Compounds, Integration, and Log-Returns

Study notes deriving continuously compounded (log) returns from discrete compounding, why log returns are time-additive, and how they relate to the force of interest.

Summary

Log returns, defined as

$$

Rlog⁡=ln⁡ ⁣(PtPt−1),R_{\log} = \ln!\bigl(\tfrac{P_t}{P_{t-1}}\bigr),

$$

are exactly the continuously compounded returns for an asset over the period from t−1t-1 to tt, since they arise as the limit of discrete compounding when the compounding frequency goes to infinity . In practice, this means that if you reinvest gains instantaneously at every moment, the growth factor is

$$

Pt=Pt−1 eRlog⁡,P_t = P_{t-1},e^{R_{\log}},

$$

and hence the log return Rlog⁡R_{\log} is the natural rate that delivers the observed price change under continuous compounding . Moreover, log returns coincide with the “force of interest” or instantaneous rate of growth, enjoy time‑additivity (so multi‑period log returns simply sum), and yield symmetric treatment of gains and losses .


1. Simple vs. Continuously Compounded Returns

1.1 Simple (Discrete) Return

  • Definition: The simple (or holding‑period) return RR for a single period is

$$

R=Pt−Pt−1Pt−1,R = \frac{P_t - P_{t-1}}{P_{t-1}},

$$

where Pt−1P_{t-1} and PtP_t are the asset prices at the beginning and end of the period, respectively .

  • Discrete Compounding: If interest or returns are reinvested mm times per year at a nominal rate rr, the accumulation over one year is

$$

(1+rm)m,(1 + \tfrac{r}{m})^m,

$$

converging to continuous compounding as m→∞m \to \infty .

1.2 Continuously Compounded Return (Log Return)

  • Definition: The continuously compounded return (also called the log return or force of interest) is

$$

Rlog⁡=ln⁡ ⁣(PtPt−1),R_{\log} = \ln!\bigl(\tfrac{P_t}{P_{t-1}}\bigr),

$$

which directly measures the instantaneous rate that transforms Pt−1P_{t-1} into PtP_t under continuous compounding .

  • Accumulation Relation: Under continuous compounding, if the constant rate is rlog⁡r_{\log}, then

$$

Pt=Pt−1 e rlog⁡Δt⟹rlog⁡=1Δtln⁡ ⁣(PtPt−1)P_t = P_{t-1} , e^{,r_{\log} \Delta t}

\quad\Longrightarrow\quad

r_{\log} = \tfrac{1}{\Delta t}\ln!\bigl(\tfrac{P_t}{P_{t-1}}\bigr)

$$

.


2. From Discrete to Continuous: The Limit Argument

Consider discrete compounding over one period t=1t=1 with rate rr and subperiods mm:

$$

(1+rm)m  →m→∞  er.\bigl(1 + \tfrac{r}{m}\bigr)^{m}

;\xrightarrow[m\to\infty]{};

e^{r}.

$$

Hence, if you observe a total simple return RR in one period, the equivalent continuous rate rlog⁡r_{\log} satisfying erlog⁡=1+Re^{r_{\log}} = 1+R is

$$

rlog⁡=ln⁡(1+R).r_{\log} = \ln(1 + R).

$$

Applied to prices,

$$

R=PtPt−1−1⟹Rlog⁡=ln⁡ ⁣(1+R)=ln⁡ ⁣(PtPt−1).R = \frac{P_t}{P_{t-1}} - 1

\quad\Longrightarrow\quad

R_{\log} = \ln!\bigl(1 + R\bigr) = \ln!\bigl(\tfrac{P_t}{P_{t-1}}\bigr).

$$

This limit is the mathematical foundation of why log returns are continuously compounded returns .

3. Interpretation as the “Force of Interest”

  • Force of Interest: In actuarial science, the instantaneous “force of interest” δ(t)\delta(t) for an accumulation function a(t)a(t) is $$

δ(t)=a′(t)a(t)=ddtln⁡a(t).\delta(t) = \frac{a’(t)}{a(t)} = \frac{d}{dt}\ln a(t).

$$

For constant rate rr, a(t)=erta(t)=e^{rt} and δ=r\delta=r. Hence, δ\delta is exactly the continuously compounded rate .

  • Connection to Log Returns: Over a short interval Δt\Delta t, δ Δt≈ln⁡ ⁣(PtPt−Δt)\delta ,\Delta t \approx \ln!\bigl(\tfrac{P_{t}}{P_{t-\Delta t}}\bigr), tying the instantaneous force directly to the log return between two instants .

4. Key Properties of Log Returns

  1. Time‑Additivity

If an asset experiences successive log returns Rlog⁡,1R_{\log,1} and Rlog⁡,2R_{\log,2}, the total log return over both periods is

$$

Rlog⁡=Rlog⁡,1+Rlog⁡,2;R_{\log} = R_{\log,1} + R_{\log,2}, since

$$

$$

ln⁡ ⁣(P2P0)=ln⁡ ⁣(P2P1)+ln⁡ ⁣(P1P0).\ln!\bigl(\tfrac{P_2}{P_0}\bigr)

= \ln!\bigl(\tfrac{P_2}{P_1}\bigr)

  • \ln!\bigl(\tfrac{P_1}{P_0}\bigr).

$$

This additive property simplifies multi‑period analysis .

  1. Symmetry

Log returns treat percentage gains and losses symmetrically: a +50% move (Rlog⁡=40.5%R_{\log}=40.5%) followed by −40.5% (Rlog⁡=−50%R_{\log}=-50%) brings you back to the start, unlike simple returns .

  1. Normality Assumption

When asset prices are modeled as following a lognormal distribution, log returns are normally distributed—a cornerstone of the Black–Scholes framework .

5. Practical Usage

  • Volatility Estimation: Historical volatility is usually computed from the standard deviation of log returns, because of their additive and normal‑like behavior over time .
  • Option Pricing: Continuous compounding underpins risk‑neutral pricing and discounting cash flows in derivatives models, making log returns the natural choice for modeling price evolution .

In essence, log returns are by construction the continuously compounded returns: they emerge from taking the limit of discrete compounding and coincide with the instantaneous rate of growth, offering mathematical elegance and powerful properties for financial modeling.

Volatility

$$ \sigma_{anual} = \sigma_{daily}\sqrt{PT} $$

Usually, $\sqrt{PT}$ can be denoted as $\sqrt{252}$ since 252 is the trading days of a year.

Powered by Ignorance.