Direct answer
A worked example of an EMA (Exponential Moving Average) shows how you compute a moving average value step by step from a fixed set of past observations, using a chosen period and explicit smoothing assumptions. The key idea is that each new EMA value is a weighted combination of the latest observation and the prior EMA.
Mechanism or definition
EMA is a type of moving average designed to respond more quickly to recent changes than a simple moving average. In a typical EMA formulation, you pick:
- A period N (for example, 10)
- A smoothing factor (\alpha = 2/(N+1))
- A starting value for the EMA (often the first observed price, or sometimes an initial simple moving average—this is an assumption)
A common recursive definition used for calculation is:
- (\text{EMA}_1 = \text{first value}) (assumption in the worked example below)
- For (t \ge 2): (\text{EMA}t = \alpha\cdot x_t + (1-\alpha)\cdot \text{EMA}{t-1})
Here (x_t) is the observation at time (t). In markets, (x_t) could be a closing price, but the worked example below treats (x_t) as an abstract series so you can verify the math without needing any real-time data.
Evidence or example
Worked numeric scenario (assumptions stated up front):
- Choose N = 3, so (\alpha = 2/(3+1) = 0.5).
- Use a made-up observation series (x_1) to (x_6): [100, 102, 101, 103, 104, 103].
- Initialize (\text{EMA}_1 = x_1 = 100).
Now compute each EMA value using (\text{EMA}t = 0.5\cdot x_t + 0.5\cdot \text{EMA}{t-1}):
- (\text{EMA}_1 = 100)
- (\text{EMA}_2 = 0.5\cdot 102 + 0.5\cdot 100 = 51 + 50 = 101)
- (\text{EMA}_3 = 0.5\cdot 101 + 0.5\cdot 101 = 50.5 + 50.5 = 101)
- (\text{EMA}_4 = 0.5\cdot 103 + 0.5\cdot 101 = 51.5 + 50.5 = 102)
- (\text{EMA}_5 = 0.5\cdot 104 + 0.5\cdot 102 = 52 + 51 = 103)
- (\text{EMA}_6 = 0.5\cdot 103 + 0.5\cdot 103 = 51.5 + 51.5 = 103)
Self-check: because (\alpha=0.5) here, each EMA step is exactly the midpoint between the new observation and the prior EMA. If you recompute and get different numbers, the cause is usually a different (N), a different (\alpha), or a different initialization choice for (\text{EMA}_1).
To connect this to general data handling (without assuming any specific platform), note the only required inputs for verification are: the period (N), the smoothing rule (\alpha), the observation sequence (x_t), and the chosen starting assumption.
Limitations and risks
-
Choice of initialization changes early values. If instead of (\text{EMA}_1=x_1) you initialize using an average of the first N observations, the early EMA series will differ, even though later values may converge.
-
Lag and noise sensitivity are intrinsic. EMA is still a moving average: it reflects past observations. In fast or choppy conditions, it may lag turning points and still move with short-term noise.
-
Market-dependent outcomes are variable. In real trading contexts, results you observe depend on data frequency, bid/ask spreads, execution timing, and costs. Even if EMA is computed correctly, those external factors can change realized outcomes.
-
Historical relationships do not guarantee future behavior. Past patterns in EMA changes do not establish that the same behavior will occur later.
Verification or next question
To independently verify any EMA you see, reproduce the calculation from stated inputs: confirm the period (N), compute (\alpha), use the exact (x_t) values from the same data source and timeframe, and confirm the starting assumption for the first EMA value. If you want, share your chosen N and the first few (x_t) values, and you can use the same (\text{EMA}_t) recursion to check whether your computed EMA matches the arithmetic above.