLTX, the annualized return from just holding $SPY is 8.82%.

$SPY buy and hold over the same period is 5.26% with a maximum drawdown of -56.47%. So even if you do not rotate into a vehicle to provide a return on your cash, you still beat buy and hold just by selling in May.

Historical Profit Table:

Sell in May Basic Profit Table

This strategy has just been killing it. I’m leery to assume it can continue killing it. Some questions:

  • Assuming government credit / bonds do not continue to perform as well, with what would we replace VBLTX?
  • Can we add additional robust timing mechanisms to improve performance of $SPY and/or VBLTX?

I’m looking forward to reading your ideas on these questions in the comments.

The next post will add in an additional timing measure for $SPY.

Let’s Look at Seasonality for April

1,038 views

Still a week to go, but I was getting a little nervous and wanted to take a look anyway.

We know from this post from CXO what to look for at the beginning of April, but what about the rest of the month?

April Seasonality

Click on the chart to enlarge. Calculations start at the close of the last trading day of March and end on the close of the last trading day of April.

$SPX data goes back to 1928. $SPY data goes back to 1993. The two sets of data demonstrate how April seasonality has changed recently, if recently is over the last 20 years.

April Statistics for $SPY ($SPX)

  • Average Monthly Profit/Loss = 2.00% (1.12%)
  • Winning Months= 70.00% (60.00%)
  • Worst April = 2002 loss of -5.9% (1932 loss of -14.8%)
  • Best April = 2001 gain of +9.4% (1933 gain +31.0%)

Profit Distributions:

profdist

Equity Curve:

1_ Portfolio Equity

April returns since 2007 have been excellent. Last year was the first losing April since 2005. My belief in mean reversion has me betting that April 2013 will be a loser.

Charting the Santa Claus Rally, 2012 Edition

375 views

Charting December to date against previous Decembers shows that the Santa Rally may have just started a few days early.

Let’s take a look at the S&P 500, starting in 1928, and $SPY, starting in 1993, to chart the average December performance.

I’ve assumed that the $SPX and $SPY was bought on the open on the first trading day of December and sold at the close on the last trading day of December.

Santa really gets flying about halfway through December. In fact Santa rarely disappoints. We can see how true this is above.

While current December action (white line) is more volatile due to not being smoothed by the averaging process, the month seems to be following a typical December pattern, except that Santa appears to have started his run several days earlier than normal.

December Statistics (using $SPX)

  • Average Monthly Profit/Loss = 0.89%
  • Winning Months= 69.41%
  • Worst December = 1931 loss of -17.27%
  • Best December = 1991 gain of +11.16%

Equity Curve: (using $SPX)

Profit Distributions: (using $SPX)

The Christmas Spirit is really building around my house…Here’s hoping that today’s rally means Santa has fired up his sleigh and is heading your way!

2011 Santa Rally: Santa Did Not Disappoint

344 views

Per a reader request, below is the chart from the previous Santa Rally post, but it has been updated to include the rest of December 2011 SPY history.

Compared to the SPY December average, this year’s Santa Rally developed a day earlier but peaked at the same day (day 18) as usual. For future reference, day 18 is the first day the market opens after Christmas.

The Santa Rally is truly a wonderful gift of seasonality. This year’s rally added better than 4% to the SPY, putting it near breakeven for the year.

Charting the Santa Claus Rally

1,744 views

While it seems ‘ole Santa has caught a case of the Bah Humbugs, charting December to date against previous Decembers shows that the  Santa Rally may just be getting started.

Let’s take a look at the S&P 500, starting in 1960, and SPY, starting in 1993, to chart the average December performance.

I’ve assumed that the $SPX and SPY was bought on the open on the first trading day of December and sold at the close on the last trading day of December.

As shown above, Santa really gets flying about halfway through December. In fact Santa rarely disappoints. We can see how true this is above.

While current December action (white line) is more volatile due to not being smoothed by the averaging process, it is easy to see that so far, the month seems to be following a typical December pattern.

Note that SPY, which represents more recent data, shows that the Santa Rally has been losing steam over the past decade or so.

December Statistics (using $SPX)

  • Average Monthly Profit/Loss = 1.50%
  • Winning Months= 71.15%
  • Worst December = 1968 loss of -4.2%
  • Best December = 1991 gain of +11.1%

Profit Distributions: (using $SPX)

Equity Curve: (using $SPX)

Let’s hope that Santa will soon be firing up his sleigh! If history is any guide, the rally should start within a few days…

May Seasonality on the S&P 500

1,408 views

With all the Sell in May talk, I thought it might be helpful to see how bad May has actually been, seasonally speaking.

May Seasonality

Click on the chart to enlarge. Calculations start at the close of the last trading day of April and end on the close of the last trading day of May.

