Level 2 Patterns & indicators · Part 6/9
Oscillators: RSI and Stochastics
Oscillators measure speed rather than direction. They fluctuate between fixed bounds – hence the name – and are therefore misread with great regularity.
How the RSI computes
The relative strength index compares the average up moves of the last n periods to the average down moves and normalises the result to a 0–100 scale. The standard is n = 14.
What follows matters: an RSI of 80 means up moves clearly dominated down moves over the last 14 periods. It does not mean price is too high. It contains no valuation information whatsoever.
The “overbought” misconception
The 70 and 30 levels are conventionally labelled overbought and oversold. Those words are the origin of the single most common mistake in all of indicator practice.
In a strong uptrend the RSI stays above 70 for weeks. Selling at 70 means selling at the start of the trend. Conversely, in a downtrend it stays below 30 for just as long. A high RSI is a sign of strength, not of excess.
The levels are meaningful only in a clearly sideways phase – which is exactly where the question matters less anyway. That is not a flaw of the indicator but of how it is usually described.
Divergences
The sturdier application: price and oscillator move apart. If price makes a higher high while the RSI makes a lower one, the new move is carried by less momentum than the previous one.
Two caveats that rarely get mentioned:
- Divergences can persist for a long time. They are not timing. In a strong trend several occur in a row with nothing happening.
- Divergences are a matter of selection. Which highs you compare is judgement. Look long enough and you will find one in any chart.
Useful as an observation, unusable as a signal – that roughly sums up the state of the evidence.
Stochastics
The stochastic oscillator measures something different from the RSI: where the close sits within the range of the last n periods. A close at the top of the range puts the value near 100.
It reacts faster than the RSI and is correspondingly noisier. The usual smoothing (%K and %D) is an attempt to claw that back – and it is the same trade as between SMA and EMA.
What oscillators are actually good for
- As a state description: “up moves have dominated for two weeks” is a legitimate, checkable statement.
- As a filter: combined with a trend filter, to find entries in the trend direction after a pullback.
- Not as a reversal signal in trends. That is the application they are famous for, and the one in which they systematically fail.
Next
MACD and ADX measure change in momentum and trend strength instead – the next article.