Emini S&P Afterhours Behaviour Part 1 – Distribution Study on AH High and Low

By Lawrence

A series of articles on Emini S&P 500 afterhours characteristics, biases, and trading strategies.

A Look at the Price Distribution

Following is the price distribution of,

  • AH High – AH Open (2nd pane)
  • AH Low – AH Open (3rd pane)

The chart covers data from mid 2005 up to yesterday.

ES 24 Hours 30-Min Distribution Study 2_20120207_082243

Notice how the positive side is fatter overall relative to the negative side. This tells us there is an up drift bias in the afterhours session. Keep in mind that there is a 1.5 year selloff from late 2007 to beginning of 2009. Thus having an up drift bias is reasonable – when the market is going higher overall, you get afterhours bias to the upside and when the market is going lower overall, you get afterhours bias to the downside.

The Gap Effect

Following is a chart of same configuration with a trading system that long at the start of the afterhours session and close the position by open next trading day. Instead of using actual dollar value, I choose to show the amount of points net gained (or lost).

ES 24 Hours 30-Min Distribution Study_20120207_083123

Overall the winning rate is just 52% with a wipe out period during the selloff period of 2008-2009. Most people will simply dismiss the concept and thought they have not found a winning system and move on. That is not what we do when we are investigating the behaviour of the market you trade. What we should do is take notes of the special characteristics so that you can better handle the market.

Notice that back from 2005 to 2007 ES rallied approximately 250 points. The gap system produced as much and even more than that by end of 2007.

Then during the selloff period, ES dropped 800+ points yet the total losses from trading the gap is less than 400 points.

After that, from 2009 bottom to now, ES rallied approximately 600 points and the gaps are responsible for 450 points.

This behaviour is consistent with earlier historical data from 1995 to 2005.

So what is this telling us?

  • In an overall up trend on weekly or higher timeframes, majority of the gains (if not all of them) in ES come from gaps.
  • In an overall down trend on weekly or higher timeframes, only half of the losses come from gaps.

Technical Notes

The charts are 30-min 24 hours ES.

The distributions are generated with the following formulas using the distribution plot indicator.

Condition

hour (time) = 10 and minute (time) = 0

Value

$was_ah := $ah;
$ah := time >= maketime (16, 30, 0) or time <= maketime (9, 30, 0);

$rth_close := if (not $was_ah and $ah, close (1), $rth_close);
$ah_open := if (not $was_ah and $ah, open, $ah_open);
$ah_high := choose (not $was_ah and $ah, high, $ah, maxList (high, $ah_high), $ah_high);
$ah_low := choose (not $was_ah and $ah, low, $ah, minList (low, $ah_low), $ah_low);

$ah_high - $ah_open;

Change last line of code to following for the distribution of AH Low – AH Open.

$ah_low - $ah_open;

So if you like to explore the data yourself you can make changes to the code above and generate statistics yourself.

Share

Comments
  • smilingsynic February 8, 2012 at 2:30 pm

    I did a study of Google years ago and determined that ALL of the gains, at the time, came overnight.

    Intraday, there was no bias.

    • Lawrence Chan February 8, 2012 at 7:05 pm

      At times, the gaps during a short period of time, will contribute to all the gains (or even more than the net gains). That topic will come up in this series later.

  • You must be logged in to comment. Log in