face down friday, page-38

ge

I think on daily charts the scale tends to not make much difference and, on any time scale, most indicators are not affected.

I think where the semi-log scale might get tricky is with line studies such as Gann Fans, Andrew's Pitchfork etc. where the lines being drawn are not semi-logarithmic but the chart scale is. Standard Fibonacci retracements and projections are fine in any scale as they use price only to plot a horizontal line.

I find I get a better perspective of the market if I look at a weekly or monthly in semi-log scale because it doesn't visually distort price moves over these longer time periods: i.e.

If a stock makes two big 50% upwards moves over a couple of years, the semi-log scale makes each 50% price rise visually look the same on the chart (whether the rise is from $10 to $15 or later on on the same chart the rise is from $40 to $60.)

A standard linear scale will make the $40 to $60 move look much bigger on the chart because it is a $20 move as opposed to a $5 move. Both are proportionally the same though - 50%.

The differences between linear and semi-log tend to disappear when timeframes are shortened.

You have noticed the difference between the trendlines on my semi-log XJO chart versus the linear XJO chart - the linear one has the long-term trendline currently at 5269. That's 1582 points below the November High or a pullback of in excess of 23%. Thats a very, very long way to fall just to get back to the trendline.

So, the linear versus semi-log debate makes a big difference in longer timeframes - semi-log trendlines would tell you that at 5269 the XJO is well and truly in a downtrend, while linear would tell you it is just reverting to the trend at 5269.

How's that for a long, boring rant!

- HBG

---

Trade the market you see - not the market you hope for!