A strange thing happens when a technical indicator becomes famous.
The formula survives.
The problem that created it disappears.
RSI is a good example.
Today, many traders first meet the Relative Strength Index as two numbers:
70
30Above 70 is called overbought.
Below 30 is called oversold.
Then the lesson often jumps straight to:
70 → sell?
30 → buy?That shortcut hides the more interesting story.
RSI came from an engineer who had already built apartments, flown between construction projects, made money in silver, lost money in commodities, and then decided that trading needed something more systematic.
His name was J. Welles Wilder Jr.
And the problem he was trying to solve was not simply:
“How can I find a buy signal?”
It was closer to:
“How can I measure the strength and weakness of different markets on one common scale?”
That difference matters.
Understanding it changes the way we read RSI.
1. Before RSI: Oscillators Did Not Speak the Same Language
By the 1970s, traders already had moving averages, momentum ideas, oscillators, chart patterns, and mechanical trading rules.
But Wilder later recalled a practical problem.
Different oscillators could use different numerical scales.
A reading from one commodity could not always be compared intuitively with a reading from another.
A raw momentum number had the same problem.
Market A
Momentum = +4
Market B
Momentum = +4Those two values do not necessarily describe comparable moves.
A four-dollar move in a low-priced instrument can be enormous.
A four-dollar move in a much more expensive instrument can be ordinary.
Wilder wanted a standardized way to express strength and weakness.
In a later interview reprinted by the CMT Association, he described RSI as a way to put commodities and stocks on the same scale.
That scale became:
0
to
100This was the design problem behind one of technical analysis's most famous indicators.
2. The Man Behind the Formula Was an Engineer Before He Was a Trader
Wilder did not begin his career on Wall Street.
In his own account, he worked as an automobile mechanic, served in the U.S. Navy as an airplane mechanic, and later studied mechanical engineering at North Carolina State College using the GI Bill.
After college, he spent years in engineering.
Then he moved into real-estate and land development.
With partners, he helped build more than a thousand apartments across several cities in North Carolina and Virginia.
By his late 30s, his partners bought his share of the projects.
Suddenly he had capital.
And time.
Wilder became interested in commodities partly because their leverage was even greater than real estate.
At first, the story seemed easy.
He made money in silver.
Then he traded other commodities.
And learned the other half of leverage:
large upside
also means
large downsideHe later said that after discovering he could lose money as well as make it, he stopped trading for a period and began studying technical analysis seriously.
This was the early-to-mid 1970s.
The engineer had found a new system to understand.
The system was the market.
3. The Problem Wilder Was Trying to Solve
Wilder was not working in today's environment.
There was no TradingView chart with hundreds of indicators.
No Python notebook could calculate 50 variations in seconds.
No online library could tell you which formula every platform used.
Wilder later recalled that before his 1978 book, the main fully automatic trading systems he knew were essentially moving-average systems and Richard Donchian's weekly rule.
He spent the mid-1970s developing new systematic ideas.
This detail is important.
Wilder's work was not centered on drawing more lines for visual decoration.
He was interested in rules that could be calculated and followed.
That engineering mindset appears throughout his work:
define the measurement
→ make the rule explicit
→ reduce subjective judgment
→ use a small number of parametersRSI fits that philosophy.
Instead of saying:
“This market feels strong.”
Wilder wanted to turn recent upward and downward movement into a repeatable number.
4. How the RSI Idea Was Built
The calculation begins with something almost trivial:
today's Close
-
yesterday's CloseThen Wilder separated that change into two sides.
positive change
→ Gain
negative change
→ LossThe two sides were averaged.
Then compared.
Relative Strength
=
Average Gain
/
Average LossFinally, that ratio was transformed:
RSI
=
100 - 100 / (1 + RS)The result is bounded.
0 ≤ RSI ≤ 100The key innovation is easier to see if we ignore the name for a moment.
RSI takes a messy sequence of positive and negative price changes and turns their recent balance into a common numerical scale.
