Valorant Statistics Help

From a large valorant dataset, I know that the average valorant game lasts on average 20.5 rounds with a median score of 13-7.
I am trying to set up a monte carlo simulation under the presupposition that each team has a roughly even probability of winning each round, but this produces a median score of 13-12 (no overtime) and an average score of 13-9. Can somebody help rectify this? I would assume that in a dataset of 10,000 games, that it averages out that each team is about equal to each other in skill. Maybe there is a snowball multiplier where the previous round winner is more likely to win the next round? Would love to hear your thoughts

Edit: For those who care(Not Cned fan) a snowball multiple of 13.55% matches the experimental values of the large data set

My thoughts are cNed clears

You probably count 13-7 as 7-13 thats why you have a mean of 12-12

it's to do with momentum, consistency, team dynamic, smurfing, eco rounds, and some more.