Why start with Dow Theory?
Dow Theory is a way to organize market observations into a provisional trend judgment. It asks what price is doing, whether related averages agree, whether volume supports the prevailing direction, and what evidence would change the conclusion. The current CMT Program Guide names Dow Theory, its key individuals, averages, trend movements, closing price, and confirmation among Level I learning objectives. Those are verified curriculum boundaries; the numeric example here is original and is not an official exam item.
This legacy primer concentrates on making an observation log from a chart. The companion Dow Theory for the CMT Exam develops the six tenets as a conceptual map.
The observational framework
Charles Dow published stock averages in the nineteenth century. Later writers organized his market observations into what is now called Dow Theory. Do not treat a modern summary as a verbatim list written by Dow. The common framework says that market averages reflect available information, trends unfold on several horizons, major trends move through phases, related averages should confirm, volume should generally support the main trend, and an established trend remains in force until convincing reversal evidence appears.
The closing price matters because it summarizes the final negotiated level of a trading period. An intraday high above resistance can be notable, but if a rule requires a close, the high alone does not satisfy it. Different data providers may use adjusted closes or different session definitions. Record which series you use.
Three scales of movement can coexist. A primary uptrend can contain a secondary decline, and that decline can contain minor rallies. The analyst's job is to label the horizon before assigning meaning. Calling every down day a primary reversal confuses noise with structure. Conversely, calling a multi-month series of lower lows “only a pullback” can keep a bullish thesis alive after its evidence has failed.
| Horizon | Example observation | Appropriate question |
|---|---|---|
| Primary | Rising multi-month swing highs and lows | Has broad direction changed? |
| Secondary | Several-week retreat within that rise | Is the pullback still bounded by the primary structure? |
| Minor | Two-day countertrend rally | Is it relevant to the chosen decision horizon? |
These durations are illustrative, not mechanical cutoffs. The same market can be in different trends on different intervals.
Two-average confirmation: a worked case
Use two related, independently quoted broad averages for this hypothetical exercise, called Industry and Transport. The former closes at a previous significant high of 4,000 and later at 4,080. The latter's previous significant high is 2,000, but its new close is only 1,960. Industry gained (4,080/4,000 − 1) × 100 = 2%. Transport remains (1,960/2,000 − 1) × 100 = −2% below its own high. One average has confirmed a new high and the other has not.
The appropriate conclusion is a nonconfirmation warning, not an automatic bearish trend reversal. It may be a temporary lag, different industry exposure, or a real loss of participation. A practical observation rule might be: retain the prior primary trend designation, flag the divergence, and seek either a Transport close above 2,000 or independent reversal evidence. If Transport later closes at 2,020, the high is (2,020/2,000 − 1) × 100 = 1% above its reference high, so the two-average warning is resolved under this example's rule. If Industry instead breaks a prior secondary low while Transport remains weak, the bullish case has more serious contrary evidence.
This example depends on selecting significant highs consistently. Moving the comparison high after seeing the result undermines the method. The averages are also imperfect proxies: sector composition, index methodology, and modern supply chains can change how well a pair represents the economy.
Volume supports, but does not vote alone
A rising primary trend commonly has more volume on advances than on retreats in the traditional framework. Suppose a five-day rally averaged 1.5 million shares daily and a five-day pullback averaged 0.9 million. The rally-to-pullback volume ratio is 1.5/0.9 = 1.67. That is a participation observation. It cannot by itself identify accumulation, prove that buyers are “smart money,” or guarantee the next price move. Modern fragmentation and changes in trading venue can complicate comparisons, so use like-for-like data.
The phase language—accumulation, broad participation, distribution—is a useful narrative for how a primary trend may develop. It is easy to label only with hindsight. A more disciplined approach records observable evidence: price trend, cross-average confirmation, volume balance, and failed breaks. Then attach a phase hypothesis cautiously. If price and volume contradict the label, revise the label.
Decision rules and invalidation
For a study case, write the primary-trend claim in one sentence: “Both averages have formed higher confirmed closing highs and higher intervening lows.” The support is the pair of highs and lows. A warning is one average failing to confirm. An invalidation is a defined reversal sequence—for example, both averages closing beneath their preceding significant secondary lows after failing to make confirmed new highs. Real analysts may use different conventions; the important skill is to state and apply one consistently.
Avoid interpreting “a trend persists” as permission to ignore risk. A long position can have a protective exit before a full primary-trend reversal is established. Trend classification and position management answer different questions.
Common errors
The most frequent conceptual error is treating nonconfirmation as a sell signal. The second is silently switching from closes to intraday extremes when convenient. The third is using a single stock to infer two-average broad-market confirmation. Also avoid claiming that Dow Theory “proves” all information is perfectly priced; its market-discounts premise is a working assumption, not a validation of every asset price. See Market Breadth Indicators for other participation measures.
Self-check
Index A's old closing high is 500; it closes at 515. Index B's old closing high is 800; it closes at 792. Has a two-average new high been confirmed? Answer: No. A is 3% above its prior high, while B is 1% below its own (792/800 − 1 = −1%). Record nonconfirmation and identify what would resolve it. Do not infer a complete reversal from those two values.
Next, choose two related index charts, mark prior significant closing highs and lows before revealing the newest month, and write separate lines for observation, warning, and invalidation.