Overbought is not a sell signal
RSI compares recent gains to recent losses and normalises the result onto a 0–100 scale. Because it is bounded, it does something that catches almost everybody:
In a sustained trend, RSI reaches its extreme and stays there.
An instrument in a strong uptrend can hold RSI above 70 for dozens of consecutive bars. The extreme is not a warning that the trend is ending — it is what a strong trend looks like through this indicator.
So the same reading means opposite things depending on regime:
| Regime | RSI above 70 means |
|---|---|
| range-bound | price is near the top of its range — a mean-reversion case |
| trending up | the trend is strong — a continuation case |
Which is why RSI cannot be used without first deciding which regime you are in. A strategy that sells every RSI-above-70 reading is short every strong uptrend, and that is a specific way to lose money consistently rather than randomly.
The 30 and 70 are conventions
Nothing derives them. They are round numbers that became standard, and they are inputs you should expect to change per instrument and per timeframe rather than constants of nature.
The useful way to pick them: plot RSI on the symbol and timeframe you intend to trade, and choose levels that give you the trade frequency you actually want. Then, because you have now tuned something, the count of levels you tried is part of your result — why the count matters.
A regime gate you can actually build
The honest form of an RSI strategy in a form-based builder is oscillator plus trend filter:
| Field | Value | Role |
|---|---|---|
| RSI length | 14 | |
| Oversold | 30 | the entry trigger |
| Trend filter | price above a 200-period moving average | only take longs in an uptrend |
| Exit | RSI back above 50, or an ATR stop |
The filter is what converts "buy oversold" into "buy a pullback in an uptrend", which is a different and more defensible idea. And you can check the filter is doing something: count the trades with it and without it. If the difference is zero, the filter is decoration — the count test.
The divergence caveat
RSI divergence — price making a new high while RSI does not — is the most-cited RSI pattern and the hardest to use in a rule-based strategy, because "a new high" and "a lower RSI peak" both require identifying peaks, which needs state carried across bars. That is squarely in the category a form cannot express, for the reasons set out in what a form cannot express.
If divergence is central to your idea, expect to extend the exported Pine by hand rather than to configure it.
Tactix AI — Studio vs Guide
Tactix AI is AlfaTactix’s product assistant brand (open Tactix AI).
- Tactix Studio turns a one-sentence strategy description into a draft across Timeframe, Signals, Filters, and Risk in the visual Strategy Builder. You review and edit every field before Code Generator writes MQL5 or Pine Script.
- Tactix Guide explains the step you are on — what to fill, what a control means, or how to phrase a rule — without dumping untested source code.
That is form-first automation: the LLM never replaces Code Generator, and you keep plan limits and real-time validation.