You have 100 requests per hour, so the question is how do you balance it between the different types of requests. I think the best option is the TweetDeck method, which allows you to set the percentage and save the rest of% for publication (because this is important too): 
Around caching, the database would be nice, and I would ignore the deleted ones - after you downloaded the tweet, it doesnβt matter if it was deleted. If you wanted, you could theoretically just try to open the page using a tweet, and if you get 404, it will be deleted. This means no API costs.
Robert MacLean
source share