RSI and momentum

EN source master · P05-L06 · 30 minutes estimated · needs_review

Offline formative study. No wallet connection, real funds, private keys, signatures, live trade or personal portfolio inputs. Visuals are specifications. This source master remains needs_review; completing the formative exercise does not issue certification.

An oscillator's familiar scale can suggest certainty even when initialization and sampling materially change its value.

Separate changes into gains and losses

The Relative Strength Index (RSI) uses average positive changes and average magnitudes of negative changes [RSI]. RS=average gain/average loss; RSI=100−100/(1+RS). Our original three-change exercise initializes averages with simple arithmetic means. Subsequent values use Wilder recurrence: new average=(2×old average+new change component)/3. This small n teaches mechanics, not a parameter recommendation.

State edge conventions

A zero average loss with positive average gain maps to 100 in this exercise; zero gain with positive loss maps to 0. When both are zero, the ratio is undefined: mark unavailable under our convention, rather than pretending every platform uses the same flat-series treatment.

Interpret the record, not the future

A high RSI describes gains dominating losses in the sampled window. Persistent gains can maintain a high value. Familiar 'overbought' vocabulary does not impose a reversal or identify fundamental value. Changing n, smoothing, price source, interval or seed can change the result. The recurrence retains memory and therefore responds with lag.

Invalidate unsupported narratives

A mismatch between new price highs and oscillator values is descriptive divergence, not proof of exhaustion. Range-like and persistent-direction regimes can produce different oscillator distributions. A false positive occurs when a claimed turning-point narrative lacks later confirmation; no rate is estimated here. Corrected closes or a different seed require recalculation. Preserve the exact initialization and reject a directional recommendation.

• Gain/loss component: positive change or magnitude of negative change, otherwise zero.
• RS: ratio of smoothed average gain to loss.
• Wilder smoothing: recurrence retaining prior average.
• Initialization: first averages and edge conventions.

ILLUSTRATIVE closes T1–T5=100,103,102,106,104. n=3 changes, initial averages at T4 use T1→T2, T2→T3, T3→T4. T5 uses recurrence; precision retained until final two-decimal display.

Price-point and RSI panels; show gain/loss decomposition and first availability at T4. Caption n=3 teaching seed, no reversal line. Open/high/low absent; do not invent candles.

The fixture's gain/loss decomposition and stated RSI calculation.

Future direction, universal platform agreement or a turning-point condition.

• OBSERVED: FIX-P05-L06 lists change T3→T4=+4.
• INFERRED: seeded T4 RSI=87.5 under the declared formula.
• UNKNOWN: next close and its effect.
• INSUFFICIENT EVIDENCE: RSI 87.5 guarantees reversal.

• Averaging only winning changes over their count.
• Using signed negative losses in RS.
• Hiding initialization or zero/zero conventions.

Compute T4 average gain, average loss and RSI, then T5 recurrence and RSI. Explain the both-zero case and rewrite 'RSI is high, so reversal is certain'.

Deliver calculations or annotations, claim/source table and limitations. Suggested allocation: study 12 minutes, exercise 8, correction/quiz 10; estimate subject to calibration.

• Use n changes, requiring n+1 closes.
• Record seed and recurrence.
• Use nonnegative loss magnitude.
• Keep level distinct from prediction.

RSI is reproducible when changes, initialization, smoothing and edge conventions are explicit; its level is not a forecast.

• Compute an explicitly initialized RSI.
• Explain why oscillator levels do not require reversals.

P05-L07 after correction review.

NONE in the authoritative catalog. The supplied offline fixture/package is sufficient; no paid feature or unverified Production capability is required. Lab/certification metadata denotes downstream associations, not access gates or live awards.

• [RSI] TradingView — RSI — RSI ratio and Wilder smoothing; no reversal guarantee. Checked 2026-10-01; locator turn10view2.