Increasing probability based on passing through first two committees ( Ways and Means, Energy ):
https://www.washingtonpost.com/powerpost/obamacare-revision-clears-first-hurdle-in-house-committee-early-thursday/2017/03/09/579586b4-04c2-11e7-b9fa-ed727b644a0b_story.html?utm_term=.9b013b62c972

The plan is clearly laid out, the president is on board and it sounds like they have a strategy to pressure republicans who rebel.

Lastly, the stated goal is to beat the April recess date. Even if they miss it they still have one week at end of April to make it happen.
http://www.nwitimes.com/business/ap-explains-how-to-transform-gop-health-care-plan-into/article_4a2404ca-ff5d-5e00-bfa2-7424eb2d5ce9.html
"Senate Majority Leader Mitch McConnell, R-Ky., says he wants to pass the bill before the April recess, sending it to President Donald Trump for his signature."

Leaving 75% probability of it not happening because the timeline seems artificial. Artificial deadlines on complicated issues are useful for pushing things along but difficult to achieve in practice.

If the prediction had no end date I would set at 99%. It's really a much less interesting forecast on exact date at this point.