Terminology:
Percent Correct
At the 50% Correct value, there is a 50/50 chance the forecast value will be within this margin of error.
At the 80% Correct value, there is a 80% chance the forecast value will be within this margin of error.
The potential range of a forecast's value is found by taking the published forecast value and both adding and subtracting the % Correct Values.
The % Correct (or error) values published are based on prior forecast performance.
For Example:
Forecast Value = 100
50% Correct Value = 10
80% Correct Value = 15
There is a 50% Chance the actual value will be between 110 and 90.
There is a 80% Chance the actual value will be between 115 and 85.
Technical Note: The Financial Forecast Center has moved away from publishing standard deviations of the forecast's performance in recognition that the distribution of value movements in the financial markets follow Levy or Cauchy distributions, not Gaussian or normal distributions. Likewise, the forecast model's errors follow similar distributions.
A Gaussian distribution significantly underestimates the probability of a large price or rate movement. A Gaussian distribution may underestimate the probabilty of a 3 sigma price movement by a factor of 10. In other words, the chance of a 3 sigma movement is potentially 10 times greater than that predicted by a Gaussian probability curve.
The above change in error reporting enables a more accurate depiction of a forecast model's potential performance.
source: Financial Forecast Center, LLC
0 komentar:
Posting Komentar