Seasonality is a recurring tendency for a market variable to behave differently at particular times of the year or another fixed calendar interval. It can appear in prices, returns, spreads, volatility, trading activity, inventories, production, consumption or other market quantities. A seasonal pattern is an empirical regularity: it describes what has tended to recur, not what must happen in the next cycle.
The calendar marks the pattern; economics sustains it
Seasonality is most meaningful when the repeated timing has an economic mechanism behind it. In commodity markets, harvest schedules, heating and cooling demand, storage injections and withdrawals, recurring consumption, transportation constraints and other periodic changes in supply or demand can create recognizable patterns within the year. Those patterns are not necessarily permanent. Changes in storage, imports, production geography, technology or demand can smooth a seasonal effect, alter its magnitude or move its timing even though the calendar itself has not changed.
This distinguishes seasonality from a broader cycle. A commodity or business cycle may unfold over several years and need not recur at a fixed calendar date. A seasonal window is narrower: it is a particular interval in which a seasonal tendency has historically been concentrated.
The measured pattern depends on construction
There is no single universal seasonal curve. Different sample periods, calendar alignments, trading-day conventions, outlier treatment and choices between price levels, returns or spreads can produce materially different profiles. Average and median patterns may also differ when a small number of unusual years exert disproportionate influence.
Futures research adds another construction issue. A continuous futures series can switch its underlying contract according to rules based on expiry, liquidity or another roll convention. A seasonal pattern measured from that series can therefore reflect both the economics of the underlying market and the way contract exposure was joined through time.
Backtests introduce a separate trap: a seasonal profile estimated from the full historical sample uses information from later years unless it is re-estimated using only data that would have been available at each historical decision date. Even without look-ahead bias, a stable historical average can conceal structural change. Seasonality is most persuasive when the tendency survives reasonable construction choices, its economic mechanism still exists and the relevant market regime has not materially changed it. It is evidence about recurring timing, not a self-contained directional signal.