So I've been making and posting these win probability distribution graphs over at SBNation accompanying Bill C's previews. They take the projected outcomes per S&P+ and perform a random walk on the season, calculating the likelihood of each total number of wins (again, per S&P+ projections). I'm written a script that turns these out pretty easily (just have to enter the projection data, which hopefully I can convince Bill to give me in an easily serializable format).

Anyway, I've hit a little bit of a stumbling block and need some feedback. I'm going to post two graphs below and you'll see what I mean:

So you'll notice these graphs contain the same information, the only difference is the color coding. They both use a red/green gradient with green as more likely than red, but they set the scale differently. The top one uses an absolute scale in which red is 0 (impossible) and green is 1 (certain to happen). This has the advantage of letting you compare rows easily, but I think it makes comparisons within the rows more difficult.

The bottom graph set red to the least likely outcome (minimum of the row) and green to the most likely outcome (maximum of the row). This makes it really easy to see which outcome in the row is most likely, but it means that green and red mean different things in each row (ergo, you can't compare between them). I personally think it makes events seem much more likely than they are (for example, 33.7% in the last row is bright green even though it's only about a 1/3rd chance of happening).

Anyway, which of these do you prefer? Is one better or worse than the other? Is there something I haven't thought of (I know I need to add the home/away column, sue me)? The charts aren't HARD to make, but since I'm having to do the entire thing by hand in Python it takes a little more work.

Any other feedback would be great.