I agree it's disgusting to think of people working for a company like Goldman, but not sure I agree with the rest. Also some people are just trying to put food on the table, and they could compromise, but they don't and it's not proper for me to judge them even though I am.
Firstly, Goldman can't correctly value a company - no one so far can. The best Goldman can do is use their position of power to effectively tell the market what the value is, but they can do that practically by creating magic numbers from the sky. Indeed, go and do an old school valuation on many companies and you'll get vastly different results. I have a degree in Finance along with various computer and math degrees. One of the big takeaways is that financial valuations are entirely subjective just like stats studies are, so it is the justification, evidence, techniques, and so on that matter. Your job is to convince someone, and Goldman exploits its position as a large player by doing just that.
If Goldman could actually set prices or guess them exactly to the millisecond, they would never have any losing bets and they could manipulate the market 100% to their will. One could say that in this sense, you are right as they do indeed manipulate things to some degree. They are so large that their actions influence the rest of the market and their failures and successes change the market, ironically screwing themselves up quite often.
Many principles though say that valuing a company is hard to impossible (just ask Black and Scholes). Random Walk and some of the associated studies like throwing darts at the financial pages (darts beat top Wall Street investors more than 50% of the time usually) demonstrates that there are elements we do not understand in terms of valuations and predictions, and that moreover people don't act rationally depending on your point of view. Given this reality, trying to build better models is a very interesting and wide open computing topic.
Another important factor is that valuations and other financial market indicators change over time. That means your models both need to account for time and be extremely fast at reacting to changes. In computing, this touches a lot of interesting topics from performance to network architecture to low-level protocols to CPU caching to distributed computing to machine learning and on and on. Outside academia, there are few companies that are focused and yet touch so many different computing topics and actually apply them to their business in production.
Personally, I would never work for a company like Goldman. My moralistic view point has cost me at times financially but I am OK with that given my internal priorities. For some people, it's all about money. Goldman might be evil to many like me, but many popular tech companies aren't much better if you take off the geek glasses.
It is not a popular sentiment around here, but I wouldn't necessarily put people at a lot of popular SV companies like Facebook, LinkedIn, Google, Uber, and so on above the Goldman people or on some higher pedestal/plane of existence. Surely some are doing interesting or altruistic things even, but most are just grinding out the day trying to keep the business afloat or make it rich.
Many of these companies on some level are still tightly tied to people like Goldman, have similar motivations, and arguably a lot of the work they are doing is intellectually less stimulating, valuable, academic, and challenging. Moreover, a lot of the "work" and "tech" that these SV companies think they are inventing has already been around the financial sector (or even tech sector itself, not to mention academia) a long time and exists in better, faster, more powerful forms with the caveats usually of being expensive or exclusive to a single company. Not many people realize what tech people at a company like Goldman are doing, and the people that work there often are bound by legal agreements not to share. That makes companies like Goldman very adversarial, anti-open source, and anti-employee. To that end, I think they are pretty disgusting, but I've experienced nearly as bad working at game companies and tech giants.
Very few companies are doing anything worthwhile with brainpower. Even academia where brainpower should be tantamount has often become more about stats like amount of papers published, grant money, or other financial results rather the brainpower. As a whole, computer science is a field that constantly reinvents wheels poorly, wastes tremendous amounts of brainpower, and is far from altruistic or progressive at times. We could say that about almost any field, so take it for what it is and remember that most of us (a select few make a smaller scale difference) aren't saving the world, even when that literally might be the company "mission." Most of us are wasting our brainpower most of the time and it's up to us to do better; working for a company is almost never the answer for maximizing your brain.