Backtesting the S&P500 Volatility Timing Model

Posted on Mi 27 Juni 2007 in Stocks • 1 min read

  • ATS
  • Neural Nets
  • Trends tags:
  • AI meta: _aioseop_keywords: 401k, Model, NeuralNets, S&P500, Investing, Timing, AI podPressPostSpecific: a:6:{s:15:"itunes:subtitle";s:15:"##PostExcerpt##";s:14:"itunes:summary";s:15:"##PostExcerpt##";s:15:"itunes:keywords";s:17:"##WordPressCats##";s:13:"itunes:author";s:10:"##Global##";s:15:"itunes:explicit";s:7:"Default";s:12:"itunes:block";s:7:"Default";} dsq_thread_id: '181041820' author:

    I'm still back-testing my S&P500 Volatility Timing Model and have come across some interesting results that I'd like to share. I back tested the model to the beginning of 2000 and let the model generate "buy" signals. I tabulated them in the first table below and then calculated a % return for each position based on Monday's closing price for the S&P500.

    You can see that the first "buy" signal on 8/28/2000 still has a negative return but gains increased as the model bought into the S&P500 by riding the market down between 2000 and 2002.

    SP500 Wins

    It's pretty easy to get euphoric over that 81.61% gain since the low on 9/23/2002 but its all a matter of perspective. Imagine your losses if you looked at this model on that day. Your return table would've looked something like this below.

    SP500 Losses

    Now that's a pretty painful ride to bottom! It's one thing to say "buy them when they ain't" but its another to actually do it. Implementing this timing strategy can be one of the toughest things you can do financially. After all, things can't get any worse after you buy, or can they? :)