new research...

Brice · May 27, 2005

<html>

<head>
<meta http-equiv="Content-Language" content="en-us">
<meta name="GENERATOR" content="Microsoft FrontPage 5.0">
<meta name="ProgId" content="FrontPage.Editor.Document">
<meta http-equiv="Content-Type" content="text/html; charset=windows-1252">
<title>Thinking Of Selling Everything Before The Close</title>
</head>

<body>

Thinking Of Selling Everything Before The
Close? Don't Pull The Trigger Yet!


 
One of the ongoing debates that we've all heard over the years is whether or not
one assumes too much risk by holding positions overnight. Is it better to go
home flat at the end of the day, therefore avoiding all the crazy possible
events which can occur when the market is closed? No Al-Qaeda, no terrorist
attacks, few--if any--economic reports to be concerned about...simply exit the
positions and start fresh every day. 
 
For pure day traders, it's a no-brainer. Day traders are out no matter what.
But, what about everyone else? What about those of us who are in a position and
asking whether or not it's worth the risk to hold it overnight? 
 
This week, I'm going to share with you over 12 years worth of results which may
open your eyes. 
 
The Test 
 
We went back to 1993 and looked at the SPYs since they first started trading.
From 1993 - May 20 2005, the SPYs have risen from just above $43 per share to
just above $119 for a total of approximately 76 points, (which is a very nice
bull move). We ran two separate tests: The first test had us buying the SPYs on
the open each day and selling them on the close (therefore taking no overnight
risk). The second test had us buying on the close each day and exiting on the
open (therefore assuming full overnight risk). How do you think the results
proved out? 
 
Well, if you listen to the individuals who hate overnight risk, you'd likely say
that the market gains came during the day and the overnight risk ate into the
gains. But, in reality, the results are just the opposite (and somewhat
shocking). 
 
The Results 
 
If one had bought the SPYs on the close over the past 12 years and sold them on
the next day's open, they would have made 143 SPY points (plus, they would have
also received all the quarterly dividends). And had one bought the SPYs on the
open each day and sold the position on the close, how would they have done? 
They would have lost money! How much? A little more than 67 points. Yes, in
spite of a solid upward move in SPY prices, the intraday move--on a net
basis--was negative! All the gains, and a great deal more, came from holding the SPYs overnight. And then a good chunk of these gains were lost while the market
was open. 
 
What Does This Mean? 
 
First, just a reminder that the above test results are simulated and it cannot
be assumed that the results will be the same in the future. Also, one cannot
trade this outright because of the transaction costs. 
 
With that said, the results show that at least from 1993, the S&P 500 index
market participants have been more than amply rewarded for holding positions
overnight. Does this mean that day trading doesn't work? No, of course not.
There are many successful daytraders and they are succeeding for a multitude of
reasons, including shorting the market at appropriate times. But, the long side
of the S&P 500 has provided no positive edge to day traders on an open-to-close
basis since 1993. The long edge (and this edge has been substantial) has gone to
those holding positions overnight. And these overnight holdings have included
9/11, the Long Term Capital/world financial crisis in 1998, the war with Iraq,
multiple major political events, corporate scandals, economic reports, and a
whole slew of other things which cause fear in most everyone. And, it appears
that these fears have created a continuous mis-pricing of the market and this
mis-pricing has allowed for overnight holders to be well rewarded.

<img border="0" src="http://images.tradingmarkets.com/2005/Connors/lc052705-01.gif" width="729" height="457"> 
 
 
 
 
Summary 
 
Should you hold a position overnight (especially a SPY position) simply for the
sake of a past edge? No, I don't believe so. These test results do show,
however, that contrary to popular belief, from 1993 - May 2005, the gains
have occurred while the market has been closed, not while it has been open. If this holds true in the future, it may help you sleep a lot
better while you're carrying your positions overnight. 
 
Have a great week trading (and enjoy the 3-day holiday weekend)!
Larry Connors

</body>

</html>

Avid_Consumer · May 27, 2005

This has been especially pronounced in the last 2 years following down days in the indices. I recently stumbled on this (as many others probably have) after watching one too many glaring morning squeezes, and have been considering posting the code in the hope of pressuring some of that accumulation back into the daytime hours (yes, I realize how ridiculous that probably is) ..but I didn't realize this pattern is 12 years old. It really got smashed from the top of the bull to the bottom of the retrace, roughly speaking. Maybe it's better suited as a long term timing indicator.

