RSI is not a buy signal at 30 or a sell signal at 70. In honest practice, RSI is a context filter — a number that describes recent momentum relative to recent range — logged alongside your setup rules on NSE charts, never replacing them. If RSI changes nothing about your enter/skip decision across twenty replay sessions, remove it from your chart.
The oversold folklore kills more practice sessions than bad entries. Traders see RSI touch 28, call it "oversold," and enter mean-reversion trades in a trend day — then rewrite the story after the stop hits. Practice means freezing what RSI is allowed to do before you load a random historical date.
Myth inventory (what to stop practising)
| Myth | Practice reality |
|---|---|
| RSI below 30 = buy | Trends can sit "oversold" for many bars on NSE momentum days |
| RSI above 70 = sell | Breakout continuations often stay elevated |
| Divergence always reverses | Divergence without structure invalidation is noise |
| Default 14 works everywhere | Period is a parameter — frozen or absent, never tweaked mid-loss |
| RSI replaces stop | Invalidation is always a price level |
Burn the table into your drill instructions. If your log says "RSI oversold entry" without a setup tag, the session is invalid for review.
RSI as context filter: allowed jobs
Pick at most two for a 20-decision block:
Job A — Trend filter: In identified uptrend, skip long entries when RSI below X (you define X) only if your written rule says momentum exhaustion matters for your setup.
Job B — Range filter: In identified range, tag RSI extremes as context for fade setups — still require level + rejection bar.
Job C — No-trade zone: When RSI mid-band (e.g., 45–55), tag RSI-neutral and apply stricter skip rules for breakout entries.
RSI does not get Job D: "override setup because RSI looks extreme." That is folklore with steps.
Write the RSI rule in one sentence
Examples (yours may differ — freeze one):
- "Long pullback entries only if RSI(14) on 15-min is above 40 during uptrend tag."
- "No fade entries when RSI(14) above 60 and benchmark Nifty trend tag is up."
- "RSI logged but ignored for decisions this month."
If you cannot write one sentence, RSI is decoration. Remove it until you can.
The RSI context log (fields)
At each decision point before advancing candles:
setup-tag: pullback-long
trend-tag: up | down | range | unclear
RSI-value: (number)
RSI-job: A | B | C | ignored
RSI-permission: enter-ok | skip-filter | neutral
decision: enter | skip
price-invalidation: (level)
Review weekly: how often did RSI-permission differ from what you would have done with RSI hidden? If never, drop RSI. If often, examine whether skips improved rule adherence — not P&L.
Failure modes in RSI practice
- Parameter fishing — RSI(9) after RSI(14) disappoints. Tag
param-drift. - Timeframe mismatch — 5-min entry, daily RSI story told afterward.
- Trend day massacre — fading "oversold" RSI on NSE opening trend leg.
- Divergence without swings — drawing divergence between redrawn lows.
- RSI as regret button — checking RSI only after missing a move to justify chase entry.
Practise tagging folklore-entry when you catch yourself using 30/70 language without a written filter rule.
20-decision RSI drill
- Random NSE date, liquid name, frozen timeframe.
- Trend tag required before RSI read — if trend unclear, default skip unless setup allows range mode.
- Twenty decisions with RSI log fields complete.
- Optional week 2: same setup, RSI hidden (cover indicator). Compare process scores only.
Indian intraday note: opening 15 minutes produce violent RSI swings on 5-min charts. Many practitioners ignore oscillator reads until 9:45. If that is your rule, write it — "no RSI reads before 9:45 IST."
RSI + structure (not RSI alone)
Valid practice chain:
Structure candidate → trend tag → RSI filter → enter/skip → stop at price
Invalid chain:
RSI touched 30 → enter → structure searched afterward
Replay exposes invalid chains fast when you log in order.
When RSI practice is not worth your month
If your setup is purely level-based (session high break, VWAP reclaim), RSI may add hesitation without filtering bad trades. The minimal stack challenge applies: run with and without, measure violations, decide.
No guru claims that RSI "works on NSE." Your private log is the only evidence you are allowed to cite to yourself.
Conceptual clarity without fabricated edge
RSI measures recent up-close vs down-close momentum. It lags. It scales between 0 and 100. That is the entire technical definition you need for practice. Everything else is implementation you must test — or ignore.
Pairing RSI practice with trend tags
RSI readings without a trend tag are orphan data. Spend one week tagging trend first (up, down, range, unclear) with RSI hidden; spend the next week adding RSI filter. If skip quality does not change, RSI is dead weight on your chart for that setup class — remove it and reclaim attention for structure.
Practise RSI as a context filter on historical NSE charts — not as oversold folklore. Join the waitlist.
Related reading
- The 300-Trade Experiment: How Much Can You Learn From 30 Days of Trading Practice?
- What Is Chart Replay? Backtesting vs Paper Trading
- Pullback vs Breakout Trading: Which Should You Practise First?
- How Many Trades Do You Need to Test a Trading Strategy?
Educational note
Not investment advice. Not stock tips. Practice results do not guarantee future results.