Building A Simple Trading System To Diversify Your Portfolio: Part 3

| About: SPDR S&P (SPY)

After a short Easter break, it is about time to continue this series and introduce our third and final seasonal edge, that this simple system is built upon. In Part 1 we had a look at the bullish bias on days of the FOMC meetings, the second Part identified a very bullish bias on the first trading days of a month and this article will examine Options Expiration Days for similar tendencies.

To get about the same amount of trading opportunities as with the first two edges, we will concentrate on every months third Friday, when weekly and monthly options expire; this way "triple witching" days (the third Fridays of March, June, September and December, when quarterly options expire as well) will be covered, too.

It turns out that these days have a bearish bias, when measured from the open to the close, and there are a few theories for why this is the case.
One claims that institutional size investors that are involved in equities and options trading simultaneously tend to only sell stock positions after their options expired, so that they can lock in the options expiration at higher prices before putting pressure on the underlying.
Another one blames this edge on a general selling pressure before the weekend, to not run extensive overnight/over weekend risk, when the markets are closed for two full days.
I personally do not really agree with those explanations, but if a steady exploitable edge reveals itself along the way I don´t mind taking my chances.

Now before we start crunching the numbers, I have to point out one particular problem with this analysis: The free available historical data on the S&P 500 is shaky, concerning the daily open price.
Last Friday (04/05/2013) for example, most sources post an opening price of 1559.98, which was the previous days close, even though the actual open was about 20 points below that level.
Those flaws are spread all over the free available historical data and make it useless for backtesting this strategy on the S&P 500.

In this case there are two alternative ways: to backtest the strategy on Nasdaq 100 data, or using the SPY as a proxy for the S&P 500.
Even though the SPY only got introduced in 1993 which makes the available data five years shorter than for the former two presented edges, I decided to go with it as I wanted to stick with the same underlying (Index) and I would have had to account for a higher level of volatility when choosing to backtest on Nasdaq 100 data.

So the simple rules for this test are the following:

  • Short the SPY on the open of every third Friday (or Thursday, if Friday is a holiday) of the month.
  • Cover the short at the close of the same day.

  • February 1993 to February 2013
    Trades: 240  
    Winning Trades: 135 56.3%
    Losing Trades: 105 43.8%
    Avg. Winner: +0.677%  
    Avg. Loser: -0.491%  
    Max. Winner: +2.603%  
    Max. Loser: -3.182%  
    Expectancy*: +0.166%  
    Performance: +47.72%  

    *=(win. percent*Avg. Winner)-(lose percent*Avg. Loser)Even though the hitrate is not as high as one could hope for, the comfortably higher ratio of average winner vs avg loser is enough to give us a constant edge and a positive expectation.
    Just as in the last articles, I have to stress the issue of having hard stops in the market, to not get completely overrun by a single bad trade; so lets again add a 1.5% stop loss.
    Trades: 240  
    Winning Trades: 132 55%
    Losing Trades: 108 45%
    Avg. Winner: +0.666%  
    Avg. Loser: -0.504%  
    Max. Winner: +2.603%  
    Max. Loser: -1.5%  
    Expectancy: +0.139%  
    Performance: +38.64%  

    A slightly lower hitrate as well as a slightly worse avg. win/avg. loss ratio, decrease the overall performance and expectancy, but this is a price one should be willing to pay for the piece of mind of ruling out catastrophic single losses.
    Lets take another step and see how the results would look if we would only trade options expiration Fridays if the previous close was above the 200 day moving average.

    above 200MA       above 200MA + Stop Loss    
    Trades: 168     Trades: 168  
    Winning Trades: 96 57.14%   Winning Trades: 95 56.55%
    Losing Trades: 72 42.86%   Losing Trades: 73 43.45%
    Avg. Winner: +0.617%     Avg. Winner: +0.6%  
    Avg. Loser: -0.332%     Avg. Loser: -0.358%  
    Max. Winner: +2.571%     Max. Winner: +2.571%  
    Max. Loser: -1.634%     Max. Loser: -1.5%  
    Expectancy: +0.21%     Expectancy: +0.184%  
    Performance: +41.83%     Performance: +35.71%  

    This combination filters out a handful of bad trades, which pushes especially the avg. loser number to a pretty low level. Only the lower trade frequency makes the overall performance lag the original version, but the experienced drawdowns are not even close as deep, which becomes obvious in this comparing chart.

    In this case the quick conclusion would be to trade the filtered version of this system; but I first of all want to structure all three edges the same simple way and second of all choosing a more imperfect, realistic version makes it easier to later introduce and demonstrate a powerful tool to decide whether to trade a system at all.
    So in the end we will just stick to the unfiltered ruleset with the added stop loss.


    This third and final setup completes the seasonal edges theme that shall be traded in this system and these are the rules in a quick overview:

      Fed Days:

    • Buy the S&P 500 on the close of the day before an FOMC meeting, sell exactly one day later.

    • Stop loss 1.5% below the entry
      (March 1988 - March 2013)

      First Day of the Month:
    • Buy the S&P 500 on the close of the last day of the month, sell exactly one day later.

    • Stop loss 1.5% below the entry
      (March 1988 - March 2013)

      Options Expiration Friday:

    • Short the SPY on the open of the third Friday of the month, cover short on the close of the same day.

    • Stop loss 1.5% above the entry
      (February 1993 - March 2013)

    It becomes obvious that this system follows some of the simplest rules possible; you could mark those days at the beginning of a calendar year and the only rule concerning risk management so far is a hard 1.5% stop loss. Theoretically there is a lot of room to optimize those parameters, on the other hand it would be questionable to tweak the rules too much in your favor as it is very easy to over-optimize a system that aims for profits as small as this one does.

    Combined Edges    
    Trades: 740  
    Winning Trades: 429 57.97%
    Losing Trades: 311 42.03%
    Avg. Winner: +0.833%  
    Avg. Loser: -0.578%  
    Max. Winner: +5.136%  
    Max. Loser: -1.5%  
    Expectancy: +0.24%  
    Performance: +470.3%

    I know it is dangerous to post a jawdropping chart like this without mentioning that these results are still of highly theoretical nature, as we neither have had a look at the obstacles one encounters when trading this system live nor accounted for slippage and transaction fees yet.
    Nevertheless it is still very encouraging to see that a few easy trading rules can replicate the S&P´s long term performance, with a fraction of the exposure and way less gnarly drawdowns.

    One thing still missing to make this system more complete is a routine that tells us when to abandon an edge completely and when to maybe up the ante when things are going well, which will be introduced in the next article.
    The last step then would be to consider the possible ways to trade this system in real life, and how it could be integrated in a more diversified portfolio.

Disclosure: I am long SPY. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: The results presented here are of a highly theoretic nature. Past performances are no guarantee for the future.

About this article:

Author payment: $35 + $0.01/page view. Authors of PRO articles receive a minimum guaranteed payment of $150-500. Become a contributor »
Problem with this article? Please tell us. Disagree with this article? .