I haven't look too closely at this space, but someone here at HN made the claim that bank APIs are sometimes inaccurate, which is why Yodlee scrapes in the first place.
I've never had a problem with them - granted I only tested with accounts I had access to through family/friends. That said, it is up to the bank to maintain their end of the API and make sure that their server has the uptime and correct code to parse the request and return the response, so when that doesn't happen, the only resort is to scrape. Do you have a link to the thread where the claim occurred?
I am not disagreeing with your claim and am using OFX, for now.
My initial goal was to provide a simple and cheap API that one can get started with in minutes vs. having to gather a server list, maintain it, read through the OFX docs, etc. In my testing, I have never had a problem with stale data, but that doesn't mean that it doesn't exist. Going forward, I am not opposed to scraping if there's a demand for it.
Can you confirm/rebut?