NuclearPhynance Archive

((In begging voice.))

Me too me too!

Anyway, I hand delivered the paper to Nero, but you need to give me the strats on the sample return histories on which you want me to try the model. What I can say is the first is potentially a candidate.

If you would be so kind, I'd love to see the paper, please... Thanks

again intriguing case...



The Dangers of Japan's Corporate Pension Funds



Anyone ever saw the track records?

Nope, but I'd like to test the detector on them.

I would like to see the results Wink

Had to drag up an old thread to ask if these returns seem fishy?



edit: deleted the wall of text, here's a google sheet link: https://docs.google.com/spreadsheets/d/1jx77vouWsG-kDUccwvzxV8v6C4wRiOS0hDHf6ZWb7kg/edit?usp=sharing

Definitely a bit odd. The following stats



Data 1-50 50 - 100 1 - 100

Benford Bias 3.29 2.45 3.71

Mu 1.67 1.96 1.45

Vol 14.5 11.5 13.1

Skew 0.1 -0.75 -0.12

Kurt 0.58 2.36 1.2



There are no data points ( returns ? ) in the + 7 to +10% range over any window period which suggests that data has been removed - but why delete 'good' data



The Kurtosis looks too low to be financial data to to me but not impossible ( knowing the dates would help ).



Vol appears to be mean reverting in the range of 5 - 19 %.



Rolling 12m CAGR falls from over 100% to 50% to 10%.



If it is financial data and I had to hazard a guess I would say it could be back-tested data for the early period and the lower return out of sample numbers. Doesn't explain the missing +7 to +10% range of data though. The fact that that is consistently missing from both periods suggests the data was generated using some sort of random number generator with the missing range excluded and some noise.

An actual fund, they claim them as actual performance. Should have mentioned those are monthly returns.

Dates and currency ?