$SPX data goes back to 1928. $SPY data goes back to 1993. The two sets of data demonstrate how May seasonality has changed recently, if recently is over the last 20 years.

May Statistics for $SPY ($SPX)

  • Average Monthly Profit/Loss = 0.70% (-0.22%)
  • Winning Months= 60.00% (54.12%)
  • Worst May = 2010 loss of -7.9% (1940 loss of -24.1%)
  • Best May = 2001 gain of +6.2% (1990 gain +9.2%)

The data show that May has historically been weak, but over the last 20 years, May has been able to grind out gains.

I want to note that the gap up shown in the $SPY results on the first day in May caught my attention. I thought that there was likely an error somewhere in my code. I have not been able to find any error, and going through charts by hand, it does seem that the first day of May has tended to see a rather large gap up, at least over the last 20 years.

Profit Distributions, Using $SPX Data

May profit Distribution

Equity Curve:

May Seasonality Equity Curve

The equity curve shows that May tended to be weak from 1928 to around 1985. Since then, the month has tended to close higher. This shows that one has to be careful blindly following market mantras.

If you are interested in this monthly seasonality type study, Market Sci has recently penned a couple of excellent posts on this subject:

Calendar Month Seasonality Debunked?

“Sell in May” Debunked?

Improving Sell in May: Use the 200 Day

1,537 views

The previous article, Long Term Investors Crazy Not to Sell in May?, we examined a basic Sell in May and Go Away strategy. This update will add an additional factor to the strategy, which is to use the 200 day simple moving average to keep the strategy out of the S&P when it is in bear market territory.

The Rules:

  • Long $SPY from the close of the last trading day of October through the last trading day of April IF $SPY is above its 200 day simple moving average.
  • Long VBLTX (Vanguard Long Term Bond Index Fund) from the last trading day of April to the last trading day of October.
  • First day of test is 6.20.1996. Last day is 3.28.2013. First day is first day of history for VBLTX.
  • No commissions or slippage included.

Note: I spent some time trying to find data for a bond index which would allow me to go back farther than 1996. I did not have any luck. If anyone knows of a suitable government bond / credit index (or mutual fund tracking such an index) with accessible data going back farther than 1996, please let me know.

The Results:

Sell in May with 200 Day Avg

The top pane is $SPY. The middle pane is the equity curve for the strategy. The lower pane shows the drawdowns in percentage terms. The blue portions of the equity curve represents time in $SPY while the green represents time in VBLTX. Compared to the basic strategy, adding the 200 day moving average requirement keeps the system in VBLTX for significant amounts of time.

Statistics:

Sell in May with 200MA Stats

Because of the way I coded his, all the trade information pertains only to the $SPY trades. The returns from VBLTX are built into the overall returns but are not considered to be trades.

If I subtract out the added return from VBLTX, the annualized return from just holding $SPY is 10.69%.

$SPY buy and hold over the same period is 5.26% with a maximum drawdown of -56.47%. So even if you do not rotate into a vehicle to provide a return on your cash, you still double buy and hold just by selling in May.

Historical Profit Table:

Sell in May with 200MA Profit Table

The strategy has performed unbelievably well, any way you slice it.

However, I have my doubts going forward. They are best encapsulated in a comment left by Kill The Banks, in the previous post:

I think there needs to be some evaluation of bonds as an appreciating asset class going forward. There was some blogoland chatter a few months ago regarding the “max theoretical gain” left in bonds (forget if it was the 10yr or the 30yr) and that number being around 17% total net or so (IIRC). Sentiment seems to be that given the current interest rate situation the only direction rates can really go is up, which translates into “bonds down”.

Me, I’d be wary of the forward usefulness of any methodology incorporating bond longs that has good backtest results.

With that comment in mind, what would we rotate into, if not for bonds? Is there another asset class that should be considered during the May – October seasonal period?

Long Term Investors Crazy Not to Sell in May?

1,044 views

I was reading a post from EconomPic called Checking in On the World’s Greatest Rotation Strategy. This strategy is essentially just Sell in May and Go Away, but has one buy the Long government / credit bond index rather than just sitting in cash from May – October.

Oddly enough, I had never tested this simple strategy before, and the results EconomPic posted were good enough that I had to take a look under the hood to see exactly what was making this system run.

The Rules are simple:

  • Long $SPY from the close of the last trading day of October through the last trading day of April.
  • Long VBLTX (Vanguard Long Term Bond Index Fund) from the last trading day of April to the last trading day of October.
  • First day of test is 6.20.1996. Last day is 3.29.2013. First day is first day of history for VBLTX.
  • No commissions or slippage included.

The Results:

Sell in May Basic

The top pane is $SPY. The middle pane is the equity curve for the strategy. The lower pane shows the drawdowns in percentage terms. The blue portions of the equity curve represents time in $SPY while the green represents time in VBLTX.

