Pairing stochastic divergence with trendline invalidation

What counts as a split

A divergence is a non-confirmation in which price prints a new swing extreme while a related average or oscillator refuses the same extreme. It is used as an early-warning condition rather than a standalone order.

A bullish split is defined as price making a new low while the oscillator makes a higher low. A bearish split is defined as price making a higher high while the oscillator makes a lower high.

The same class of non-confirmation

When one price average records a new high and a related average cannot exceed its prior high, the strength of the prevailing primary trend is treated as unconfirmed.

Absent volume or open-interest expansion alongside a price trend is treated as the same class of non-confirmation as an oscillator that fails to ratify a new price extreme.

Screen with a bounded stochastic

A stochastic oscillator is a zero-to-one-hundred momentum reading whose slow %D line is compared with price for the split and whose faster %K cross later times the test.

Bounded oscillators that travel between 0 and 100 supply explicit extreme bands. Unbounded momentum histograms do not, which makes their splits harder to screen.

The overbought-oversold-screen is a requirement that the confirming oscillator already sit in an extreme band as the split begins to form, used to discard marginal patterns. A confirming oscillator is required to sit already in an overbought or oversold band as the split begins to form, so that incomplete or mid-range patterns can be discarded.

Time the test after the cross

On a 14-period slow stochastic, the split is read on the %D line and the hypothesis is timed only after the faster %K line crosses %D.

State where the hypothesis is wrong

After a split completes, invalidation is placed a few ticks beyond the pattern extreme: above the highest high for a bearish case and below the lowest low for a bullish case. Swing-extreme-invalidation is that protective level.

A trendline is a structural line that may break with a divergence, later act as a retest barrier, and supply a location for invalidation when the oscillator split is otherwise clean. A completed split can coincide with a trendline break or a later pullback to that same line, giving a second structure-defined place to test or abandon the hypothesis.

Price-oscillator splits carry no built-in measuring rule, so objectives and later stop-tightening come from other structure such as prior support, resistance, flags, or trendlines.

Sequences and scanner misses

Turning areas may print a sequence of splits rather than one two-swing ideal, so an early completed pattern can be stopped out before a later, cleaner split appears. Automated scanners can also mark false splits and miss obvious ones.