Forecasting Toolkit
Diagnostic tools to test the health of your process Forecast Bias Indicator
What
Gives you rating of th
Forecast Run Indicator
in a set of f into accoun
How
Works ou the avera
What it
Gives you a si test for identify
Forecast Error Ratio
Single point forecasts
fo re casts
with a pro pr o duc i n g
How used?
To quickly
set of fore (complian
Policy).
forecast proce exhibiting bias.
How it
Sets out the nu con secu tive f o
the same sign
indicative of th bias.
How m used?
As a „rule of th
identifying that process is goin
What it
Gives you a si test for forecas
Forecast Quality Indicator
What it does
What it does it look like?
How it
Gives you an objective rating of the quality of a
Sets out the m positive/negati
forecast process.
errors consiste
unbiased forec level of confide
How it does it
7
5
How mi used?
As a „rule of th
identifying fore which are bias
Comparing the spread of forecast data with the spread
of actual data. The greater
the spread of the forecast the poorer the forecast process.
How might it be used?
fScore
3
Goog Low level of volatility
Poor Moderate level of voatility
Poor Moderate level of volatility
To give a fuller measure of
1
forecast performance. To identify where the forecast
process (as opposed to behaviours around forecasts)
4
8
12
16
20 Sample Size
24
28
32
needs to be improved.
Range Forecast Error Indicator
What it does
Gives objective guidelines for Range Forecast
Range Forecast Quality Indicator
accuracy.
What it does
What it does it look like?
How it does it
Works out what percentag outcomes should fall withi
the middle 50% of the ran
Gives you an objective rating of the quality of a
Range Forecasting process by taking into account the
1.00
0.95
Probable Wide Range
Range forecasts
forecast if the Range Forecast is to be conside
sou nd .
How might it used?
To give an early warning
failures in the Range Forecasting process.
sample size
How it does it
Works out the probability of a sample of Range Forecasts
being consistent with a
process capable of meeting the target rate (90%)
How might it be used?
Success Rate
0.90
0.85
0.80
0.75
0.70
0.65
0.60
Probably Ok
Poor
Significantly Poor
To quickly identify whether the
Range Success Rate is acce pt abl e
0.55
4
8
12
16
20 Sample Size
24
28
32
36
36
Tools to help build forecasts
Range Forecasting Methodology
What it does
Gives you.
Statistical Forecasting Methodology
What it does
What it does it look like?
How it does it
Works out.
Creates a statistically based
forecast  using a robust Excel based methodology
Qua te ly Overhead
R y=1.20L(x)+3.75 2 = 0 . 4 3
How it does it
12.0
How might it be sed?
o
qu
Uses Mathematical and
Statistic e un
s to r
10.0
.0
i
4.0
wh
ailed
s
tr
or
re
cr
ey
e
r en
d et
cas
n pr
)
s
To
„baselineā for
2.0
0.0
1 2 3 4 5 6 7 8 9 10 1112 13 1415 16 1718 19 2021 22 2324 25 2627 28 2930 31 3233 34 3536 Qua ter (Code )
Quartery ctul
4MA LL
UL
Log. (4M )
Tools to help you manage your process Forecast Bias Performance Guide
What it does
Gives you the ability to s targets for forecast bias
Forecast Error Chart
given a desired level of confidence
How it does i
An Excel tool which, taki
into account the historica volatility of forecasts erro
works out the range of
What it does
Gives you a set of ranges (based on historic forecast
accuracy) which represent limits within which unbiase
Forecast Credibility Corridor
acceptable values for ne for a set sample size and
forecasts should fall
What it does
What it does it look like?
specified degree of con fi de nce .
How might it used?
To set realistic and reas ta rg ets.
How it does it
Based on historic forecast error works out the ranges
within which 90% of
individuals forecasts shoul fall if there is no bias
How might it b
Gives you a set of ranges (based on historic trends)
within which your forecast should fall  with a defined
set of probabilities
How it does it
Takes a Moving Annual Total and works out typical
10.0
5.0
ig y i i ic
ozenFood ot y oer 19972005
Op ti mi sti c
bi
rept into your
f
rocess. To test
bi
wh et her
ecasts you receiv
rates of change for this
m ar ket/ bu sin ess o ver di ff er en t h or izo ns
How might it be used?
r o er I cre s / ecre se
0.0

5.0

10.0
ig y esi i ic
Li ke ly
y=0.0x0.06
Pessimistic
To test the credibility of
forecasts you produce To test the credibility (and
123
4
5
6
7
89 oso d
10
11
12
13
14
15
16
potential bias) of forecasts you receive
To focus attention on the
significant elements of fo re casts
N
68%CL
99.97%CL
er
)