RSI Extremes and Divergence: Context Before a Reversal Signal

Video: RSI Overbought Doesn’t Mean Sell. Here’s What Does.
Creator: Verified Investing
Original Video: Watch the original video
Difficulty: Intermediate

The Big Idea

RSI above 70 does not compel a short, and RSI below 30 does not compel a long. Strong trends can remain at extreme readings. Divergence asks a more specific question: is price making a new extreme while the momentum measure fails to confirm it?

What RSI Measures

The relative strength index compares smoothed gains with smoothed losses in the same price series. It is not the stock’s performance relative to a market index, and it does not directly measure trading volume or institutional interest.

In its usual formulation, RSI = 100 − 100 ÷ (1 + RS), where RS is the ratio of smoothed gains to smoothed losses. The common fourteen-period setting is the indicator default here; the supplied lesson does not establish it as an optimized parameter.

Readings lie on a zero-to-one-hundred scale. The 30 and 70 references are index levels, not dollar prices or probabilities. The conventional overbought and oversold labels describe momentum conditions; they do not establish that an asset is fundamentally mispriced.

Why Extremes Can Persist

A sustained run of gains can keep the smoothed gain/loss relationship elevated while price continues rising. Selling solely because the oscillator crossed 70 can therefore oppose an ongoing trend. The mirror risk applies to buying a continued decline below 30.

A later pullback does not prove that an earlier short was economically viable. The adverse movement before that pullback, the stop location, and the available holding period all matter. Record the sequence rather than counting any eventual reaction as a successful signal.

The source uses selected charts to illustrate both the persistence problem and several divergences. Those examples teach the distinction; they do not supply a consecutive sample or independently establish a success rate.

Regular Divergence Uses Matched Swings

Compare corresponding points in time. Choosing one pair of price lows and an unrelated pair of oscillator lows can manufacture a divergence. The included code samples RSI at confirmed price-pivot bars rather than independently choosing whichever oscillator turning points fit.

Divergence can occur without RSI crossing 30 or 70. If extremes are an additional requirement, define and test that filter separately. The source’s main warning remains relevant even when an extreme and divergence coexist: a reversal is not guaranteed.

Worked Example: A Clue Is Not a Fill

A simple comparison made after the chart is complete can make the 98.00 low look like an obvious buy. But the subsequent bars used to confirm it did not exist at that moment. If a different early-entry rule is desired, it needs a different information and failure analysis.

Divergence Can Extend Before Price Turns

Price can make another lower low after bullish divergence, sometimes while RSI continues improving. A bearish divergence can likewise appear repeatedly during a continuing advance. The original observation may be genuine while the proposed reversal trade still loses.

That is why the source discusses combining divergence with other chart evidence, including a trend-line break. Decide what constitutes confirmation before seeing the outcome. A line that is only obvious after the reversal cannot authorize an earlier trade.

Possible research definitions include a close beyond a preidentified minor swing, a reclaim of support, or a separately defined failure test. These are proposed definitions to compare, not a complete rule set supplied by the lesson.

Choose the Swing Rule Before Measuring Reliability

A pivot requiring three bars on each side becomes known only after the third subsequent bar finishes. A more responsive one-bar rule changes the set of swings and can create more unstable candidates. Neither parameter is automatically best.

The distance between compared pivots also matters. Comparing points a few bars apart may describe a local hesitation; comparing distant extremes may span different conditions. Record the gap rule and keep it fixed when reviewing later data.

The indicator compares consecutive confirmed pivots of the same type. It does not search through every possible earlier pivot to find the best-looking divergence. That conservative choice makes the output easier to explain, while missing some discretionary comparisons.

Separate the Observation From Risk Management

An oscillator reading does not choose position size. Once an independent entry and invalidation exist, calculate distance from the actual fill, include costs, and check available room before the next obstacle.

A tight stop chosen merely to make an attractive R multiple may test a different idea from a structural stop beyond the relevant low. A wider stop at unchanged size increases exposure. Preserve the denominator used to compare results rather than changing it after a loss.

Keep the Scope of the Evidence Clear

The lesson’s promotional segment concerns unrelated products and is not part of the RSI method. This guide focuses on the supplied momentum concepts without converting a sponsor claim into a recommendation.

For further platform detail, TradingView’s RSI divergence reference (paid link) explains the regular divergence relationship and cautions against acting on it alone. Use that as a definition reference, not proof that a particular trade will work.

