Ok so somebody further up the thread said that I could replace the host with a computer program so I just did that.
The rules I used are these.
1) car is placed randomly in one of three places
2) The first guess is made as 1, 2 or 3
3) Host chooses one of the unchosen places and eliminates it
4) contestant either sticks or changes
5) results are evaluated and spit out as count rates
Once put into code form it becomes pretty obvious what the result will be without even running the simulation.
The logic is inescapable.
First of all there is one thing to check
1) Was my guess correct? (remember there are only 2 choices now since the host already opened a curtain/door that wasn't the car)
YES (1 in 3 chance) --> Sticking wins, changing loses so add 1 to StickCount
NO (2 in 3 chance) --> Sticking loses, changing wins so add 1 to ChangeCount
The results of 1 million games comes out as follows
StickCount = 332654
ChangeCount = 667346
Until putting this into code form I would have sworn that the chances were 50:50 but as soon as I wrote the problem in code form it's obvious that changing is going to win twice as often as sticking