Calculating phase shift between IQ samples and Rotating constellation points

Hello everyone,

I captured some wi-fi signals and with the help of WLAN System Toolbox, decode them and try to regenerate the same packet again with waveformgeneration function of WLAN toolbox. The packet generation seems to be correct, since there is no bit error between captured and regenerated one. So if I get a reference symbols for my regenerated packet with the help of matlab and calculate EVM from this reference symbols and symbols of my regenerated packet should not the result be zero?In my case it is not zero what could be the reason? they seem to be rotated since when I plot the reference constellation of matlab and my regenerated one they are at the same points but evm is not zero.

Any help would be appreciated.

Kind Regards Seref