What it means
Most business measures move around a central level rather than wandering off in one direction forever. A restaurant group's food cost percentage, a software firm's win rate and a factory's scrap rate all have a normal band, and unusual readings often reflect temporary conditions such as weather, a single large order or a one-off supplier problem.
Mean reversion is the name for the pull back towards that normal band. This matters because managers routinely treat one exceptional quarter as the new baseline.
If a sales team beats target by 40% because a large contract landed a month early, building next year's plan on that figure sets everyone up to miss. The same trap works in reverse, where a single terrible month triggers panic cost cuts that damage a business whose underlying performance was fine.
Using the idea in practice starts with working out what the long-run average actually is, usually from three to five years of data covering at least one full business cycle. You then ask how quickly the metric has historically closed the gap between an extreme reading and that average, which analysts call the speed of reversion.
Finally you fade the current extreme part of the way back rather than assuming it either persists or corrects fully. The important nuance is that averages themselves can shift permanently.
If a competitor exits the market, a tax rate changes or a business moves from one-off sales to subscriptions, the old average is no longer the right anchor and waiting for reversion becomes an expensive mistake. Telling a temporary swing apart from a structural break is the hard part, and it is a matter of judgement rather than arithmetic.
Mean reversion is also the reason careful forecasters distrust straight-line extrapolation of any recent trend. Compounding an unusually strong growth rate for five years produces a number that almost never arrives, which is why sensible plans fade growth towards an industry norm.
The same discipline applies to margins, where competition tends to grind unusually high returns back towards what everyone else earns.
In practice
Real-world examples.
Example
An online homeware retailer sees its product return rate jump from a steady 8% to 14% in December. Rather than rebuilding the whole returns process, the operations lead notes that December always runs hot because of gift purchases and forecasts a return to roughly 9% by March. The rate lands at 8.6%, and the team avoids a costly overreaction.
Example
A fund manager reviews a listed engineering group whose shares fell 30% after one weak half-year result. Because the company's return on capital has averaged 14% across a decade and nothing structural has changed, the manager treats the weak half as a temporary dip and buys, expecting profitability to revert.
Example
A subscription fitness app records monthly churn of 2.1% in January against a two-year average of 3.4%. The finance team resists raising the annual revenue forecast, since January signups are unusually motivated, and models churn drifting back to about 3.1% by mid-year.
Think of it
“Mean reversion is things returning to normal-extreme values moving back toward average.
Formula
Calculation
Expected next value = Long-run average + (Current value - Long-run average) x (1 - Speed of reversion)
A specialist distributor has earned a gross margin averaging 42% over the past five years. This year the margin came in at 36% because of a freight cost spike, and the firm's history suggests roughly half of any gap closes within a year, giving a speed of reversion of 0.5.
Gap to the average = 36% - 42% = -6 percentage points.
Faded gap = -6 x (1 - 0.5) = -3 percentage points.
Expected margin next year = 42% + (-3) = 39%.
Running the same step again gives 42% + (39% - 42%) x 0.5 = 42% - 1.5% = 40.5% for the following year. On revenue of $20,000,000, the difference between planning at 36% and planning at 39% is 3% x $20,000,000 = $600,000 of gross profit, which is why the assumption is worth arguing about.Case study
Seen in the real world.
In this illustrative example, Harborline Coffee Roasters, a fictional wholesale roaster, posted a gross margin of 49% in a year when green coffee prices collapsed. The founder took the number as proof that a new sourcing strategy had permanently changed the economics and used it to justify hiring six extra sales staff and signing a larger warehouse lease.
The following year green coffee prices returned to their usual range and the margin fell to 41%, close to the 40% the business had averaged for six years. The extra fixed costs did not revert, so operating profit fell by more than the margin change alone would suggest, and Harborline spent two quarters unwinding commitments.
The finance director rebuilt the planning model afterwards so that any margin more than three percentage points away from the six-year average was automatically faded halfway back before it fed the hiring plan. The change was unglamorous, but it stopped one lucky year from being treated as permanent.
Watch out
Common mistakes.
- Treating mean reversion as a guarantee that a falling share price must bounce back, when a genuine decline in the business means the old average no longer applies.
- Calculating the long-run average from too short a period, so that a two-year average simply captures the same temporary conditions you are trying to correct for.
- Applying reversion to a metric that has a real trend, such as unit costs in a business steadily gaining scale, where the average is moving rather than fixed.
Questions
People also ask.
How fast should I expect a metric to revert?
It depends entirely on the measure, so estimate the speed from your own history rather than assuming a single year, since margins often take two or three years while daily traffic can normalise in weeks.
Is mean reversion the same as regression to the mean?
They are close cousins, with regression to the mean describing what happens statistically when you re-measure an extreme observation, while mean reversion usually describes a series moving back towards its average over time.
Can I use this in a budget?
Yes, and the simplest way is to fade any unusually strong or weak input part of the way towards its multi-year average before it drives headcount, pricing or capital spending decisions.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%Related