This. This is one of the reasons I left my last PM job. I was being told to spend less time with the team and more time on the road with clients. I don't know what the right ratio is, but I do know that my rapport with the team and deep familiarity with the stories/tasks led to pretty solid estimates. Also knowing the patterns of the dev lead helped a lot, he was a chronic underestimater so I'd do my best to push him to consider things like coordination, legacy, etc.
Personality of the estimator plays a huge part and the voltile/stable essay was a great shortcut for understanding this. A voltile will hit the estimate or beat it. It'll be cowboy all day long and god help the company if it needs to be maintained. I've had a few dev leads like this. They'll take the tricky bits and do it alone. At night. On the weekend. Half asleep. Then, there is the stable. They'll never hit their original estimate. The most egregious won't get beyond the proof of concept without some serious prodding. When the project is done, it's meticulous, but we rarely get to that point because of the daily 6 hour long discussions about naming...