Hacker News new | past | comments | ask | show | jobs | submit login
Why a Hedge Fund Started a Video Game Competition (nautil.us)
127 points by dnetesn on Dec 1, 2017 | hide | past | favorite | 58 comments



Annoyed to see a unit error in the first paragraph.

Bulgaria nominal GDP is $50 billion PER YEAR.

Two Sigma assets under management is $50 billion. (Not per year).

If your car goes 100km/hour, and Shelbyville is 100km away, your car is not the distance of Shelbyville.


Hyperbolic comparisons like these are very off-putting for me too, but I see them everywhere, presumably "for effect". The effect in this instance is that I stopped reading.


I'd saying having the yearly GDP of Bulgaria as AUM is still pretty impressive. And the point is more to contextualize. It's hard when you get past the millions to understand exactly how much we're talking about.


It's a bit odd though because I doubt the majority of readers will have any direct experience with Bulgaria to put it in context. It's like trying to explain X in terms of Y, when Y is equally unfamiliar.


"The yearly GDP of a sovereign nation, albeit a relatively small and poor one" is a lot more meaningful than $50b without context, I'd think.


It is 1/4th of Toyota's or Volkswagen's revenue of one year, so not that much. And last time I checked they did not create a game for recruiting tech talent either.


Oh, only a quarter of the yearly revenue of the world's fifth-largest company by revenue, or the second-largest if we exclude state-run enterprises (c.f. https://en.wikipedia.org/wiki/List_of_largest_companies_by_r...). Alright then.


Still stock vs flow.


So what? I'm not trying to claim they're bigger than Bulgaria.

If somebody compared the distance traveled by some object to the length of a football field, you could similarly object that the football field is stationary, the object isn't as long as that, etc., but none of those things are the point of the comparison, right?



I mean, if your reference point is a state fund run by the government of an OECD country I suppose it's not that big.


That is being overly sensitive.


The comparison is mainly to put their available resources into perspective. Imagine the question: 'We have the entire economic output of Bulgaria in liquid assets at our disposal, what is the optimal portfolio we can construct'? And then re-ask yourself this question every second and just make sure you account for transaction costs as you shift stuff around. Also remember this is pre-levered money so deployed capital is a lot higher.


>We have the entire economic output of Bulgaria in liquid assets at our disposal.

This is still a unit error. They do not have the entire economic output of Bulgaria, they have the yearly economic out of Bulgaria.

The correct comparison would be to say that they have 1 Bulgaria-Year worth of liquid assets.

In this context, where the number is used simply for effect, this is not that big of a deal. However, in conversations where the GDP is what is actually important this unit error causes alot of misunderstanding. For example, what happens when the debt to GDP ratio reaches 1? For a lot of people, this seems like a particuarly important point; but in reality, it is just the debt reaching 365 days. Possibly too high, but nothing particularly special about this threshold.


Hey, but with the right ship you can make the Kessel Run in less than twelve parsecs.


And then make some more movies to invalidate the clever Expanded Universe reinterpretation of that line.


If I have a sense of the distance of Shelbyville, then I'd say juxtaposing these two facts gives me a better sense of how fast your car goes.


That's true, it's partly why units like kilowatt-hours are used. They definitely do lead many people to confusing a unit and it's time derivative though.


How far you mean?


I live in Shelbyville, therefore I am an hour.


I thought it was called Morganville, ya?


GDP is the amount of value created by the economy in a year, and it is measured in USD so there is no unit error.

A better comparison would he the Hour Record in Cycling (how far can you cycle in an hour). It's determined by speed, obviously, but the value is still a distance.


The unit error is in the comparison between USD/year (a rate) and USD (a value at a single point in time)


Do you think the Hour Record is measured in km or km/h?


If you want a fair comparison, do GDP vs value produced by that investment fund in a year.


Alternatively, give AUM/GDP, i.e. the time it would take for the accumulated GDP to equal the AUM.


I'm not suggesting its anything other than a ridiculous and meaningless thing to write, I was just saying that both GDP and AUM are measured in the same units (USD).

Apparently several people decided this fact was necessary to downvote ¯\_(ツ)_/¯


No GDP is in $/time. Gross domestic product (GDP) is the monetary value of all the finished goods and services produced within a country's borders in a specific time period.

Though GDP is usually calculated on an annual basis, it can be calculated on a quarterly basis as well. https://www.investopedia.com/terms/g/gdp.asp


distance/h the per hour is implied when recording distances, however should the race not exactly record 1h the difference is not simply ignored.

Another example, look a records when the physical pools are slightly off of regulation distances for 100m/200m etc. Record keepers in no way ignore those differences.


It's obviously measured in km/h with h=1


I want joules of oil.


And you can. Vegetable oil is around 33 MJ per liter.

https://en.m.wikipedia.org/wiki/Energy_density


So the hedge fund started a video game competition to gain recognition and acquire talent. Well, they are at the top of HN now, so they have achieved their goal.


im actually not sure why this article is doing so well. HN ranking is weird.


