I always have a hard time finding VMware benchmarks. The most frequent reason I hear is that VMware doesn't allow publicly posting them (maybe it's in the TOS too but I always just gloss over it). They seem to also go out of their way to contact blogs and news sites that they're not allowed to post benchmarks of their products. At least that's what the editors says. I don't have my own tech blog.
As for actual numbers on the impact of memory speed (latency should also be considered too) for applications in general, you could look for benchmarks done by AnandTech and Phoronix. In general, GPU intensive tasks on systems with GPUs that use the main system memory, like Intel GPUs and AMD APUs (the integrated GPU in AMD APUs are more memory speed limited than Intel ones), are the ones that get affected the most. CPU bound tasks don't get affected much or at all. Any task that aren't CPU nor GPU intensive don't noticeably perform better with faster memory since their active times are so short that speed ups aren't that big in comparison to the total active time (ie. a 2x speed up on an in-frequent 1 second task isn't really noticeably by most people).
It doesn't seem like you're going to be playing games and the virtual GPU for VMs might be the more limiting factor than memory speed. I don't know what kind of day trading you do but speed and latency would likely be more important for those doing high frequency trading where latencies between trades are calculated in milliseconds.
Here are some recent memory scaling articles:
http://www.anandtech.com/show/7364/memory-scaling-on-haswell
[Phoronix] DDR3 Memory Scaling Performance With AMD's Athlon 5350
The AnandTech article talks about avoiding DDR3-1600 memory for enthusiasts and competitive overlockers but the actual numbers reveal near 0 improvement on CPU tasks and up to 5% improvement on GPU tasks running on Intel GPUs. Not really noticeable.
The Phoronix article also shows negligible performance differences with CPU tasks. For the AMD APU, GPU intensive tasks do noticeably benefit from faster memory. These GPU numbers may or may not translate to VMs.
