Topics

Basics of price returns calculation – Quant Finance

31 March 2020

In financial terms, a return is nothing but that the amount of money that can be made or lost on a given investment. In this post, we briefly see basic concepts about the calculations of the price variations of whatever asset quoting (in some exchange) and changing in price over time.

Percentage price returns

Suppose we are looking at the price of certain stock where we have some money invested in. What is the definitions of its price price returns? This is going to be how much we gain, or loose, from the price variation from $t$ to $t+1$.

Let $P_0$ and $P_1$ be our stock price at time $t_0$ and $t_1$, respectively. To move from $P_0$ to $P_1$, the stock has moved by a quantity denoted here by $R_{0,1}$, i.e., we can write $P_1$ as: $$ P_1 = P_0 + R_{0,1}P_1 = P_0(1 + R_{0,1}). $$ Therefore, the percentage price return is given by $$ R_{0,1} := \frac{P_{1} - P_0}{P_{0}} = \frac{P_1}{P_0} - 1. $$ For example, if $P_0=\$100$ and $P_{1}=\$104$, then $R_{0,1}=104/100-1=0.04=4\%$. If we were long the stock, we gained $4\%$ of our initial investment.

Compound return

If we now look at two time periods, $(0,1)$ and $(1,2)$, the overall return is not just given by the sum of the returns over the two periods. For example, consider the price variations: $$ P_1 = P_0 + R_{0,1}P_0 \qquad\text{and}\qquad P_2 = P_1 + R_{1,2}P_1. $$ where $R_{0,1}$ and $R_{1,2}$ denote the percentage price returns during the periods $(0,1)$ and $(1,2)$, respectively. By substituting $P_1$ into $P_2$: $$ P_2 := (P_0 + R_{0,1}P_0) + R_{1,2}( P_0 + R_{0,1}P_0 ) = P_0 + R_{0,1}P_0 + R_{1,2}P_0 + R_{1,2}R_{0,1}P_0. $$ Hence, diving by $P_0$, the overall return over the total period $(0,2)$ is given by: $$ R_{0,2} = \frac{P_2}{P_0} - 1 = 1 + R_{0,1}+R_{1,2}+R_{1,2}R_{0,1} - 1 = (1 + R_{0,1})(1 + R_{1,2}) - 1. $$ This is going to be a compound return which is the rate of return of the total time period expressing the cumulative effect of the two returns (positive/negative) $R_{0,1}$ and $R_{1,2}$.

The definition can be easily generalized. Given a certain time period $(t, t+k)$, with $k>1$, the corresponding compound return is given by; $$ R_{t,t+k} := \frac{P_{t+k}}{P_t} - 1 =(1+R_{t,t+1})(1+R_{t+1,t+2})\cdots(1+R_{t+k-1,t+k}) - 1. $$ For example, suppose we buy a stock that on the first day goes $10\%$ up whereas on the second day goes $3\%$ down. At the end of the second day, our total (compound) return is $$ R_{0,2} = \frac{P_2}{P_0} - 1 = (1 + R_{0,1})(1 + R_{1,2}) - 1 = (1 + 0.10)(1 - 0.03) - 1 = 0.066999 = 6.7\%. $$ As we can see, this is not simply $10\%$ minus $3\%$, i.e., $7\%$, but it is something lower$\dots$

Note that if the returns are the same over all time periods, say $R$, then the compound return over the entire period becomes: $$ R_{t,t+k} = (1+R)^{k} - 1. $$

Return per month

Suppose we have a set of monthly returns, $(R_{0,1}, R_{1,2}, \dots)$, where each timeframe, $(0,1), (1,2), \dots$ represents a time period of a month.

In the example above, for the period $(0,2)$ we had two months, with prices $P_0$ (at the beginning of the first month), $P_1$ at the end of the first month, and $P_2$ at the end of the second month.

We saw that the compound (total) return after two months is $R_{0,2} = (1+R_{0,1})(1+R_{1,2})-1$, where $R_{0,1}$ and $R_{1,2}$ are the returns of the first and the second month, respectively.

What is now the return per month, that is, the fixed return for each month that will give the same total return $R_{0,2}$? That is, we are asking for the value $R_m$ such that: $$ R_{0,2} = (1 + R_{m})^2 - 1, $$ where we have used the simplified formula since $R_{2}$ is going to be the same for the $2$ months considered. We get: $$ R_{m} = (1 + R_{0,2})^{1/2} - 1 = ( (1+R_{0,1})(1+R_{1,2}) )^{1/2} - 1. $$ In general, in the equation above, we divide by the number of given months (in this case, $2$).

