In the HyperBitBit initial experiments slide, the table with the results and estimates is throwing me off guard in regard to the ratios.

From my quick calculations, I'm getting [1.03, 0.96, 0.96, 1.05] but the table specifically has [1.05, 1.03, 0.96, 1.03]

Am I missing something here? Specifically the 2M case has a positive ratio when the HBB estimate is higher than the actual count.