Understanding SMA and Data Smoothing

What is smoothing and why do we use it?

At CryptoInflation, we smooth our data because circulating supply changes day by day, often in unpredictable ways. To show the underlying trend in cryptocurrency emissions—and to give users a clearer, more stable experience—we use the Simple Moving Average (SMA) smoothing algorithm.

A quick definition of SMA

For a window of (w) days, the (backward-looking) SMA at time (\(t\)) is

$$ \begin{align} \mathrm{SMA}_t(w) = \frac{1}{w}\sum_{i=0}^{w-1} x_{t-i} \end{align} $$

This averages the current value with the previous (\(w-1\)) days. Smaller (\(w\)) reacts faster but is noisier; larger (\(w\)) is smoother but introduces more lag.

Try it yourself

Below you can see Ethereum’s emission without any smoothing applied. Open the Smoothing tab to adjust the emission and supply windows and watch how the chart responds.

＋ Compare

Inflation Emission Supply Price Market Cap

1W 1M 1Y 3Y 10Y ALL Custom

Linear Log

Smoothing ?

Emission window (days): Supply window (days): Apply Reset

Tip: Supply smoothing affects how supply is first smoothed on the server. Emission smoothing smooths the emission (supply delta) series. Larger numbers = smoother but slower to react.

What does a “window” mean?

The window is the number of days included in the average.
For example, if you set the emission window to 10, each day’s emission is averaged over the current day plus the prior 9 days (10 values total).

• Smaller windows → quicker to show changes, but the chart looks grainier.
• Larger windows → slower to react, but the chart is smoother.

We set the default values to emission window = 14 and supply window = 70. In our experience this strikes a good balance: emissions are naturally jumpier and benefit from a two-week average, while supply moves more slowly and looks best with a longer window. Of course, this can vary by asset, so feel free to adjust the sliders to your preference.

Tips for choosing a window

• Want to spot short-term moves? Try smaller emission windows (e.g., 7–14).
• Focusing on long-term trends? Increase both windows (e.g., 30 / 90).
• Remember: larger windows add lag, so peaks and troughs may appear slightly shifted in time.

A note on data coverage

At the very start of a series there might be fewer than (w) days available. In those cases we compute the average over the available data; as more days accrue, the smoothing becomes more stable.