Combine Trendiness and Volatility

Discussion in 'Technical Analysis' started by BENG, May 11, 2006.

  1. BENG


    I know some ETers have developed their own trendiness indicators. For example, acrary knows how to measure the trendiness of any market based on the number of days a trend persist (I hope I understand him correctly).

    I wonder if anyone knows how to combine trendiness and volatility.

    For example, a 10 days bullish trend can go up smoothly, or in a pretty volatile manner.

    Any input on how to approach this measurement? Or somehow, the way you measure trendiness has filtered out the volatile movement? Hence you would only trade the trendy market without volatile action.
  2. man


    try dividing the return of the market by the standard deviation of daily returns. if you annualise both you can abstract from the trend length.
  3. BENG


    Thanks man. When you mean return of the market, do you mean the moving 10 days return in my example?

    I think I should restate my question:

    Is there ONE measurement to rank the markets by the 10 days trendiness and how volatile within that 10 days?

    Imagine if I know the 10 days will be very trendy and smooth, I can use a tighter trailing stop. If the 10 days will be very trendy, but very choppy before reaching the 10 days holding period, I need a looser stop to capture the whole 10 days trend.

    Of course, 10 days is just an example, it can be 3 days, 5 days, etc.
  4. agpilot


    Hello BENG

    TC2005 Worden Bros software has a volatility sort function. It sez that it determines hi-lo range for each of the last 13 weeks and then averages those 13 weeks. It then sorts all stocks in your list. You would have to generate a personal sort formula to select both volatility AND percent gain in the watchlist you are looking at. I take this volatility into account on most sorting because I don't care for a wild ride even tho the end of month P&L may be the same. Most of the time I prefer to do a visual sort of those stocks I make my buys from. It's quite easy to spot those stocks with a wild daily swings if you have a preselected style chart set up just for that volatility sort. The vertical scale should have a common percent change when doing a visual. hope this helps.
  5. cnms2


    If I understood you correctly, you were asking how to determine the volatility beforehand. You can't.

    On the other hand you'll see that price movements in the same direction with the next higher time frame (thrusts) usually exhibits less volatility / price rate of change, than retracements.

    Hence it's better to "trade with the trend", as the conventional wisdom says. With the same stop size you'll get better price movement.
  6. Trending versus range bound days are impossible to predict, but easy to anticipate. Simple volatility mean reversion tends to hold over all periods.

    The way I do it is compare today's pivot range width (derived from yesterday's data) as a percentage of the 30-day median pivot width. When the pivot range widens to excessive levels, it means that we just had a trending day and to anticipate some consolidation. Vice versa holds true.

    The gray area using this technique is very wide, so the only real signals are at the extremes. It's very often the case that when the pivot range narrows to less than 40% of the 30-median values, the following day contains sizable moves if not a new high value.

    Hence, I am slower to take profit when anicipating a trending day because 1) I can simply make more money and 2) the fact that I am anticipating a trending day means I'll get fewer entry points, so I hold on to those that I get.
  7. BENG


    No, I'm not asking how to determine volatility beforehand, I'm asking how to find the trendiness of a market from the historical data. Imagine from the 2005 data, I know that S&P doesn't trend a lot, and the most trendy it can get from point A to point B would be 3 days and assume the 3 days trendiness value is 1. In the same period, Oil is more trendy, with a 3 days trendiness of 1.7 (just an example).

    I hope I could use the 2005 data to apply on the 2006 trading methods. Or I hope I could use the last quarter's data to apply on the next quarter's trading strategies.

    Now given that both markets have a high enough 3 days trendiness value to trade, the path they finish from point A to point B could be different. S&P could be up one day strong, do nothing the next day, and up another strong to complete the 3 days trend. This is a smooth 3 day trend. While Oil could be up one day strong, gapped down next day and trade lower due to some news, then gap up and run big on the third day due to another news to complete the 3 days trend.

    Both S&P and Oil complete their relative path from A to B, but the way they complete the path are different. One is smooth, one is volatile.

    I already know some ETers can measure trendiness of different markets based on the number of days a trend persist. I'm asking how would you measure the volatility within the 3 days trend from the past data, but not the volatility of the market itself.

    I assume knowing how to measure the 3 days trendiness will help, but I'm open to some other clever methods.
  8. Bollinger Bands..

    If they are narrow, volatility is low.

    If the middle trend line is slanting up it is an uptrend.


    Combine the two and study a hundred charts and look for ways to trade these patterns ;)
  9. cnms2


    Then ... flip a coin! :)
  10. Actually my system is more complicated than that and is automated. I have a coin simulator flip 2 coins at once and display the outcomes within seconds.
    #10     May 11, 2006