Velvia is a brand name that Fujifilm employs for three different products. I suppose you mean Velvia 50 or RVP50.
From what I understand (and what I understand is not Gospel at all) look the graph "characteristic curves", page 8.
The more-or-less straight portion of the curves goes from around 0.3 (right side, highlights, "foot") to around 1.7 (left side, shadows, "shoulder").
That means 2.0 as measured on a log base 10 exposure scale. That means 10^2.0 = 100.
In stop terms, that is between 6 and 7 stops. 6 stops would be 64, 7 stops would be 128.
Outside of the linear portion, you still have texture, but the response of the film is not linear (which helps you in not burning the highlights) and is not chromatically correct (see in the shadows how the red layer clips much before the other two layers).
The answer to your question would be that Velvia 50 (RVP50) has the correct and nice response that you expect some 3 and something stops on both sides of middle gray, after which the non-linear portion begins. Considering that highlights burn faster and with greater overall damage to your image, you can - as most do as I understand - consider the dynamic range as being 2.6 EV above middle grey, and 3.3 below middle grey, and still be using 6 stops of dynamic range (your film has something more in the quasi-linear portion, but not much, the rest is foot and shoulder).
In practical life, in a contrasted situation, with this film you meter with a spot light reflected meter the brightest part of the scene which you would consider a failure if it came burned out (lacking texture, looking "washed") and open 2.6 stops more than what the meter measures. That places that brightest part of the image in the part of the characteristic curve where the film answer is still polite
The shadows will fall as they will fall and if you are outside and have no control over light that's just a thing you stoically accept.
In practical terms, you would "scan" with your spot light meter the shadows and see where they fall, and how large it is their extent. You pay particular attention to any spot which is darker than 3.3 stops than your previously calculated exposure.
If the extent of those zones is too large and the final slide is going to come out a bit too sepulchral, you have three choices:
- bracket a bit toward opening more, accepting some highlight burning and hoping the shoulder will save your picture;
- wait for some clouds;
- save film.
PS People don't use slides because they guarantee better chances of success. They don't. Slides can be problematic and sometimes are a pain where it aches. People use slides because when they work, they work a lot. It's not important how bad are images that you don't show, because you don't show them. It's important how good are images that you show, because you show them. Slides are very good at producing very bad and very good images. Just toss the bad ones (or save film, or use negative).
PPS If you consider your highlights limit to be 2.7 and your shadows limit to be 3.3 you are already "skewing" your measure toward a slight underexposure. I would meter film at its nominal ISO value. I never "rate" slide film any different than ISO value, but I do consider the dynamic range "asymmetrical".
PPPS If you use an incident light meter, in low-contrast situation you just expose according to your light meter indication (at nominal ISO). If you have one of those situations with a lot of contrast and the brightest marble receiving the most abundant light, you might close some 1/3 or 1/2 stop to prevent highlight burning.
(it goes without saying that if the frame matters, and if the situation is tricky, you should do some bracketing).