A moving average looks so simple that it is easy to imagine a trader invented it while staring at a price chart.
That is not what happened.
The idea came from a much broader problem:
noisy data
↓
find the underlying movementStatisticians used averaging to smooth changing data before moving averages became a standard tool on trading screens. Traders later borrowed that idea and asked a different question:
Can a smoother price series
help us follow a market trend?That small change in purpose turned a statistical tool into one of the most recognizable ideas in technical analysis.
1. The Moving Average Did Not Begin in Trading
Early statisticians were already trying to remove short-term variation from economic and social data around the beginning of the twentieth century.
R. H. Hooker's 1901 work on economic data is part of that early history. G. Udny Yule discussed related calculations in 1909, and historical accounts trace the expression moving average to this period.
By 1912, Willford I. King's textbook The Elements of Statistical Method included a section titled Index Historigrams with Moving Average.
early 1900s
many observations
↓
local averages
↓
smoother seriesThe original problem was not:
When should I buy a stock?It was closer to:
How can I see the broader movement
inside data that jumps around?That distinction matters. A moving average is first a data transformation. A trading rule is something people may build on top of it.
2. So Who First Used It for Trading?
This question has no perfectly clean answer.
Traders experimented with charts long before every method was documented carefully. So it would be too strong to say that we know the first person in history who ever traded using an average of prices.
But one name is especially important when we ask about systematic moving-average trading:
Richard DonchianDonchian became a pioneer of rule-based trend following in futures markets. CME Group records that he started Futures, Inc. in 1949, widely described as the first publicly managed futures fund.
Historical accounts of technical trading also associate Donchian with diversified trend-following rules using 5-day and 20-day moving averages.
short average
↕
long average
↓
rule-based trend decisionThis was an important step.
The moving average was no longer only a line that made data easier to see. It became part of a repeatable decision process.
The safest historical statement is therefore:
Richard Donchian was not the inventor of the mathematical moving average, but he is one of the earliest and best-documented pioneers of using moving averages systematically in trend-following trading.
3. Donchian Changed the Question
Once a moving average is on a price chart, several measurable relationships appear.
price above or below average
average rising or falling
short average above or below long averageThese are observations.
Donchian's contribution was to turn observations like these into consistent rules that could be followed across many markets.
That is the beginning of an idea that later became central to systematic trading:
do not decide from a story every day
measure
→ define a rule
→ follow it consistentlyThis does not prove the rule will make money. It simply makes the rule testable.
4. Granville Helped Make the 200-Day Line Famous
Another important name is Joseph E. Granville.
His 1960 book A Strategy of Daily Stock Market Timing for Maximum Profit discussed many market indicators, including the 200-day moving average.
The long moving average gave traders a simple way to compare today's price with a much slower estimate of the market's recent level.
price
versus
200-day averageGranville did not invent averaging. His importance was helping turn the long moving-average line into a widely known market-timing framework.
This is why the 200-day moving average still appears so often in financial commentary today.
5. Then Came a Different Kind of Memory
The Simple Moving Average has one obvious limitation.
Inside its window, every observation receives equal weight.
SMA
old price = equal weight
new price = equal weightDuring the 1950s, forecasting researchers developed exponential smoothing methods in which recent observations receive more weight and older observations fade gradually.
Charles Holt's work dates to 1957, Robert G. Brown's major book to 1959, and Peter Winters extended the approach in 1960.
recent observation
→ larger influence
older observation
→ smaller influenceThis is the same core memory idea you learned when we built the EMA from scratch.
The important historical point is that exponential weighting was part of a wider time-series and forecasting revolution. Traders later applied the same kind of smoothing logic to market prices.
6. Computers Changed Moving Averages Again
A moving average is easy for a computer.
But before modern computers, calculating and updating many averages across many markets was repetitive work.
By the 1970s, systematic traders began using computers to test trend rules across historical data. Ed Seykota is a well-known early example; accounts of his work describe testing Donchian-inspired ideas on early computers.
hand-drawn chart
↓
mechanical rule
↓
computer backtest
↓
many markets + many yearsThis changed the discussion.
Instead of asking only:
Does this chart look convincing?researchers could ask:
How did the rule behave
across thousands of historical observations?7. Academic Research Asked the Harder Question
Once computers made large tests practical, moving averages stopped being only a chartist's tool. They became an object of academic research.
A famous 1992 paper by William Brock, Josef Lakonishok, and Blake LeBaron tested simple moving-average and trading-range rules on a long history of the Dow Jones Industrial Average.
Their results gave serious attention to the possibility that simple technical rules contained information about returns.
But later research made the standard higher.
one good backtest
is not enoughResearchers increasingly asked about:
- transaction costs,
- data snooping,
- false discoveries,
- out-of-sample performance,
- and whether an apparent edge survives after many traders know about it.
A CFA Institute review of thousands of technical rules found that much apparent predictive power became weak after accounting for false discoveries and trading costs. Other research has found that moving-average predictability can change through time as markets adapt.
That leads to an important Alphesta principle:
indicator
≠
proven edge8. Modern Moving Averages Are Really a Family of Filters
Today, "moving average" no longer means one formula.
SMA
→ equal weights in a fixed window
EMA
→ exponentially fading weights
adaptive averages
→ change their response as market conditions changePerry Kaufman's adaptive moving-average work in the 1990s is one example of this next step. Instead of choosing one fixed speed forever, an adaptive method tries to respond differently when price movement is directional versus noisy.
Modern quantitative research often describes moving averages more generally as filters used to extract trend from noisy data.
market price
↓
filter
↓
compressed view of trendThat language brings us surprisingly close to where the story began more than a century ago.
The original goal was to make noisy data easier to understand.
The modern goal is still very similar.