That was much closer to Wilder's original design problem than the modern shortcut:
RSI > 70
→ sellThe modern TA-Lib implementation still follows Wilder's basic structure: positive and negative changes are separated, the first gain/loss averages are initialized from a simple average, later values use Wilder's recursive smoothing, and the standard period is 14.
If you want to build that calculation step by step in Python, see What Is RSI? Compare Recent Gains and Losses with Python.
5. Why 14? And Why Smoothing?
The number 14 became part of RSI's identity.
But we should be careful about creating a story that the historical record does not support.
Wilder's 1978 book used 14 periods, and modern implementations commonly preserve that default.
But later technicians have asked why 14 was chosen.
John Ehlers, for example, recalled that he could not determine a satisfying market-based reason for the fixed 14-period choice and later pursued adaptive indicators partly because of that question.
That does not mean 14 is wrong.
It means:
14
=
a historical design choice
not
=
a law of natureThe smoothing choice is more revealing.
Wilder did not simply throw away the oldest 14-day window and recalculate everything from scratch.
After the initial average, the previous average is carried forward and updated with the new gain or loss.
That gives the indicator memory.
past influence
does not disappear suddenly
it fadesThis same smoothing logic appears elsewhere in Wilder's work.
RSI was not an isolated trick.
It was part of a broader way of turning noisy market movement into recursively updated measurements.
6. The Name "Relative Strength" Can Mislead You
There is a historical naming trap here.
The phrase relative strength existed in investment research before Wilder's RSI.
Robert A. Levy published *Relative Strength as a Criterion for Investment Selection* in *The Journal of Finance* in 1967.
That work belongs to an older relative-strength tradition concerned with selecting stronger securities.
Wilder's RSI is a different construction.
Inside RSI:
Relative Strength
=
this market's average gains
/
this market's average lossesIt is not:
AAPL
/
S&P 500So when you hear "relative strength," ask:
relative to what?For Wilder's RSI, the comparison is internal:
upward movement
versus
downward movementThat distinction prevents a surprisingly common conceptual error.
7. The 1978 Moment: A Book Met the Personal Computer
In 1978, Wilder self-published *New Concepts in Technical Trading Systems* through Trend Research.
The book introduced a family of ideas that included RSI, Directional Movement, Parabolic systems, True Range / volatility concepts, and other systematic methods.
Wilder later described 1978 as his first major success in the commodities industry.
The timing was unusually favorable.
He recalled placing an RSI article and a full-page advertisement in the June 1978 issue of *Commodities* magazine.
At roughly the same time, small computers were becoming accessible to individual traders.
That created a perfect historical fit.
Wilder had produced calculations.
Traders suddenly had machines they wanted to program.
new formulas
+
new personal computing power
=
rapid diffusionIn the interview, Wilder said the book sold thousands of copies around the world and eventually more than 25,000 copies.
That figure is his own retrospective account, not an independently audited sales record.
But the broader historical outcome is easy to verify:
RSI became a standard technical-analysis function and remains implemented in modern libraries such as TA-Lib.
This is historical success.
It does not yet tell us whether an RSI trading rule reliably makes money.
8. Wilder's Philosophy Was More Interesting Than "70/30"
If RSI were only:
above 70 → sell
below 30 → buyWilder's philosophy would be much less interesting.
In his later interview, when asked what RSI signals mattered, Wilder specifically pointed readers toward the Failure Swing concept in his original book.
That is a clue.
The original use of RSI was richer than treating 70 and 30 as automatic order buttons.
Wilder also expressed broader views about system design.
He argued that markets adapt.
A trading system can work for a period and then deteriorate as market behavior changes and participants become more sophisticated.
He also favored systems that worked across more commodities with fewer parameters.
That philosophy sounds surprisingly modern:
avoid overfitting
prefer robustness
expect decay
follow the system
separate measurement from emotionSo a historically faithful reading of Wilder is not:
“Find the magic RSI threshold.”
It is closer to:
“Build a clear measurement, understand what it means, use few assumptions, and do not confuse one historical setting with a permanent law.”