Who does the big overnight buying? ..foreigners... sneaky gorillas?

Avid_Consumer · May 31, 2005

today seems a good a day as any to give away some code. here's a demonstration of the above mentioned relationship. See attached curve with $2.40 comms and $0 slip assumptions. ... the way I see it, if the stock market must go up, better for us that it happen during the day and not via large players, ppt or whoever in the middle of the night. Will be interesting to observe what effect if any this information can elicit

{apply this to ES 1 min bars. Generates curves on all the indices to varying degrees with appropriate param tweaks. Because of the somewhat low number of trades, it's relatively insensitive to slippage. If the params bother you, imo it's a light application of curve fit, and still curves fine off proper daily close and open prices}

input:
ETime(0),
XTime(8);

if closed(0) < opend(0) then
if time = calctime(1600, (-1 * ETime)) then
buy this bar;

if time >= calctime(931, XTime) and time < calctime(1600, (-1 * ETime)) then
sell this bar;

Hydroblunt · May 31, 2005

It's actually not surprising at all. Gap & churn is way too common in this market.
To make serious money, gotta hold overnights.

5yrtrader · May 31, 2005

Wow, I think this is a really interesting study. This shows that indicators that include volume should probably be tossed out the window. All the volume is during the day, but all the money is made overnight.

tradermaji · May 31, 2005

I have done some preliminary studies on using a Day of Week as a filter for buying and selling stocks.

I analyzed 3 scenarios in Wealthlab using the last five years of data on SPY. I used $10,000 per position and deducted $20 commissions per RT.

Here are the results in a nutshell:

1. Buy at close and sell next day at open:

Profit/Loss $ $6,691.29
Total Winners 772
Total Losers 725

2. Buy at Close of Wednesdays, Thursdays and Friday's and sell on the next day's open (no trades on option expiration weeks):

Profit/Loss $ $22,441
Total Winners 437
Total Losers 399

3. Bought on a Friday and Sold on Open on Monday

Profit/Loss $ $3,153.10
Total Winners 150
Total Losers 143

Just playing with days and seeing how much I can curve fit data

Maji

Quiet1 · May 31, 2005

what's new about this? last time i saw this study would have been at least 3 years ago and i think a lot longer than that since I first knew it.

still amazing though when you think about it.

Q1

Dustin · May 31, 2005

Quote from Brice:

One of the ongoing debates that we've all heard over the years is whether or not
one assumes too much risk by holding positions overnight. Is it better to go
home flat at the end of the day, therefore avoiding all the crazy possible
events which can occur when the market is closed? No Al-Qaeda, no terrorist
attacks, few--if any--economic reports to be concerned about...simply exit the
positions and start fresh every day.

The debate about holding overnights doesn't apply to SPY imo. The risk is in stocks that could open sharply against you due to stock specific news.

Kind of interesting study anyway.

Threei · May 31, 2005

I've read about similar study at least three times over several years. I haven't run any kind of testing on it so can't cite anything negating the idea but I always wanted to ask this:

If as study says

If one had bought the SPYs on the close over the past 12 years and sold them on the next day's open, they would have made 143 SPY points (plus, they would have also received all the quarterly dividends). And had one bought the SPYs on the open each day and sold the position on the close, how would they have done? 
They would have lost money! How much? A little more than 67 points.

then why not simply make money with no thinking involved: buy every close, sell AND short every open, cover AND buy again every close? No second-guessing, no exceptions, just let statistics do all the work. Have we finally found Holy Grail?

Anyone ran statistics on drawbacks? Defined the capital required to sustain them?

My point is, all these statistics say is (providing they are correct): positive gaps overweight negative ones over period of time involved in study and negative closings overweight positive ones over the same period. This is not a strategy yet, this is just a foundation for a strategy. So why noone employed it if this is known for quite some time? I think I know why but I am not big on system developments so I will let someone more knowledgeable to take a shot.

Avid_Consumer · Jun 1, 2005

what do people think about trying to squeeze a relationship like this out of the market by exposing it in full view? is that impossible. if everyone publicized code that expresses relationships counter to their preferred trading time and risk profiles ...

maybe this is a different thread. it sounds like most of you have been aware of this specific pattern for a long time, but the general public, probably not at all. i think i need a thread titled 'free code' or something

new research...

Attachments