PyTorch wraps THNN, not Torch. Moreover, even if this was true, it wouldn't matter at all. Practically 0% of the overhead is related to Python in the first place, all of the time is dominated by the underlying C implementation.
1,354 karma · joined August 18, 2018
Always interested in opportunities in New York, Chicago, Tanzania, South Africa, or Hong Kong.
There’s a couple of things wrong with this argument so let’s go over it:
1) Low or uncorrelated return streams are very valuable due to the central limit theorem. As you add these return streams together, they form an increasingly tighter normal distribution. This gives us two advantages: we can convert non-normal distributions into normal ones, and we reduce our standard deviation with every stream we add as long as the correlation is less than one.
2) Returns aren’t the defining factor of whether a strategy is worth it. Instead institutional investors measure investments by their Sharpe ratio: the slope of the line between return and risk (the axis is the risk-free rate). The reason for this is that since these kinds of investors can not just lend, but borrow at the risk-free rate, they can lever up their returns to anything they want. This is called risk-adjusted returns.
4) The biggest risk in investing is exposure to Beta, “the market,” which in the US is usually measured by the S&P 500. Reducing exposure to Beta is often goal #1 of institutional investors. Maybe not a 100%, you should definately not be exposed to a 1.0 Beta coefficient. This is especially important for endowments, pensions, and other funds that have monthly or yearly pay-outs. The auto-workers would be super pissed if the economy tanked and they no longer got their retirement pensions. After all, that’s one of the things that pensions are supposed to hedge against.
Before I started working in finance, I definately had the same questions and theories as you. After all, the field is very baroque and complicated and the explanation of the industry being stupid is very simple and straightforward. Chesterson’s Fence is a good way to view a lot of things. Understanding is always a prerquisite for change.
Moreover, why not just make the workers salaried so they can’t complain?
Also we could also talk about: dry clean suits, taking showers, cutting hair, shaving, putting on make-up, etc.
Oh man. What is this person talking about?
Though who am I kidding, no one probably evaluates production like that. Though I imagine chip makers are pretty serious about that kind of stuff.
2) Even today, bonds are traded OTC and mostly by people. In the 1990's, 100% of bonds were traded by people. Bond market cap dwarfs equity market cap.
3) Exotic OTC options were a big thing in the 80s and 90s. While that's not true today, when they existed, all of those were traded by people as well.
4) Most hedge funds today manage their positions with excel and trade based on analyst recommendations. They might have some fancy factor ops engine or something, but it's basically an after thought.
5) Around the dotcom boom, Goldman's equity desk had over 500 traders.
Fundamentally we are talking about a lot of different things. Different asset classes, buy-side vs sell-side, NYSE systems vs pit traders, etc etc. We can slice and dice these numbers in many different ways: total number of discretionary firms vs quant, trade volume vs assets, buy-side vs sell-side, etc etc. There's many ways to look at it that result in different conclusions. All I'm saying is that computerized trading is probably not yet the dominant paradigm and it certainly wasn't the dominant paradigm in 2007, much less 1990.
While no concrete evidence has been provided, there is a lot of statistical evidence (he’s on a practically impossible part of the bell curve), Occam’s Razor implies that he’s cheating. Moreover he only plays when twitch is rolling. The most likely explanation is that he’s giving a percentage of his winnings to some tech guy processing the feed, probably more than 50%.
Of course this only US and only equities, but it's still well above 1%.