"Ok, lets say the market consists of 100 people. Microsoft has 90 people, Apple has 9 people."
ok so thats...
msft 90%
apple 9%
"Tomorrow, 6 more people join. 5 of them go to Microsoft, 1 to Apple. Microsoft now has 96/106 people, Apple has 10/106."
hang on, microsoft should have 95/106!
and apple 10/106.
ignoring that..
OK I see where the confusion lies, I think you are looking at market share as accumulated over a period of time, which is ok, but then you are also comparing daily rise and falls.
You should really compare by the same timeframe otherwise it's possible to skew the comparison in favor of one or the other due to microblips in the stats.