The Mean Value Theorem

If a stock's price starts the day at $100 and ends at $106, its average rate of change over the day is $6. It's tempting to think that says nothing about any specific instant — after all, the price could have zigzagged wildly in between. But if the price path is smooth and unbroken (no gaps, no sudden jumps), a surprisingly strong guarantee holds: at some exact instant during the day, the price's instantaneous rate of change must have been exactly $6 per day too — not approximately, exactly. The mean value theorem is what proves this must happen.

An analogy: a speed camera can't be fooled by average speed

Drive 60 miles in one hour and your average speed is 60 mph. If a stretch of road is monitored so it's known your speed varied continuously — no teleporting, no instant speed changes — then at some exact moment during that hour, your speedometer must have read exactly 60 mph, even if you spent part of the time going faster and part going slower. You can't average 60 mph over a continuously-varying trip without hitting 60 mph exactly at least once. The mean value theorem is the rigorous version of this everyday fact, stripped of any reference to cars or speed and stated purely in terms of functions and derivatives.

The math, one symbol at a time

If $f$ is continuous on the closed interval $[a,b]$ and differentiable on the open interval $(a,b)$, then there exists at least one point $c$ in $(a,b)$ such that

$$f'(c) = \frac{f(b) - f(a)}{b - a}.$$

In words: somewhere strictly between $a$ and $b$, the function's instantaneous rate of change (the left side, a derivative) exactly matches its overall average rate of change across the whole interval (the right side, a simple slope-of-the-secant-line calculation). The theorem doesn't tell you where $c$ is — only that it must exist, as long as the function is smooth enough (continuous, and differentiable in the interior) over the interval in question.

Worked example 1: finding the guaranteed point exactly

Let $f(x) = x^2$ on $[1,5]$. The average rate of change is $\frac{f(5)-f(1)}{5-1} = \frac{25-1}{4} = 6$. The theorem guarantees some $c \in (1,5)$ with $f'(c) = 6$. Since $f'(x) = 2x$, solve $2c = 6$, giving $c=3$ — right in the middle of the interval in this symmetric case, and indeed strictly between 1 and 5 as promised. Checking: $f'(3) = 6$, matching the average rate exactly.

Worked example 2: bounding a price move without knowing the exact path

A stock's price function is known to have derivative (instantaneous rate of change) bounded between $-2$ and $3$ dollars per day at all times over a 5-day stretch — a limit on how fast the price can move at any given instant, perhaps from a circuit-breaker-style trading rule. By the mean value theorem (applied in reverse, as a bound), the price's average rate of change over the 5 days must also fall between $-2$ and $3$ (since it equals $f'(c)$ for some $c$ in the interval, and $f'$ never leaves that range). So the total price change over 5 days must lie between $-10$ and $15$ dollars — a firm bound derived purely from a limit on the instantaneous rate, without ever tracking the actual path the price took.

Pick two points on the curve and note the slope of the line connecting them (the average rate); somewhere between them, the curve's own tangent slope must match it exactly — try to spot where visually as the curve bends.

What this means in practice

The mean value theorem is the bridge between "the derivative at a point" and "the average behavior over a range," and it underpins error bounds used in numerical methods (bounding how far an approximation can be off, given a bound on the derivative), and it's the proof engine behind other results quants rely on, including why a function with a derivative that's always zero must be constant, and pieces of the reasoning behind Taylor series error terms (see Taylor Series Expansion).