Backtesting

Backtesting is the process of applying a stock screener to stock data from past markets (e.g. 6 months ago) and then determining the performance of the screener through a given time period (e.g. calculate gain through today). Backtesting is extremely important to the design of customized screeners because it allows you to see how your screener would have fared under past market conditions. Before we discuss each backtesting parameter in detail, let's look at an example. The screenshot in the previous section shows a screener set to use data from six months ago and to compute the gains through today

The results of this screener look like the results from any other screener, with one exception: a table is provided at the top of the report cards which compares the gain (or loss) of the dow, s&p 500, and nasdaq with the average price gain of the screening results over the period which you specified in the backtesting form.

The summary interpretation of these results is as follows: if you had purchased and held shares in a dow, nasdaq, or s&p 500 index fund six months ago, you would have gained 2.61%, 12.64% and 5.71%, respectively. If, on the other hand, you had purchased equal dollar amounts of the results of such a screener six months ago, you would have gained 10.72%. Note that this is a pure 'buy and hold' result. It does not take in account any advantage that you would gain by automatically managing buys and sells using the autoinvestor.

Now that we've established some context for discussing the backtesting parameters, let's discuss the settings:

Backtesting setting 1: use stock data from

The 'use stock data from' setting establishes the date from which the screening data is drawn. The default setting will use data from today (no backtesting). Other convenient presets are provided such as 1 week ago, 1 month ago, etc. Selecting 'custom' will enable the date selection widget to the right of the pulldown so that you can set an arbitrary date.

Backtesting setting 2: calculate gain through

The 'calculate gain through' setting determines the period for gain calculation. Gains will be calculated between the 'use stock data from' date and the 'calculate gain through' date. This setting will have no effect on the screening results (it will only affect gains displayed in the gain table). The screening results are only influenced by the 'use stock data from' date.

Backtesting setting 3: save as sliding/fixed date

When you save a stock screener, you can run the same screener today, tomorrow or a year from tomorrow. If you run a saved screener on a future date, then we need to know what date range you intended to save when you set up the backtesting.

The most common option is a sliding date. This means that if, for example, you selected 6 months ago for the 'use data from' option, then it is always 6 months ago from the day that you are running the screener. If you run the screener a year from now, then the stock data will come from 6 months from today (which is the same thing as saying 6 months ago from a year from now).

The fixed date option, on the other hand, will store the backtesting settings as fixed dates. If you ran the previous example a year from now with the fixed date option, then the data would be still come from 6 months ago (which means that when you run the screener in a year, the data will be 18 months old). This option is most often used when you are trying to find screeners which perform well over the course of market-impacting events, such as the terrorist attacks on September 11, 2001.

If you set a custom date, this date may not be a valid trading date (the markets may not have been open), so the screener will automatically adjust the custom date for you, if necessary. Also, if you set a custom date and then use the sliding date setting, the screener will calculate the number of days ago that this custom date is from today and the sliding date will always be represented by the number of days ago from the date that the screener is run.

Notes

The notes field gives you a place to type in any notes which are relevant to the current screener settings. Notes will be saved with the screener and displayed on the screener list page.