Jumping Those Hurdles
The Chartmill Value Indicator by Dirk Vandycke
Oscillators identify overvalued and undervalued situations.
They have their advantages, especially in short-term valuations,
but also struggle with recurring disadvantages. In this,
the first of a series, we will look at how to overcome these
Any technical analyst is familiar with the concept of an
oscillator. Oscillators like relative strength index (RSI),
moving average convergence/divergence (MACD), and
stochastic try to capture a relative short-term valuation of
their underlying series. Almost all classical oscillators fall into
two categories. On the one hand, there are range compression
oscillators like RSI, which basically try to squeeze a price chart into a fixed range like [0, 100] or [‑1, +1]. Smoothing
oscillators like MACD, on the other hand, use moving averages
to get rid of noise. Neither set of these classic oscillators
interprets the relative value of a stock, which accounts for
almost all of their shortcomings.
How do you cope?
Not only that, range compression oscillators just give you
the same information you see if you squint at the original
price chart. They also have a stickiness problem, which is
the tendency to stick to the minimum or maximum of their
interval during strong trends, as they try to fit trends of any
length into the same narrow range. Those trends almost always
get compressed in the small oversold or overbought zones.
That is because the biggest part of their range is reserved for short-term reactions to price. This gives extended overbought
or oversold signals with the trend just marching on, making
those signals useless. Smoothing oscillators on their part have
the same lagging problems as moving averages, the primary
components on which they are built.
To make things even worse, they all need parameters, which
leaves them open to a lot of subjectivity in their usage and
interpretation. This puts traders who are backtesting these
oscillators in harmís way, as they might fall victim to curvefitting.
This lack of transparency can be seen in the numerous
ways they are used, while in fact they are often nothing more
than a small, distorted version of the original price chart.
This leaves us with the question: Can we cope with the lag,
stickiness, and subjectivity to come up with a better oscillator?
Yes. The solution to this problem of building such an oscillator
lies in the statistics of what short-term value is.