Return per year

If we have a fixed return per month, we can obtain the corresponding return per year, or annualized return, i.e., the compound return after $12$ months: $$ R_{y} := (1 + R_{m})^{12} - 1. $$ On the other hand, if we don't compute the return per month but we have, however, a set of monthly returns, we can infer the return per year from their total compound return.

For example, by using the formula for $R_{0,2}$ from the first two months, we get $$ R_{y} = \left(1 + \underbrace{R_{m}}\right)^{12} - 1 = \left(1 + \underbrace{(1+R_{0,2})^{1/2} - 1}\right)^{12} - 1 = (1 + R_{0,2})^{12/2} - 1. $$ In general, given a series of $n$ monthly returns, there holds $$ R_{y} = (1 + R_{0,n})^{12/n} - 1, $$ where $R_{0,n}$ denotes the compound (total) return over all available months.

That is, if we had a compound return $R_{0,6} = 5\%$ for a 6-months period, this is a return per year of $R_y = (1 + 0.05)^{12/6}-1 = 10.25\%$. On the other hand, if we had $R_{0,24} = 10\%$ for a 24-months period, then it means we have a return per year of $R_y = (1 + 0.10)^{12/24}-1 \approx 4.8\%$.

General formula for annualizing returns

The previous formula can be generalized to be used for any kind of available returns that we may have, e.g., daily, weekly, or monthly returns. The difference would be in the numerator of the power.

For example, suppose we have a series of daily returns, then the annualized return is given by: $$ R_{y} = \left(1 + \underbrace{R_{d}}\right)^{252} - 1 = \left(1 + \underbrace{ (1+R_{cp})^{1/N} - 1}\right)^{252} - 1 = (1 + R_{cp})^{252/N} - 1, $$ where again, $R_{cp}$ is the compound (total) return of the series of daily returns that we have, $N$ is the lenght of the series of returns, and $252$ is the number of (trading) days in one year.

In general, given a set of $N$ returns and $R_{cp}$ being the compound return: $$ R_{y} = (1 + R_{cp})^{p/N} - 1, \quad\text{where}\quad p = \begin{cases} &252 & \text{if we have daily returns},\\ &52 & \text{if we have weekly returns},\\ &12 & \text{if we have monthly returns.} \end{cases} $$ The variable $p$ can be called the periods per year.

Nominal Interest Rate

Previously, we have seen that given a fixed return $R$, the compound return after a period of length $t+k$ (starting from $t$) is given by: $$ R_{t,t+k} = (1+R)^{k} - 1. $$

For example, suppose that we borrow some money, say $P := \$1$, during one year at an annual interest rate of $r := 10\%$. Also, suppose that we have to pay back the money in one payment at the end of the year. Then, the return after one year is $10\%$ and we will have to pay back: $$ P \cdot (1 + r) = 1 \cdot (1 + 10\%) = \$1.1. $$

Now, suppose that we still borrow $P := \$1$ at the same interest rate $r := 10\%$ but now we have to make two semestral payments. In this case, the second interest rate to pay is accumulated on top on the first interest rate paid after six months.

In other words, the rate $r=10\%$ is split into two semi-annual rates $r/2 = 5\%$. Hence, the total compound return after one year is: $$ \left(1 + \frac{r}{2}\right)\left(1 + \frac{r}{2}\right) - 1 = \left(1 + \frac{r}{2}\right)^2 - 1 = \$0.1025, $$ This means that we will have to pay back $1 + 0.1025 = \$1.1025$.

If payments were instead due every month, we have: $$ \left(1 + \frac{r}{12}\right)\cdot\;\; ...\;\; \cdot \left(1 + \frac{r}{12}\right) - 1 = \left(1 + \frac{r}{12}\right)^{12} - 1 = \$0.1047, $$ and then we have to pay back $1 + 0.1047\cdot 1 = \$1.1047$.

We can see that due to compounding, the more payments due over the entire period the larger the total compound return, hence more money have to be paid back.

In general, given a nominal interest rate $r$ (also called the instantaneous interest rate) and $N$ periods (i.e., the number of payments, of the investment, loan, and so on), the total return is given by: $$ R = \left(1 + \frac{r}{m}\right)^{mt} - 1. $$ where $t$ is the total period in years ($t=1$ year in our examples). Such a return $R$ is nothing but that the annualized return (as we used to call it so far) or the effective annual interest rate obtained from discrete compounding.

Continuous compounding

What if the discrete compounding becomes continuous? If $m$ tends to infinity, we have: $$ \lim_{m\to \infty} \left(1 + \frac{r}{m} \right)^{mt} = e^{rt}, $$ and then $$ R = \lim_{m\to \infty} \left(1 + \frac{r}{m} \right)^{mt} - 1 = e^{rt} - 1. $$ See here.