Statistics:

Sell in May Basic Stats

Because of the way I coded his, all the trade information pertains only to the $SPY trades. The returns from VBLTX are built into the overall returns but are not considered to be trades. Perhaps in future tests I’ll break out both $SPY and $VBLTX as separate trades.

If I subtract out the added return from VBLTX, the annualized return from just holding $SPY is 8.82%.

$SPY buy and hold over the same period is 5.26% with a maximum drawdown of -56.47%. So even if you do not rotate into a vehicle to provide a return on your cash, you still beat buy and hold just by selling in May.

Historical Profit Table:

Sell in May Basic Profit Table

This strategy has just been killing it. I’m leery to assume it can continue killing it. Some questions:

  • Assuming government credit / bonds do not continue to perform as well, with what would we replace VBLTX?
  • Can we add additional robust timing mechanisms to improve performance of $SPY and/or VBLTX?

I’m looking forward to reading your ideas on these questions in the comments.

The next post will add in an additional timing measure for $SPY.

Let’s Look at Seasonality for April

1,038 views

Still a week to go, but I was getting a little nervous and wanted to take a look anyway.

We know from this post from CXO what to look for at the beginning of April, but what about the rest of the month?

April Seasonality

Click on the chart to enlarge. Calculations start at the close of the last trading day of March and end on the close of the last trading day of April.

$SPX data goes back to 1928. $SPY data goes back to 1993. The two sets of data demonstrate how April seasonality has changed recently, if recently is over the last 20 years.

April Statistics for $SPY ($SPX)

  • Average Monthly Profit/Loss = 2.00% (1.12%)
  • Winning Months= 70.00% (60.00%)
  • Worst April = 2002 loss of -5.9% (1932 loss of -14.8%)
  • Best April = 2001 gain of +9.4% (1933 gain +31.0%)

Profit Distributions:

profdist

Equity Curve:

1_ Portfolio Equity

April returns since 2007 have been excellent. Last year was the first losing April since 2005. My belief in mean reversion has me betting that April 2013 will be a loser.

Charting the Santa Claus Rally, 2012 Edition

375 views

Charting December to date against previous Decembers shows that the Santa Rally may have just started a few days early.

Let’s take a look at the S&P 500, starting in 1928, and $SPY, starting in 1993, to chart the average December performance.

I’ve assumed that the $SPX and $SPY was bought on the open on the first trading day of December and sold at the close on the last trading day of December.

Santa really gets flying about halfway through December. In fact Santa rarely disappoints. We can see how true this is above.

While current December action (white line) is more volatile due to not being smoothed by the averaging process, the month seems to be following a typical December pattern, except that Santa appears to have started his run several days earlier than normal.

December Statistics (using $SPX)

  • Average Monthly Profit/Loss = 0.89%
  • Winning Months= 69.41%
  • Worst December = 1931 loss of -17.27%
  • Best December = 1991 gain of +11.16%

Equity Curve: (using $SPX)

Profit Distributions: (using $SPX)

The Christmas Spirit is really building around my house…Here’s hoping that today’s rally means Santa has fired up his sleigh and is heading your way!

2011 Santa Rally: Santa Did Not Disappoint

344 views

Per a reader request, below is the chart from the previous Santa Rally post, but it has been updated to include the rest of December 2011 SPY history.

Compared to the SPY December average, this year’s Santa Rally developed a day earlier but peaked at the same day (day 18) as usual. For future reference, day 18 is the first day the market opens after Christmas.

The Santa Rally is truly a wonderful gift of seasonality. This year’s rally added better than 4% to the SPY, putting it near breakeven for the year.

Charting the Santa Claus Rally

1,744 views

While it seems ‘ole Santa has caught a case of the Bah Humbugs, charting December to date against previous Decembers shows that the  Santa Rally may just be getting started.

Let’s take a look at the S&P 500, starting in 1960, and SPY, starting in 1993, to chart the average December performance.

I’ve assumed that the $SPX and SPY was bought on the open on the first trading day of December and sold at the close on the last trading day of December.

As shown above, Santa really gets flying about halfway through December. In fact Santa rarely disappoints. We can see how true this is above.

While current December action (white line) is more volatile due to not being smoothed by the averaging process, it is easy to see that so far, the month seems to be following a typical December pattern.

Note that SPY, which represents more recent data, shows that the Santa Rally has been losing steam over the past decade or so.

December Statistics (using $SPX)

  • Average Monthly Profit/Loss = 1.50%
  • Winning Months= 71.15%
  • Worst December = 1968 loss of -4.2%
  • Best December = 1991 gain of +11.1%

Profit Distributions: (using $SPX)

Equity Curve: (using $SPX)

Let’s hope that Santa will soon be firing up his sleigh! If history is any guide, the rally should start within a few days…