RSI is taught with two lines and one instruction: below 30 is oversold, buy it; above 70 is overbought, avoid it.
On PSX, over 10.6 years, that instruction is backwards.
| RSI band | sample | touched +20% | vs the market |
|---|---|---|---|
| Below 30, "oversold" | 2,370 / 1,351 | 23.8% / 31.5% | 1.36× / 1.09× |
| 30 to 50 | 16,687 / 12,872 | 15.1% / 27.5% | 0.86× / 0.95× |
| 50 to 70 | 14,466 / 13,798 | 17.8% / 28.3% | 1.01× / 0.98× |
| Above 70, "overbought" | 2,527 / 3,027 | 26.8% / 35.3% | 1.52× / 1.23× |
| Above 80, "very overbought" | 582 / 795 | 28.2% / 38.9% | 1.60× / 1.35× |
The single best RSI band for reaching +20% was above 80. The reading that every course tells you means "get out". It beat the market by 1.60× and 1.35×, in both periods.
And "oversold" decayed badly: a respectable 1.36× in the design years fell to 1.09× in the held-out years, barely above random.
Notice also the dead middle: RSI between 30 and 50 measured 0.86× and 0.95%, the worst zone on the table. That is 29,559 observations saying the boring, mid-range, "healthy pullback" stock is the least likely place for a large move to start.
Why the textbook is inverted here
RSI above 80 means the stock has been going up, hard, for weeks. On PSX, where a single operator can control the float of a small company and the exchange caps daily moves at roughly 10%, that persistence is often the campaign itself, not its exhaustion. High RSI is a symptom of the thing you were looking for.
The part you must not skip
This lesson is about the probability of touching +20%, and only that. It is not a recommendation to buy overbought stocks and hold them. The same measurement set shows the median 20-session closing return for RSI-above-80 stocks was −1.76% in the design years and +0.14% in the held-out years, roughly nothing, while the touch rate was the highest on the board.
Both facts are true at once: these stocks reach high prices unusually often and end around where they started. That gap is not a paradox. It is the entire reason ZEME teaches a resting sell order at your target. The move is real; keeping it requires the order to already be there.
And a hard rule that overrides all of the above: ZEME does not enter after a same-day circuit lock, and applies a 3-5 day cooling period. A high RSI reading is not a licence to chase a stock that has already locked today.
Do this now, practical activity
- Open SuperCharts, add RSI and set the range to 1 year.
- Mark every day RSI crossed above 70. For each, note the price then and the highest price over the following month.
- Do the same for every day RSI fell below 30.
- Compare your two lists.
What to watch for in your own reaction: if the RSI>70 list looks better and you still feel the RSI<30 entries were "safer", you have just found the gap between what you measured and what you believe. That gap costs money, and noticing it is the whole point of this Academy.
Measured: RSI above 80 touched +20% within 30 sessions in 28.2% of 582 cases (1.60×) and 38.9% of 795 held-out cases (1.35×). The best RSI band in both periods. RSI below 30 measured 1.36× then decayed to 1.09×.