The headline "lowest rates in a week" may not be too sensational or even that informative at first glance, but in today's case, it's big news.  You'd have to go back to last Thursday to see the average lender quoting anything better, and that's the first time we've been able to say that since the most recent rate spike began on March 7th.

March 7th might not seem like that long ago, but it's highly unusual for rates to move higher so consistently.  There are plenty of examples of rates going a week or two without hitting a 5-day low, but there are no recent examples of anything longer.  In fact, you'd have to go back to March 2021 to see a similar scenario.  

The average lender is quoting conventional 30yr fixed rates in the upper 4% range, but the exact number can vary quite a bit depending on the lender, borrower-related variables, and the options chosen by the borrower.  For example, paying a bit more upfront can make more sense than normal right now due to the bond market that underlies mortgage rates.  Those bonds are offered in half point increments (3.0, 3.5, 4.0, etc) and each bond is like a bucket that is only allowed to carry certain interest rates. 

One of the most active recent buckets contains rates of 3.75 to 4.625.  The next bucket (the one that holds rates of 4.25-5.125) hasn't seen nearly as much action over the past 3 years.  This means investors command a bit of premium before making loans that can only fit in that bucket.  As such, 4.625% (being the top of the easier-to-use bucket) is much closer to earning lenders the amount of income as 4.75%.

Don't worry if this is confusing. The bottom line is that it doesn't cost a borrower as much as it normally would to drop their rate by 0.125% assuming we're talking about the 0.125% from 4.75 to 4.625. 

In other words, this means that 4.625% with some upfront cost may be a better option than 4.875% with no cost.  This isn't true for every lender, but it's one example of why mortgage rates don't always move in lock step with the broader bond market.