Back Testing Methodology

Back testing is performed against data for breakouts from our CwH (cup-with-handle) watchlist for the number of days you specify up to a maximum of 365 calendar days (1 year). The default is 90 days (13 weeks) which provides a reasonable estimate of returns that can be expected under current market conditions for confirmed breakouts that meet your filter criteria.

To perform the backtest, firstly, all breakouts are summarized by their performance on the day of breakout and then by their subsequent performance. The statistics reported are:

  • No. of Breakouts - the total number of confirmed breakouts over the last 3 months. This is a rolling figure and will vary each day.
  • Average Gain at B/O day close - this is the average gain over breakout price on the day the breakout was confirmed.
  • Average Max. B/O Day gain - this is the average maximum gain, calculated using the the high on the breakout day, over the breakout price on the day of the breakout.
  • No. of Failures - the number of stocks that failed to rise at least 5% before dropping to -8% of the breakout price
  • Avg. Max Gain since B/O - the average maximum gain, using the highest intraday high, achieved over breakout price, since the breakout. If a breakout rose past 5% and then subsequently fell below 8% from the pivot price, then the highest high reached before the collapse is used.
  • These statistics are reported on the first line of the report

Secondly, the same statistics are obtained after applying your filters to the same 3 months of historical data and reported on the second line of the report.

The third line of the report shows the variance as the arithmetic difference between the two.

The fourth line shows the variance as a % improvement of your filter performance against all breakouts.

A positive variance on the percentage gain numbers means that your filters are selecting breakouts that perform better than the overall averages. Through using very selective filters, it is possible to get substantial improvement over the averages but this usually comes at the cost of substantially reducing the number of breakouts that you will be alerted to.

The statistics for the breakout day and subsequent performance are provided so you can adjust the filters to meet your own trading style, and/or to match prevailing market conditions.

© 2005 breakoutwatch.com