Every moving average is a low-pass filter: it removes high-frequency noise (daily chop) but introduces phase lag — the average always trails the price. Lag is the price you pay for smoothness, and it is quantifiable.

For a simple moving average of length n, the average age of the data points is (n−1)/2 bars. An EMA of the same length has an effective center of mass of (n−1)/2 as well, but weights recent bars far more heavily, so its practical lag is roughly half that of an SMA.

Shorter n → less lag, more whipsaw, more false signals.
Longer n → more lag, fewer signals, later entries and exits.
Lag is why MAs confirm trends rather than predict turns.

MA length	Avg data age (bars)	Behavior
SMA 10	4.5	Fast, noisy
SMA 50	24.5	Medium trend
SMA 200	99.5	Slow, regime filter