No registration required! (Why?)

# Not Smooth Enough?

If you wouldn't mind, explain to me how, if at all, market condition criteria factors into determining the exponent of your EMA equations please. Insofar as I'm aware in the technical analysis field there isn't a consenus as to how much weight is given to smooth data in differing timeframes or even market conditions.

I, like many of you, use Brown's simpliest weight of 1 added to an SMA, divided by 2, &c., and it works well for me. I'm just curious if any of you that have experimented using differing weights have found an edge in doing so, and if so, if you'ld consider writing a little bit about your findings.

Thanks.

I, like many of you, use Brown's simpliest weight of 1 added to an SMA, divided by 2, &c., and it works well for me. I'm just curious if any of you that have experimented using differing weights have found an edge in doing so, and if so, if you'ld consider writing a little bit about your findings.

Thanks.

Today's Exponential Moving Average (or Average True Range)=(current day's closing price (or trading range) x Exponent) + (previous day's EMA (or ATR) x (1-Exponent)).

The exponent is the contstant, and weight taken from the oldest value is an infinite regress, since the exponent is a number below zero, but will always be factored in as a variable in determining the current data. The data is then smoothed as they say always showing the truest function of the current data. You want this, because if plotted on a line chart the difference between an SMA and an EMA is drastic. The SMA is farther away from price action and therefore increases the probability of a) not being useful at all in making trading decisions, unless coupled with one or more EMAs, and b) providing false signals when thought to be useful.

The longer the period used in determining an EMA the fewer the signals, however, of those signals lay greater statistical probability of a winning trade. The reverse is true for shorter timeframes.

But how do you find the exponent? First you require the current data from a previous average, be it an SMA or EMA; you then need to determine how many periods you want to mean weight, then add 1 to the period range, and divide the total by 2. So for a 200 period exponent you'ld have .0099502, and not 100. Then you round to get .01.

The rest is all about personal preference and experimentation.

I've the mind to begin experimenting with additional Fibonacci weighting exponents. I'd bet Jim already has.

You can't go wrong with medium range period averages when first starting out. For an ATR I'd start at 13 or 14.

Using ATRs you can make a mint.

Here are a few of my ATR strategies: for stops I use a 2 ATR multiple (current ATR times 2) or whichever is greater for longs compared to 1/2 percent of the index. For targets, 1%.

You only want to use ATRs on breakouts, ideally. That way while you haven't entirely maximized on capital gains, you're on the right side of the trend and haven't been burned on not picking the right bottom or top. If you have an ATR bouncing back and forth, at say 8 and 9 for a period of time, and after a new day's closing data shows 10, go long at the bottom of the next best availiable retrace at the open. The reverse is true for shorts.

The exponent is the contstant, and weight taken from the oldest value is an infinite regress, since the exponent is a number below zero, but will always be factored in as a variable in determining the current data. The data is then smoothed as they say always showing the truest function of the current data. You want this, because if plotted on a line chart the difference between an SMA and an EMA is drastic. The SMA is farther away from price action and therefore increases the probability of a) not being useful at all in making trading decisions, unless coupled with one or more EMAs, and b) providing false signals when thought to be useful.

The longer the period used in determining an EMA the fewer the signals, however, of those signals lay greater statistical probability of a winning trade. The reverse is true for shorter timeframes.

But how do you find the exponent? First you require the current data from a previous average, be it an SMA or EMA; you then need to determine how many periods you want to mean weight, then add 1 to the period range, and divide the total by 2. So for a 200 period exponent you'ld have .0099502, and not 100. Then you round to get .01.

The rest is all about personal preference and experimentation.

I've the mind to begin experimenting with additional Fibonacci weighting exponents. I'd bet Jim already has.

You can't go wrong with medium range period averages when first starting out. For an ATR I'd start at 13 or 14.

Using ATRs you can make a mint.

Here are a few of my ATR strategies: for stops I use a 2 ATR multiple (current ATR times 2) or whichever is greater for longs compared to 1/2 percent of the index. For targets, 1%.

You only want to use ATRs on breakouts, ideally. That way while you haven't entirely maximized on capital gains, you're on the right side of the trend and haven't been burned on not picking the right bottom or top. If you have an ATR bouncing back and forth, at say 8 and 9 for a period of time, and after a new day's closing data shows 10, go long at the bottom of the next best availiable retrace at the open. The reverse is true for shorts.

Copyright © 2004-2018, MyPivots. All rights reserved.