# ACF - Autocorrelation Function

Calculates the sample autocorrelation function (ACF) of a stationary time series.

## Syntax

ACF(X, Order, K, Method)
X
is the univariate time series data (a one-dimensional array of cells (e.g. rows or columns)).
Order
is the time order in the data series (i.e. the first data point's corresponding date (earliest date=1 (default), latest date=0)).
Order Description
1 ascending (the first data point corresponds to the earliest date) (default)
0 descending (the first data point corresponds to the latest date)
K
is the lag order (e.g. 0=no lag, 1=1st lag, etc.). If missing, the default lag order of zero (i.e. Lag=0) is assumed.
Method
is the calculation method for estimating the autocorrelation function (0= Sample Autocorrelation (Default), 2=Periodogram-based estimate , 2=Cross correlation).
Value Method
0 Sample autocorrelation method.(default)
1 Periodogram-based estimate. method
2 Cross-correlation method

## Remarks

1. The time series is homogeneous or equally spaced.
2. The time series may include missing values (e.g. #N/A) at either end.
3. The lag order (k) must be less than the time series size, or else an error value (#VALUE!) is returned.
4. The ACF values are bound between -1 and 1, inclusive.
5. Method 1: The sample autocorrelation is computed as:$$\hat{\rho}(h)=\frac{\sum_{k=h}^T{(y_{k}-\bar y)(y_{k-h}-\bar y)}}{\sum_{k=1}^T(y_{k}-\bar y)^2}$$Where:
• $y_{t}$ is the value of the time series at time t.
• $h$ is the lag order.
• $T$ is the number of non-missing values in the time series data.
• $\bar y$ is the sample average/mean of the time series.
$\bar y=\frac{\sum_{i=1}^{N} y_i}{N}$
6. Note that we subtract the full sample mean $\bar y$
7. Method 2: Periodogram estimate.
In this method, we compute the spectral density (periodogram is an estimator) of the sample data set, and use it to compute the sample auto-correlation.
8. Although the ACF estimate using periodogram-based is usually biased, it generally exhibits a smaller standard error.
9. Method 3: cross correlation method :$$\rho(h)=\frac{\sum_{i=1}^{N-h}(y_i-\bar y)\times (y_{i+h}-\bar y_h)}{\sqrt{\sum_{i=1}^{N-h}(y_i-\bar y)^2 \times\sum_{j=h}^N (y_j-\bar y_h)^2}}$$Where:
• $y_{t}$ is the value of the time series at time t.
• $h$ is the lag order.
• $\bar y=\frac{\sum_{i=1}^{N-h} y_i}{N-h}$
• $\bar y_h=\frac{\sum_{i=h}^N y_i}{N-h}$
• $T$ is the number of non-missing values in the time series data.
• $\bar y$ is the sample average/mean of the time series.
10. Special Cases:
• By definition, $\hat{\rho}(0) \equiv 1.0$

## Examples

Example 1:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
A B
Date Data
1/1/2008 #N/A
1/2/2008 -1.28
1/3/2008 0.24
1/4/2008 1.28
1/5/2008 1.20
1/6/2008 1.73
1/7/2008 -2.18
1/8/2008 -0.23
1/9/2008 1.10
1/10/2008 -1.09
1/11/2008 -0.69
1/12/2008 -1.69
1/13/2008 -1.85
1/14/2008 -0.98
1/15/2008 -0.77
1/16/2008 -0.30
1/17/2008 -1.28
1/18/2008 0.24
1/19/2008 1.28
1/20/2008 1.20
1/21/2008 1.73
1/22/2008 -2.18
1/23/2008 -0.23
1/24/2008 1.10
1/25/2008 -1.09
1/26/2008 -0.69
1/27/2008 -1.69
1/28/2008 -1.85
1/29/2008 -0.98

Formula Description (Result)
=ACF(\$B\$2:\$B\$30,1,1) Autocorrelation of order 1 (0.235)
=ACF(\$B\$2:\$B\$30,1,2) Autocorrelation of order 2 (-0.008)
=ACF(\$B\$2:\$B\$30,1,3) Autocorrelation of order 3 (0.054)