9. Did RSI Actually Work?
We need to separate three different meanings of success.
Historical success
Yes.
RSI survived.
It entered charting software, books, training material, and technical-analysis libraries around the world.
A measurement created before the modern personal-computing era is still a standard function decades later.
Conceptual success
Also yes.
RSI gave traders a compact way to express the balance of recent positive and negative price changes on a common 0–100 scale.
Its structure also influenced later oscillators and variations.
Trading-performance success
This answer is much more complicated.
Some empirical studies have found profitable RSI rules in particular markets and periods.
A 2008 study of the London FT30 reported that tested RSI and MACD rules often outperformed buy-and-hold in its historical sample.
A 2014 study found significant abnormal returns for particular RSI and MACD parameterizations in some OECD stock indices.
But other evidence is much less encouraging.
A 2015 study across five Southeast Asian equity markets found significant gross results in several markets, yet transaction costs eliminated most net profits outside Thailand.
And a much broader 2023 study examined 6,406 technical rules across 41 developed and emerging equity markets. It found that apparent predictability declined sharply over time, was sensitive to moderate transaction costs, and did not persist well out of sample.
That does not prove RSI is useless.
It proves something more important:
RSI
is an indicator
RSI(14) < 30 → buy
is a strategy
Those are not the same thing.10. How RSI Became Easier to Misuse as It Became More Famous
Success created a paradox.
The more widely RSI spread, the easier it became to teach it in a compressed form.
The compressed lesson became:
70 = overbought
30 = oversoldThen:
overbought = price must fall
oversold = price must riseBut a strong trend can keep an oscillator elevated or depressed for a long time.
And Wilder himself did not describe RSI as a machine that guaranteed the next move.
A threshold tells us something about the recent balance of gains and losses under a chosen smoothing period.
It does not force the market to reverse.
The correct research question is not:
“Does RSI work?”
It is:
“Does this exact RSI rule, with this period, threshold, market, execution rule, cost model, and test period, hold up out of sample?”
That is the Alphesta question.
Before You Use RSI
It was designed to measure:
the balance of recent gains and losses
on a common bounded scale.
It was not designed to guarantee:
the next price direction.
Its original context was:
systematic technical trading in the 1970s,
especially commodities and stocks,
before modern charting software.
Later evidence suggests:
some RSI rules have worked in some markets and periods,
but performance is highly conditional and often weakens
after costs or out-of-sample testing.
A trading rule using RSI is still:
a hypothesis that must be tested.What to Remember
Wilder's most important contribution may not be the number 14.
It may not even be 70 and 30.
The deeper idea is this:
messy price changes
→ separate gains and losses
→ smooth them
→ compare them
→ put them on one scaleAn engineer wanted markets to speak a common numerical language.
That language became RSI.
And the safest way to use it today is to remember the problem it was built to solve before turning its output into a trading command.
Sources
1. J. Welles Wilder, *New Concepts in Technical Trading Systems*, Trend Research, 1978. Google Books bibliographic record.
2. CMT Association, “J. Welles Wilder,” reprint of a Trader’s Journal interview, *Technically Speaking*, August 2006.
3. TA-Lib — RSI.
4. Robert A. Levy, “Relative Strength as a Criterion for Investment Selection,” *The Journal of Finance*, 1967.
5. Chong & Ng, “Technical analysis and the London stock exchange: testing the MACD and RSI rules using the FT30,” *Applied Economics Letters*, 2008.
6. Chong, Ng & Liew, “Revisiting the Performance of MACD and RSI Oscillators,” *Journal of Risk and Financial Management*, 2014.
7. Tharavanij, Siraprapasiri & Rajchamaha, “Performance of technical trading rules: evidence from Southeast Asian stock markets,” *SpringerPlus*, 2015.
8. Rink, “The predictive ability of technical trading rules: an empirical analysis of developed and emerging equity markets,” *Financial Markets and Portfolio Management*, 2023.