The choice of what scale to use on your charts makes a big difference to how trend lines sit. Which scale is correct?

The chart above shows Gold 2 weekly on an arithmetic scale. Notice the bear market trend line has been breached, but did not show where price exactly found support and resistance in the process.

The chart above shows Gold 2 weekly on a semi-logarithmic scale. So far price remains below the bear market trend line.

An arithmetic scale is best used for short term price movements. But for long term movements a semi-logarithmic scale is more correct, particularly for markets like Gold which can exhibit blow off tops and selling climaxes.

Any long term movement a year or more should always use a semi-logarithmic or ratio scale.

From Magee (“Technical Analysis of Stock Trends”, 9th edition, page 11):

“Our own experience indicates that the semilogarithmic scale has definite advantages in this work; most of the charts reproduced in this book employ it… Percentage relations, it goes without saying, are important in trading in securities… certain trend lines develop more advantageously on the ratio scale.”

From Pring (“Technical Analysis Explained”, 4th edition, page 68):

“Arithmetic scaling is not a good choice for long-term price movements, since a rise from 2 to 4 represents a doubling of the price, whereas a rise from 20 to 22 represents only a 10 percent increase… For this reason long term price movements should be plotted on a ratio or logarithmic scale. The choice of scale does not materially affect daily charts, in which price movements are relatively small in a proportionate sense. For periods over 1 year, in which the fluctuations are much larger, I always prefer to use a ratio scale”.

*Published @ 04:28 p.m. EST.*

Lara, thanks for this post.

The question that I still debate within my own mind (yes I hear voices lol!) is whether or not the arithmetic scale breakout of such a long held trendline has a significant affect on social mood to affect the overall outcome? i.e. will price also breakout of the simi-log scale trendline within a short period of time?

We will know very soon if price will breakout of the simi-log scale trendline, but if the breakout does occur, that may still not answer the question of whether the breakout of the arithmetic trendline predicted or foretold the breakout of the simi-log trendline.

Comments welcome…

I’ve wondered that too.

Because the arithmetic scale trend line would always come before the semi-log scale trend line, we shall never be able to answer the question.

But looking at how price behaved about the arithmetic scale trend line, it didn’t appear to be where price was finding resistance and support during a consolidation. The line was overshot and undershot repeatedly, it appears to be in the middle of a zone of resistance and support.

For the trend line to be working properly price should react at or very close indeed to the line.