Help measuring "trendiness"

Discussion in 'Strategy Development' started by Candace, Feb 20, 2010.

  1. Candace


    I want to look at different stocks/markets to see how frequently tradeable intraday trends were present, and what market conditions were like at the time (bull/bear/sideways and volatility). I have taken a stab at an approach but I often sit starry eyed looking at the quality minds posting here so thought I would ask for some input. Below is a very basic approach, if you have suggestions for improvement or another approach altogether, I'm all ears.

    I have a spreadsheet that tallies HOD and LOD in 30 minute periods. I have assumed that a "smile" (more highs/lows at the beginning/end of the day) means that there were some intraday trends present.

    I calculate a ratio that is the sum of the HOD/LOD tallies for 3pm to 4pm plus the tally for 9:30am to 10:30am divided by the tally for 10:30am to 3pm. I assume that the higher the ratio, the better the trend. For the stock/period I looked at the ratio was 2.5 for HOD's (big smile) and 0.7 for LOD's (little smile). By comparison, if highs or lows happened to be evenly distributed throughout the day the ratio would be 0.4 (2hrs/4.5hrs).

    I plan to plot these ratios in a MA format (20, 50, 200day) against an SMA for price and ATR.

    Even as I type this I wonder if this wouldn't be better done with some best fit line and std devs. Anyone have suggestions (remembering that I only have Excel available)?
  2. didn't you figure out that stocks mean revert in short timeframes?
  3. Candace


    Trend days certainly seem to be more rare than range days but I was hoping to try to quantify this. 1 in 8? 1 in 12? Thinking of giving up my beloved but risky rtm ways and convert to trend following.