Testing relative-strength-index reversal rules against trend continuation

How the forecast is built

The relative-strength-index is formed by averaging up closing-price net changes and down closing-price net changes over a lookback, taking their ratio, and mapping that ratio onto a scale that runs from 0 to 100.

A 14-bar first reading divides the up average and the down average each by 14. Later bars update each average by multiplying the prior average by 13, adding the latest up or down close, and dividing by 14.

A chart overlay, not a stay-in-the-market routine

The measure was presented as a chart overlay meant to mark recent highs and lows shortly after they form, not as a mechanical routine that stays continuously in the market.

What can be coded as a unique test

A discretionary divergence-setup waits for price to extend a prior extreme while the oscillator fails to confirm, then withholds action until the oscillator breaks the intervening swing. That sequence is hard to encode as a unique historical test.

Threshold use as an overbought-oversold reading is readily coded, and that coding typically places a counter-trend trade. An elevated zone is treated as a coming top and a depressed zone as a coming bottom.

Three procedures on the same reading

One tested short procedure used a change-of-direction-rule. It required the oscillator to trade above an elevated parameter illustrated at 70, then turn down for one bar, with entry on the next open and an n-bar high/low channel trailing-stop.

A second tested short procedure used a back-through-rule. It required the oscillator to first exceed that same elevated parameter and then recross it from above before selling on the next open, again with an n-bar high/low channel trailing-stop.

A third momentum-strategy used an always-in-continuation-rule. It bought when the oscillator reached an elevated reading and sold when it reached a depressed reading, remaining always in the market and using no stop.

The evaluation compared lookbacks of 9, 14, and 19 bars across the three procedures.

What the write-up concluded

The write-up concluded that an extreme oscillator reading more often marked further continuation than an imminent reversal, and that the oscillator therefore identified trends more consistently than overbought or oversold turning points.