9. A Timeline Worth Remembering
1901–1912
moving averages emerge in statistical work
↓
1930s–1940s
systematic trend-following ideas develop
↓
1949
Richard Donchian starts Futures, Inc.
↓
1950s–1960s
exponential smoothing develops
Granville popularizes long moving-average timing
↓
1970s
computers make large rule tests practical
↓
1990s
academic tests + adaptive moving averages
↓
today
moving averages are treated as filters,
strategy components, and research hypothesesDo not memorize every date.
Remember the direction of development:
SMOOTH DATA
↓
SEE TREND
↓
MAKE A RULE
↓
TEST THE RULE
↓
ADAPT THE FILTER10. What History Does Not Tell Us
History can explain where an idea came from.
It cannot tell us that a moving average will predict the next price.
A 20-day, 50-day, or 200-day average is not useful because it is old or famous.
Its usefulness is a hypothesis that must be tested for a particular market, rule, time horizon, and cost structure.
history
→ explains the idea
data
→ tests the ideaThis distinction is one of the most important habits you can build before we move deeper into technical indicators.
Check Your Understanding
You are ready to move on if you can explain these ideas in your own words:
- Moving averages began as a way to smooth changing data, not as a buy-or-sell signal.
- There is no securely documented single "first trader" who can be proven to have used a moving average before everyone else.
- Richard Donchian is one of the earliest and best-documented pioneers of systematic moving-average trend following.
- Joseph Granville helped popularize long moving-average market timing, especially the 200-day line.
- Exponential smoothing gave recent data more weight and helped shape the logic behind EMA-style filters.
- Computers changed moving averages from hand-updated chart lines into rules that could be tested over many markets and years.
- Modern research treats moving averages as filters and hypotheses, not guaranteed sources of profit.
What You Just Learned
statistics
↓
smoothing
↓
trend following
↓
systematic rules
↓
computer testing
↓
adaptive filtersIf one idea stays in your head after this lesson, let it be this:
A moving average is not an old trading trick. It is a century-old idea for compressing noisy data that traders gradually turned into a testable trend filter.
Now that we know where moving averages came from, we can leave this family for a moment and ask a new question:
Instead of smoothing price, what if we measure how much price itself is changing?
Sources and Further Reading
For this history lesson, the references below are arranged so you can inspect the source itself whenever possible. Digitized books, original journal pages, working papers, and readable reproductions are listed before general historical summaries.
Primary and Read-It-Yourself Sources
- R. H. Hooker, "The Suspension of the Berlin Produce Exchange and its Effect Upon Corn Prices," Journal of the Royal Statistical Society, 1901. This is the original journal article page. Read the journal record / article page
- G. Udny Yule, "The Applications of the Method of Correlation to Social and Economic Statistics," Journal of the Royal Statistical Society, 1909. This is the original journal article page from the period in which moving-average terminology was emerging. Read the journal record / article page
- Willford I. King, The Elements of Statistical Method, 1912. Google Books provides a digitized copy and its table of contents explicitly lists Index Historigrams with Moving Average on page 195. Open the digitized book
- Richard D. Donchian, "Donchian's 5- and 20-Day Moving Averages." A later readable reproduction preserves the article text and publication details. Because this is a reproduction rather than the original magazine host, it should be read as a convenient access copy. Read the reproduced article The original 1974 Commodities Magazine citation is also listed in the bibliography of Trading Systems and Methods. Check the bibliographic record
- Joseph E. Granville, A Strategy of Daily Stock Market Timing for Maximum Profit, 1960. Google Books provides a digitized preview and its searchable terms include 200-day moving average. Open the digitized book preview
- Charles C. Holt, "Forecasting Seasonals and Trends by Exponentially Weighted Moving Averages." The 2004 International Journal of Forecasting article is a reprint of Holt's 1957 Office of Naval Research report and was published specifically to make the work more accessible. Read the article page and reprint
- Peter R. Winters, "Forecasting Sales by Exponentially Weighted Moving Averages," Management Science, 1960. The publisher page includes the abstract and a PDF download link. Read the original paper page / PDF
- William Brock, Josef Lakonishok, and Blake LeBaron, "Simple Technical Trading Rules and the Stochastic Properties of Stock Returns." The Santa Fe Institute hosts the 1991 working-paper version with a PDF. It tests moving-average and trading-range rules using Dow Jones data from 1897 to 1986. Read the working paper / PDF The final article appeared in The Journal of Finance in 1992.
- Perry J. Kaufman, Smarter Trading: Improving Performance in Changing Markets, 1995. Google Books provides a digitized preview whose searchable text includes Adaptive Moving Average, Efficiency Ratio, and related filtering concepts. Open the digitized book preview
- Benjamin Bruder, Tung-Lam Dao, Jean-Charles Richard, and Thierry Roncalli, "Trend Filtering Methods for Momentum Strategies," 2011. SSRN provides the paper and downloadable PDF. It treats moving averages within the broader problem of extracting trends from noisy financial time series. Read the paper / download PDF
Historical and Research Context
- CME Group, "The Miraculous Growth of Managed Futures." Useful for the historical record of Richard Donchian, Futures, Inc., and the development of managed futures. Read the CME historical note
- CFA Institute, "Technical Trading Revisited: False Discoveries, Persistence Tests, and Transaction Costs." This review is useful for understanding why a technical rule that looks successful in one test still needs controls for multiple testing and trading costs. Read the research digest
Reading tip: You do not need to read every source from beginning to end. Start with King's 1912 digitized book, the Donchian reproduction, Granville's book preview, Holt and Winters on exponential weighting, and the Brock-Lakonishok-LeBaron working paper. Together they show the main path:
smooth data
↓
use the smoother series on market prices
↓
turn observations into rules
↓
test the rules with data
↓
treat moving averages as filters