Saturday, November 7, 2015

Original post from October 17, 2015

I have not been posting for about 10 weeks after I decided to concentrate on switching my sentiment analysis from Excel to a database system using Delphi/Pascal.

Now, I'd like to share some of the results.  I started earlier this year collecting info on about 10 indicators including various put/call, vxx volume and skew data. Looking over the results in the new system, none of the indicators worked consistently alone so I decided to combine several as standardized variables to produce a composite. So I will show the results in two charts.

The first chart is a backtest from Feb to mid-Aug 2015 in a low volatility period. Sentiment composite (gray) is left axis, SPX close prices (black) is on right. More important are the EMAs of composite with 5 day (green), 10 day (red) and 20 day (blue). The horizontal lines are + and - 1 stddev from mean (sample varies from pop mean of 0 from 7/2010). + 1 std dev (lime) is oversold extreme (buy), - 1 std dev (pink) is  overbought extreme (sell). During this period three sells were generated when the 5 day ema reached - 1 std dev, and two buy signals were generated at the + 1  std dev line. Somewhat better on the sells, so I will probably be looking for a better buy composite.

The second chart is longer term from mid-2014 with the 1st chart period cut out to compare the last half of 2014 to today. Here the two rallies in 2014 after volatility spikes flattened out when the 5 day ema (green) reached - 1 std dev, then continued upward in a choppy fashion until the 10 day ema (red) reaches - 1 std dev and finally  stalls out and begins a decline. The current indicators are at an almost identical position to 11/25/2014, two weeks before the Dec top and a 100 pt SPX decline. This is more interesting considering the fractal from Mar-Oct 2014 where 2015 is leading by 6 weeks.  July top preceded Sept 2014 top by 6 weeks, late Aug crash preceded Oct 2014 crash by 6 weeks, and last week in Oct precedes Dec 2014 top by six weeks. So a choppy market into FOMC top Oct 27/28 may be good time to short.

No comments:

Post a Comment