Relative Seasonality vs Absolute Seasonality
Direct Answer
Relative seasonality tests whether an asset outperforms a benchmark during a given calendar period, rather than testing whether the asset's own return was simply positive during that period (absolute seasonality). Because broad markets have historically risen in most months regardless of season, an asset's absolute win rate is confounded with the market's general drift — a stock "up 70% of Januaries" is unremarkable if the market itself is up in roughly 60% of all months. The relative outperformance rate isolates any asset-specific calendar effect and is the methodologically stronger test.
Key Takeaways
- Absolute seasonality is confounded with market drift: A high historical "up rate" for a given month partly reflects the market's long-run upward tendency, not an asset-specific calendar effect.
- Relative seasonality subtracts the benchmark return for the same period, isolating whatever calendar pattern is specific to the asset rather than shared with the market.
- The benchmark must be chosen before the test is run — selecting whichever benchmark makes the result look strongest is a form of researcher degrees of freedom.
- Relative seasonality does not solve the small-sample problem. A monthly pattern tested over 20 years still has only 20 independent yearly observations, and the same multiple-testing and regime-instability concerns from calendar-based seasonality research still apply.
- Sector and industry seasonality is a natural application of relative seasonality, since a sector is typically compared against the broad market it sits inside.
- A relative outperformance rate near 50% with wide dispersion across years is evidence against a real seasonal effect, even if the absolute up-rate looks impressive.
Core Concepts
What is the difference between absolute and relative seasonality?
Absolute seasonality asks a simple question: in a given recurring calendar period (a month, a week of the year, a day around a holiday), was the asset's own return positive more often than chance would suggest? It is calculated directly from the asset's raw historical returns for that period, with no reference to anything else.
Relative seasonality asks a different, narrower question: in that same recurring period, did the asset's return exceed a chosen benchmark's return? It is calculated as the asset's return minus the benchmark's return for each historical instance of the period, then examined as its own distribution — a hit rate, an average, a standard deviation — independent of whether the raw asset return itself was positive or negative.
The distinction matters because equity markets have a well-documented long-run upward drift. A broad index has historically finished positive in a clear majority of calendar months across long sample periods, for reasons that have nothing to do with any specific month — economic growth, earnings growth, and the equity risk premium compounding over time. Any individual stock's absolute up-rate in a given month is measuring a mixture of two things: this market-wide drift, plus whatever asset-specific or seasonal effect might exist. Absolute seasonality cannot separate the two.
Why relative seasonality is the methodologically stronger test
Subtracting the benchmark return removes the shared market-wide component from each observation, leaving (approximately) only the asset-specific effect: general market conditions, sector-wide moves, and macro drift are common to both the asset and the benchmark in the same period, so they largely cancel out in the subtraction. What remains is closer to an estimate of whatever return pattern is unique to the asset itself during that calendar window — which is what a seasonality claim is actually supposed to be about.
This also makes relative seasonality more directly actionable for a trader who is choosing between holding the asset itself versus holding a broad market position, since the decision that matters is not "does this go up" but "does this beat the alternative I'd otherwise hold."
Relative seasonality is not a cure for every methodological problem in seasonality research. It still requires the same statistical scrutiny as any other seasonal claim: a large enough number of independent historical instances, out-of-sample validation where possible, and an honest accounting for how many other calendar periods or assets were tested before this one was reported (see multiple testing and researcher degrees of freedom). A relative-return test with 20 years of data is still a 20-observation test.
Worked Hypothetical Example
The following figures are an illustrative, hand-constructed hypothetical — not a real historical statistic for any specific ticker — used only to demonstrate the arithmetic of relative versus absolute seasonality.
- Absolute seasonality claim: "Stock XYZ has finished January positive in 14 of the last 20 years (70%)." Presented alone, this sounds like a strong seasonal edge.
- Benchmark context: Over the same 20 years, the broad market index finished January positive in 12 of 20 years (60%) — and finished any randomly chosen calendar month positive in roughly 12 of 20 years (60%) on average across all twelve months. The market's own baseline up-rate is already high in any given month, not just January.
- Relative calculation: For each of the 20 Januaries, subtract the index's January return from XYZ's January return. Suppose XYZ outperformed the index in 11 of the 20 Januaries (55%), with an average outperformance of +0.4 percentage points and a standard deviation of outperformance around 4 percentage points.
- Interpretation: A 55% relative hit rate with a small average edge and wide dispersion is close to what a coin flip would produce over 20 trials — it does not support a meaningful January-specific edge for XYZ once the market's own January tendency is accounted for. The headline 70% absolute up-rate mostly reflected the fact that January, like most months, tends to be positive for the broad market, not a distinctive XYZ-specific pattern.
- Contrast case: If XYZ had instead outperformed the index in 17 of 20 Januaries (85%) with a tight, consistently positive spread, that would be a materially more interesting relative result — worth further scrutiny (sample size, regime stability, and multiple-testing correction), but at least not confounded away by market drift in the first step.
