New Members: Be sure to confirm your email address by clicking on the link that was sent to your email inbox. You will not be able to post messages until you click that link.

RSI divergence scan help

Hi,

I am trying yo build a weekly RSI divergence scanner. This is what i have so far. It didn't produce any search results so far. Also would like to know how to implement this on the daily monthly hourly and 30minute time lines if possible.

Thanks for your help.

[weekly RSI(14) > last week's weekly RSI(14)]
and [weekly close < last week's weekly close]

Comments

  • Weekly scans are best run after the close Friday and before Monday's open. Otherwise, if your run it, say Tuesday, the current week's data is only as of Tuesday at the time you run the scan. That's not a full week, so you are comparing apples to oranges (partial week data to full week data). Same for monthly scans run on any day other than the last day of the month after the market close and before the next open. Daily scans should be run after the market close. You can't scan for hourly and thirty minute data. You can run daily scans during the day, but they use daily data for past days and they treat the most recent update (say 1:12 pm eastern) as end of the current day even thought the market isn't closed.

    To test whether your weekly scan above is producing accurate results, run just the first line and see if there are any that closed down. Use a weekly chart style on the results to verify.

    You can test prior weeks by setting the calendar back to prior Fridays.
  • i tried this scan yesterday, and it didn't produce any results. Back tested to couple fridays and mondays with no results.
  • I tried it a few different random weeks going back a year and didn't get any hits either.

    Nothing wrong with the scan though. It looks like it's just a rare event. Most of the time, the RSI(14) dips with the price.

    You could try re-writing to skip a few weeks and see what happens - like this week's close < 3 weeks ago and RSI > 3 weeks ago.
  • Is something like this a good idea? I get hits:


    and [Weekly PctChange(2,weekly RSI(14)) > 0]
    and [Weekly PctChange(2,weekly close) < 0]

    I was thinking of using Slope too.
  • mark, how would the three weeks ago be implemented in the code? Its a nice idea to increase the duration.
  • Here's one possibility. Try @Kevo 's, too.

    [group is sp500]
    and [weekly RSI(14) >3 weeks ago weekly RSI(14)]
    and [weekly close < 3 weeks ago weekly close]
Sign In or Register to comment.