It was in the past that we do use past record to predict the future, but this season proved everyone wrong that past history doesn't guarantee future results. I see Chelsea as the favorite in this match, because City have lost form and are unpredictable when playing with big clubs. Imagine how they lost their only chance of qualifying to the round of 16 to PSG, despite on two goals lead.
Literally it's not that we completely depend on the past to make out predictions, basically it's supposed to be that the past was to give us a insight of a possible outcome when certain teams meet and not an absolute to an expectation from both teams, but some persons makes it the absolute and. That could be reason for some of their losses this season. Chelsea has got a very fierce record with Manchester city of which entails mostly an epic game between the two sides , so for fixtures like this expectations usually are kept at its minimal.
Past matches or like H2H are just one of the events that can be used to strengthen predictions. There is nothing wrong with anyone predicting the results of a match using past meetings. We also know that the results are not always correct, but we as bettors must do a lot of analysis. So even though H2H does not provide any guarantee, at least we have other analysis to strengthen the results of the match.
And regarding the Man City match that will face Chelsea, I am more realistic that Man City's performance is in bad condition. While Chelsea is in an inconsistent condition. However, Chelsea has a slight advantage in defense that can be relied on. My prediction is that the match will most likely end in a draw.