Measurement Framework
| Measurement | What it tells you |
|---|---|
| Absolute up-rate for the period | How often the asset's raw return was positive — confounded with market drift. |
| Benchmark up-rate for the same period | The baseline against which the asset's up-rate should be judged. |
| Relative outperformance rate | How often the asset beat the benchmark in that period — the confound-adjusted hit rate. |
| Average relative return and standard deviation | The size and consistency of any asset-specific edge, once market-wide drift is removed. |
| Number of independent historical instances (N) | The real sample size the relative statistic is based on — typically the number of years in the dataset. |
Common Misconceptions and Failure Modes
Treating a high absolute up-rate as evidence of a seasonal edge
An asset finishing a given month positive 65–75% of the time over a couple of decades sounds compelling in isolation, but broad equity markets have historically finished a plurality of individual months positive as well. Without a benchmark comparison, there is no way to know whether the observed up-rate reflects anything specific to the asset or the calendar period at all.
Choosing the benchmark after seeing the result
If several candidate benchmarks (broad index, sector index, a peer group) are tested and only the one that produces the strongest relative result is reported, the benchmark choice itself becomes a form of researcher degrees of freedom, and the reported relative edge is inflated in the same way an unreported multiple-testing search inflates a strategy's significance. The benchmark should be selected on theoretical grounds — the natural alternative exposure an investor would otherwise hold — before the relative statistic is computed.
Assuming relative seasonality solves the sample-size problem
Subtracting a benchmark return removes a confound; it does not create additional independent observations. A monthly relative-seasonality test over 20 years of data has the same 20 independent yearly data points as the absolute version. The relative test is a better-specified question, not a bigger dataset, and still requires the same scrutiny for regime stability and statistical significance described in statistical significance in seasonality research.
Ignoring regime instability in the relative relationship itself
The relationship between an asset and its benchmark can itself shift across regimes — a stock's sector composition, its business mix, or its market-cap weighting relative to the index can all change over a 20-year sample. A relative seasonal pattern computed across a period spanning multiple structural regimes for the asset-to-benchmark relationship is measuring a moving target, not a stable effect.
Frequently Asked Questions
What is relative seasonality?
Relative seasonality tests whether an asset's return in a given calendar period exceeds the return of a benchmark over the same period, rather than testing whether the asset's own return was simply positive. It answers "does this asset outperform in month X" instead of "does this asset go up in month X," which isolates any asset-specific calendar effect from the market's general upward drift.
Why is absolute seasonality a weaker test than relative seasonality?
Absolute seasonality measures whether an asset was up or down in a given period without adjusting for the broad market's own tendency to rise over time. Because equity markets have historically risen in a majority of months regardless of season, an asset being up 70% of Januaries is unremarkable if the market is up roughly 60% of all months. The absolute win rate is confounded with long-term market drift, so it overstates the strength of any asset-specific calendar pattern.
How do you calculate relative seasonal return?
Relative seasonal return is calculated by subtracting the benchmark's return for a period from the asset's return for the same period, for every historical instance of that period, then examining the distribution of those differences. A positive average relative return with a high hit rate across many independent years is more meaningful than a positive average absolute return, because the benchmark subtraction removes the shared market-wide component.
Does relative seasonality eliminate the small-sample problem?
No. Relative seasonality removes the market-drift confound but does not increase the number of independent observations. A monthly seasonal pattern tested over 20 years of data still has only 20 independent yearly observations for that calendar month, which remains a small sample subject to the same multiple-testing and regime-instability concerns as absolute seasonality.
What benchmark should be used for a relative seasonality test?
The benchmark should represent the broad exposure the asset would otherwise be compared against — a broad market index for an individual stock, a sector index for a stock being tested against its sector, or a sector index against the broad market for sector-level seasonality. The benchmark choice should be fixed before running the test, not selected after seeing which benchmark produces the strongest result, since benchmark selection is itself a researcher-degrees-of-freedom choice.
Related Reading
- Sector and Industry Seasonality — a direct application of relative seasonality, since sector performance is typically measured against the broad market.
- Statistical Significance in Seasonality Research — how to evaluate whether a relative or absolute seasonal pattern is statistically meaningful given the sample size.
- Multiple Testing and Researcher Degrees of Freedom — why testing many calendar periods, assets, or benchmarks inflates the apparent strength of any single reported seasonal result.
Educational Disclaimer
This guide is for educational purposes only and does not constitute investment, financial, or trading advice. Seasonal patterns described here, including the worked example, are illustrative and hypothetical, not a recommendation to trade based on calendar effects. Historical patterns, including seasonal ones, are not reliable predictors of future returns and are subject to small sample sizes and regime change. Consult a qualified financial professional before making investment decisions. Trading involves significant risk of loss.