Are IV values calculated by QuantLib scaled by a factor?

Posted by Morpheous on
URL: http://quantlib.414.s1.nabble.com/Are-IV-values-calculated-by-QuantLib-scaled-by-a-factor-tp270.html

I am using QuantLib to calculate IV values. I am suprised at the magbitude of the difference between my calculated historical volatility (HV) values, compared to the implied vol (IV) values calculated by QuantLib.

I am using QuantLib to calculate implied volatilities.

I am trying to understand the calculated figures (especially, when compared to historical volatility). The calculated implied volatility numbers are seldom below 0.5, whilst the historic volatility numbers* is never above 0.5.

Here is an output from my program:

DEBUG: 20 day historic volatility: 0.10
DEBUG: 20 day implied vol: 0.519485358338
DEBUG: 30 day historic volatility: 0.10
DEBUG: 30 day implied vol: 0.515820883886
DEBUG: 40 day historic volatility: 0.12
DEBUG: 40 day implied vol: 0.624451849413
DEBUG: 50 day historic volatility: 0.16
DEBUG: 50 day implied vol: 0.692403434852
DEBUG: 60 day historic volatility: 0.30
DEBUG: 60 day implied vol: 0.492372372425
DEBUG: 70 day historic volatility: 0.27
DEBUG: 70 day implied vol: 0.544712487074
DEBUG: 80 day historic volatility: 0.31
DEBUG: 80 day implied vol: 0.579945073422
DEBUG: 90 day historic volatility: 0.12
DEBUG: 90 day implied vol: 0.489174212819
DEBUG: 100 day historic volatility: 0.31
DEBUG: 100 day implied vol: 0.563068062254
DEBUG: 110 day historic volatility: 0.24
DEBUG: 110 day implied vol: 0.608231639138
DEBUG: 120 day historic volatility: 0.38
DEBUG: 120 day implied vol: 0.62748262992

The IVs are several multiples of the HV figures. I haven't looked under the bonnet and delved into the code just yet, but the consistently large difference (with IV constantly several multiples above HV) does suggest some scaling.

Finally, is a IV reading of 0.62748262992 to be interpreted as 0.6275% or 62.75% (the latter interpretation seems outrageously high)

*Note: Historic vols shown are annualized and calculated as the AVG square root of the variance of log returns.