Following the Tides

Opening Scene

A current tells a navigator where the water is generally drifting. The tides tell a different, equally important story: a regular, predictable rise and fall that repeats on a known schedule, tied to the moon rather than to any lasting drift. A skilled navigator doesn’t confuse the two — mistaking a temporary tidal swing for a genuine change in current would produce a badly wrong course. Time series carry this same regular, repeating rhythm, and mistaking it for something else is a common, costly error.

In Plain English

Seasonality is a regular, repeating pattern in a time series tied to a known, fixed period — daily, weekly, monthly, or yearly cycles are the most common. Retail sales spiking every December, electricity demand peaking every afternoon, website traffic dropping every weekend — all of these are seasonality. Unlike trend, which is a persistent drift, seasonality is a predictable, repeating oscillation around that drift.

The Old Way

Before formal seasonal decomposition existed, people tracked repeating cycles through direct, informal comparison:

A farmer planning around the same planting and harvest windows every year, based on accumulated seasonal experience rather than a formal model.
A retailer stocking up ahead of the same holiday rush every year, based on remembered past patterns.
A utility company anticipating the same daily demand peaks, based on operational experience rather than formal seasonal modeling.

The instinct to expect a known cycle to repeat is ancient; formal seasonal decomposition simply made that instinct precise and automatable.

What’s Changing (and Why AI Is the Reason)

Formal decomposition methods, covered directly in Article 7, now separate seasonality from trend and noise mathematically, rather than relying on informal, memory-based pattern recognition.
Multiple overlapping seasonal cycles — daily and weekly and yearly patterns simultaneously — can now be modeled together explicitly, something informal, memory-based approaches struggled to disentangle.
As forecasting scales to thousands of sequences at once, covered in Article 19, automated seasonality detection has become essential, since manually identifying every sequence’s seasonal pattern by eye simply doesn’t scale.

The Metaphor, Fully Extended

The Voyage	Seasonality Concept
The tides, rising and falling on a known, fixed schedule	A regular, repeating pattern tied to a known period
Mistaking a tidal swing for a genuine change in current	Mistaking seasonality for a genuine trend shift
Multiple overlapping tidal and lunar cycles	Multiple overlapping seasonal patterns (daily, weekly, yearly)
A tide table, precomputed and reliable	A formally estimated, mathematically separated seasonal component

For Beginners: What to Actually Do

Plot a time series over at least two or three full seasonal cycles to visually confirm a suspected pattern before assuming it’s real.
Practice distinguishing a genuine seasonal pattern from a one-time event that happens to resemble one.
Learn to check for multiple overlapping seasonal periods in the same sequence, not just the most obvious one.

For Practitioners and Leaders: The Deeper Layer

Require explicit seasonality analysis, alongside trend analysis from Article 4, as a standard step before any forecasting model is chosen.
Watch for seasonal patterns that shift over time — this is a genuine, distinct concern covered directly in Article 15.
Invest in automated seasonality detection at the scale covered in Article 19, rather than relying on manual review of individual sequences.

Quick Recap

Seasonality is a regular, repeating pattern in a time series tied to a known, fixed period.
It’s distinct from trend — a predictable oscillation rather than a persistent drift.
Multiple overlapping seasonal cycles can occur in the same sequence and need to be modeled together.
Automated seasonality detection is essential once forecasting scales beyond a handful of sequences.

Where This Fits in the Series

Article 5 covered the tides — a time series’ regular, repeating rhythm. Article 6 covers what’s left over once both current and tide are accounted for: the unpredictable part.