(From https://www.investopedia.com/terms/k/kratio.asp)

QUOTE:
The K-ratio takes into account the returns, but also the order of those returns in measuring risk. The calculation involves running a linear regression on the logarithmic cumulative return of a Value-Added Monthly Index (VAMI) curve. The results of the regression are then used in the K-ratio formula. The slope is the return, which should be positive, while the standard error of the slope represents the risk.

In 2003, Kestner introduced a modified version of his original K-ratio, which changed the formula of the calculation to include the number of return data points in the denominator. He introduced a further modification, which added a square root calculation to the numerator, in 2013.

Which version of the K-Ratio is calculated by WL? I am attempting to compare the WL result with another person's calculation and I think we are using two different versions of Lars's formula.

Thanks!
Vince

Apparently it's the 2003 version.

Thanks Eugene!

Is WL7 planning to use the most recent version that Lars published?

Vince

You're welcome Vince.

No, the Extended Scorecard in WL7 will use the existing version.

I would like to lobby that WL7's K-Ratio calculation match that being used in the rest of the industry to maintain an ability to compare results.

I realize that this is a LONE voice! :)

Vince

If there's a C# version somewhere you're aware of we may leverage it eventually.

I will look for one.

Vince

It appears that Lars most-recent "adjustment" in 2013 is a relatively simple modification to the original formula

https://papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID2230949_code1745920.pdf?abstractid=2230949&type=2

Vince

Apparently yes, it's a matter of multiplying the K-Ratio (2003) by Math.Sqrt(252)/TimeSeries.Count.

I believe that the Math.Sqrt(252) factor is just for a Daily Scale and that other appropriate factors would be required for Weekly and Monthly Scales.

Vince