Pine Script Indicator

//@version=6
// Guide: https://candlechasers.com/rsi-extremes-divergence-confirmation/
// Independent educational approximation. Confirmed observations, no simulated orders.
indicator("CC RSI Confirmed Divergence", overlay=false)
length = input.int(14, "RSI length", minval=2)
pivotBars = input.int(3, "Price pivot bars each side", minval=1)
minGap = input.int(5, "Minimum bars between pivots", minval=1)
maxGap = input.int(60, "Maximum bars between pivots", minval=2)
if maxGap < minGap
    runtime.error("Maximum pivot gap must be at least the minimum.")
r = ta.rsi(close, length)
ph = ta.pivothigh(high, pivotBars, pivotBars)
pl = ta.pivotlow(low, pivotBars, pivotBars)
var float previousLow = na
var float lowRSI = na
var int lowBar = na
var float previousHigh = na
var float highRSI = na
var int highBar = na
bool bull = false
bool bear = false
if barstate.isconfirmed
    if not na(pl) and not na(r[pivotBars])
        int pivotBar = bar_index - pivotBars
        int gap = not na(lowBar) ? pivotBar - lowBar : 0
        bull := gap >= minGap and gap <= maxGap and pl < previousLow and r[pivotBars] > lowRSI
        previousLow := pl
        lowRSI := r[pivotBars]
        lowBar := pivotBar
    if not na(ph) and not na(r[pivotBars])
        int pivotBar = bar_index - pivotBars
        int gap = not na(highBar) ? pivotBar - highBar : 0
        bear := gap >= minGap and gap <= maxGap and ph > previousHigh and r[pivotBars] < highRSI
        previousHigh := ph
        highRSI := r[pivotBars]
        highBar := pivotBar
plot(r, "RSI", color=color.purple, linewidth=2)
hline(70, "70 context", color=color.gray)
hline(50, "50 midpoint", color=color.new(color.gray, 60))
hline(30, "30 context", color=color.gray)
plotshape(bull, title="Bull divergence known now", style=shape.triangleup, location=location.bottom, color=color.lime)
plotshape(bear, title="Bear divergence known now", style=shape.triangledown, location=location.top, color=color.red)
plotshape(bull, title="Bull confirmation on price chart", style=shape.triangleup, location=location.belowbar, color=color.lime, force_overlay=true)
plotshape(bear, title="Bear confirmation on price chart", style=shape.triangledown, location=location.abovebar, color=color.red, force_overlay=true)
alertcondition(bull, "Confirmed bullish divergence", "CC: regular bullish RSI divergence becomes known now on {{ticker}} {{interval}}.")
alertcondition(bear, "Confirmed bearish divergence", "CC: regular bearish RSI divergence becomes known now on {{ticker}} {{interval}}.")
var table status = table.new(position.top_right, 1, 2, bgcolor=color.new(color.black, 15))
if barstate.islast
    table.cell(status, 0, 0, "RSI: " + str.tostring(r, "#.##"), text_color=color.white)
    table.cell(status, 0, 1, "Divergence marker delay: " + str.tostring(pivotBars) + " bars", text_color=color.white)

What This Indicator Shows

The indicator always displays RSI and the 30/50/70 references in its own pane. It compares successive confirmed price lows and highs, sampling RSI on those same pivot bars. By default, pivots need three bars on each side and must be five to sixty bars apart.

Divergence markers appear on the confirmation bar in the oscillator pane and on the price chart, with no negative offset. They are not moved back to the earlier swing. The RSI line can develop while the live candle is open; divergence decisions wait for a confirmed chart bar.

What This Indicator Does Not Show

It does not create a separate price-action entry, stop, or target and does not require an extreme RSI reading for every divergence. It omits hidden divergence, independent oscillator-pivot matching, and a search across nonconsecutive pivots. It reports observations rather than simulated trades or a success rate.

Setup, Alerts, and Validation

1. Add: paste the complete script into Pine Editor, save, and add it to a standard-candlestick chart.
2. Configure: follow this guide’s timeframe, direction, and session requirements. Load enough history for the longest lookback and pivot delay.
3. Inspect: check the reference plots and status table. A strict rule can correctly produce no signals; a visible line or status helps distinguish that from missing data.
4. Alert: select the relevant named condition and Once Per Bar Close. Recreate alerts after changing inputs.

Validation status: formulas, historical references, state rules, and HTML/code preservation were reviewed statically. The indicator has not been compiled or chart-tested in TradingView here. Live alerts, browser copying, and responsive rendering have not been tested. No simulated orders, fills, or measured performance results are claimed.

Practice and Research

1. Threshold versus divergence: collect all first 70/30 crossings and all confirmed divergences separately. Compare later price paths over the same predefined horizon.
2. Confirmation delay: compare pivot strengths of two, three, and five without backdating any marker. Record how available entry price changes.
3. Added trigger: apply one predeclared price confirmation to the divergence sample. Compare lost opportunities, adverse movement, and reward/risk before claiming it improves results.

Use later dates to review the frozen rules. Include divergences that extend and never generate a useful reversal.

Check Your Understanding

Does above 70 mean sell? No. Strong trends can retain extreme readings.

What makes a regular bullish divergence? A lower price low and higher RSI reading at the corresponding comparison points.

When is a three-right-bar pivot available? After those subsequent bars close, not at the historical low itself.

Final Takeaway

Use RSI extremes as context and divergence as a defined comparison. Require timely price evidence and an explicit risk plan before turning the observation into a trade hypothesis.

Risk Disclosure

This guide is for education and research, not personalized financial advice or a recommendation to trade any security or contract. Trading can cause substantial losses, and leveraged products may produce losses beyond the amount initially committed. Short selling can involve unlimited loss. Stops and displayed prices do not guarantee execution; gaps, spread, slippage, fees, liquidity, halts, and platform interruptions can materially affect outcomes. Historical charts, indicators, hypothetical calculations, and creator-reported results do not establish future performance. Understand the instrument, execution assumptions, and total exposure, practice before risking capital, and use only risk capital you can afford to lose.

CandleChasers provides independent educational analysis and is not affiliated with, endorsed by, or sponsored by the source creator or any platform mentioned.