Moving Averages for the CMT Exam: Trend, Lag, and Execution

A moving average is a filter

A moving average smooths noisy price observations. It is valuable for describing trend, but it lags because it is computed from past data. The 2026 CMT Program Guide covers indicators and trend analysis at Levels I and II and integrated indicator use at Level III. Understand the inputs and behavior; the official FAQ says candidates are not required to memorize all technical-indicator formulas, although working knowledge of volatility, risk, and performance calculations may be required.

A simple moving average (SMA) gives equal weight to the last n observations. An exponential moving average (EMA) weights the most recent observation more heavily and carries information from earlier observations. The common EMA update uses alpha = 2/(n+1): new EMA = alpha × current close + (1 - alpha) × previous EMA. Initialization and data adjustments can make platforms differ slightly.

Worked SMA and EMA

Five closes are 20, 22, 21, 23, and 24. Their five-period SMA is (20 + 22 + 21 + 23 + 24)/5 = 22. If the next close is 26, the next five-period SMA drops the oldest 20: (22 + 21 + 23 + 24 + 26)/5 = 23.2. The SMA changes even if the incoming close is not extraordinary because an old value leaves the window.

For a five-period EMA, alpha = 2/6 = 1/3. If the prior EMA is 22 and the new close is 26, the next EMA is (1/3 × 26) + (2/3 × 22) = 23.33 approximately. This faster response can reduce some lag but can increase false turns. Neither method is universally superior.

A close above a moving average can support an uptrend description; it does not imply a profitable buy at any price. A rising average and price above it may be more informative than one crossover in isolation. The choice of length determines the horizon being measured. A 200-day average describes something different from a 10-day average, even when both use daily bars.

Crossover rule with a real failure condition

Suppose a tactical system buys at the next open when a 20-day SMA crosses above a 50-day SMA at a daily close, provided the 50-day SMA is rising. It exits at the next open after a reverse cross or a predetermined risk breach. The signal is known only after the close; filling at that same close in a backtest can introduce look-ahead bias. Use the next realistically executable price and include costs.

A possible invalidation rule is that the 20-day average falls back below the 50-day average for two consecutive closes. Requiring two closes may reduce whipsaws but delay exits; compare the tradeoff rather than asserting a universal best setting. If the slower average slopes down, the filter blocks entry even if a short crossover occurs. This is one method, not the official exam's preferred strategy.

Moving averages can act as dynamic reference zones, yet they are calculated lines, not guaranteed support or resistance. In a sideways range, repeated crosses can produce losses. Volatility and range measures can identify conditions in which a trend strategy should reduce size or stand aside.

Testing without fooling yourself

Set asset universe, adjustment method, signal time, execution price, costs, and rebalancing before evaluation. Split data by time to check out-of-sample behavior. Repeatedly trying lengths until one looks best turns the historical fit into a selected result. Compare with a simple baseline and report drawdown, turnover, and exposure as well as return.

A crossover often overlaps heavily with price trend and momentum signals. Treating all three as independent confirmation exaggerates confidence. If a result depends on one extraordinary trend period, say so.

Choose the length from the decision horizon

A moving-average length is a design choice tied to holding period and market noise. A short average will react to new closes quickly and may cross a slower one repeatedly in a range. A long average gives a smoother regime description but responds later to a turning market. There is no single length that makes a crossover reliable across stocks, futures, and currencies.

Suppose the five-period SMA in the worked example is 23.2 and the next close is 18. The new SMA drops 22 and adds 18: (21 + 23 + 24 + 26 + 18)/5 = 22.4. A decline in the average does not necessarily mean a confirmed downtrend; it is one update in a five-bar window. A reader should inspect the sequence and chosen horizon rather than react to one slope change.

When comparing signals, avoid a false sense of independent confirmation. Price above a 20-day average, a rising 20-day average, and a 20/50 crossover are all functions of overlapping prices. Volume, breadth, or relative performance may add different information. If a strategy depends on a 20-day average, write the rule for missing bars, holidays, and adjusted prices before evaluating returns.

For an exam-style comparison, describe exactly which old observation exits an SMA window and why the EMA responds differently. If the latest close rises while an SMA falls, a very high old close may have left the window; the current bar alone does not determine the slope. For an EMA, all past observations remain implicitly represented through the prior EMA, though their influence decays. This distinction helps interpret unusual-looking indicator changes without memorizing a picture.

Common traps and self-check

Do not call EMA "leading"; it is still based on past observations. Do not compare averages from differently adjusted price series. Do not assume an average crossing is tradable at the close where it was computed. Avoid moving a parameter after every losing trade.

Self-check: The last four closes are 8, 9, 10, and 11. What is the four-period SMA after the next close of 12? Answer: Drop 8 and average 9 + 10 + 11 + 12 = 42; 42/4 = 10.5.

Next, calculate three successive SMA values by hand, describe the trend each measures, then test the indicator practice topic. Use the official objectives to identify the level of calculation and interpretation required.