I found Alfred’s answers insightful and elegantly stated.


An OK marketing piece for Two Sigma, but I don't see why being a Hedge Fund has anything to do with it. BTW, Robocode has been there for years and there is a huge community behind it.

P.S. I work in finance and sometimes find companies in this industry try way too hard to appeal to the younger generation. Lots of marketing pieces sound like dad jokes.


i am new to robocode but from skimming through it it seems you are just controlling one tank? this two sigma game involves controlling a swarm of bots with colonizing involved so its a different problem. in any case, just 'cause something exists shouldnt block you from trying your own twist on it.


Yeah, I agree that Robocode is very different from Halite.

There are some competitions that are somewhat similar. - Visually, generals.io is pretty similar to Halite-I, but in actuality ends up being a pretty different game. - Halite-I is most similar to the Ants AI Challenge of 2011 (which inspired it). - Halite-II was conceived of as sort of a mix of Halite-I and Planet Wars. - MIT Battlecode is similar in controlling lots of pieces, but the style of game is very different because Halite is all about emergent behaviors in the game whereas traditionally Battlecode hasn't been. They're also very different from the implementation side for the end-user. - Codingame can range from somewhat similar to very, very different.


Probably because people would be surprised at the juxtaposition.


Does anyone have a good feel for what real compensation/career trajectory looks like for a software developer at Two Sigma? I always hear about how great the compensation is in finance, but it doesn't look like that extends to software engineers. Glassdoor[1] says ~130k base/~180k total comp, which is good but not great for a software developer in a major metro area.

[1]: https://www.glassdoor.com/Salary/Two-Sigma-Software-Engineer...


Maybe they're trying to become more of a tech company so they can pay lower wages relative to finance firms. :v


Being a quant generally pays more than being a dev, but even still the Glassdoor numbers are significantly lower than what I’ve heard through the grapevine for many companies in New York. I wonder how recent the Glassdoor data points are and for what level of experience.


If the goal of this was looking smart to attract smart talent, I feel it didn't work particularly well. At least, by reading the article, I didn't get that feeling.

Any game where the last survivor wins are based on the idea to get involved as little as possible early on (unless extremely high rewards are given along the way).


From the 2016 competition, there was at least one Two Sigma hire directly resulting from Halite and several more that were probably resulted indirectly.

Regarding your point about the last survivor winning, that's partially true, but partially not. What you're essentially describing is a proto-version of the non-aggression pact we saw develop in the last weeks of Halite 2016 (and which was indeed largely successful). However, even there, it wasn't quite so simple; competition winner mzotkiew had an excellent write-up of his strategy at https://github.com/mzotkiew/HaliteBot/blob/master/writeup.pd.... Halite-II makes non-aggression even more difficult due to the lack of discretization found in Halite I.

source: conceived of and co-developed Halite.


I recently discovered halite too and I think the origin story is pretty interesting.

from https://halite.io/about/

> Halite I was conceived of and developed by Benjamin Spector and Michael Truell in 2016. Two Sigma, having had a history of playful programming challenges for its mathematical and software-oriented teams (e.g., see the Robotic Air Hockey Competition) retained Ben and Michael as 2016 summer interns to develop Halite, run an internal Halite Challenge, and ultimately open Halite up to human and non-human coding enthusiasts worldwide. Halite I was a great success, developing a flourishing community of bot builders from around the globe, representing 35+ universities and 20+ organizations.

> As a result of the community’s enthusiasm, the Two Sigma team decided to create Halite II, an iteration of the original game with new rules but a similar structure and philosophy. With Ben and Michael as creative advisors, Halite II was developed by David Li, Jaques Clapauch, Harikrishna Menon, Julia Kastner as an evolution of Halite I. The team considered simply reviving Halite I, but given the progress the community made and the number of open source bots that had been published, the team decided to create Halite II with new game mechanics and a fun background story that fleshes out the Halite universe. Halite involved moving pieces around a board with only up-down-left-right options. In 2017’s Halite II, bots battle for control of a virtual continuous universe, where ships mine planets to grow larger fleets and defeat their opponents.

not bad for just a second iteration. cynical people can view it as just a marketing thing, but you can also think of it as a very successful intern project that I personally would find hard to pull off. in particular i am surprised that there are more "professional" players than university/high school players. (https://halite.io/programming-competition-leaderboard)

in the forums there are high schoolers doing well in this game and the fund is choosing not to work with them. i think this is shortsighted. (https://forums.halite.io/t/two-sigma-interview/312)

disclosure - am interviewing with 2sigma.


That's interesting because both of the developers attend Horace Mann (still)


odds on them having a job when they graduate? :)


Ben Spector here; both Michael and I are planning to attend college, but we'll see in ~five years! :)


Hey Ben! i'm going to start at 2Sigma soon. would love to meet up with you if you are ever in new york. swyx@swyx.io


What I thought this was going to say was that they had built a system that translated market actions into a video game context and conversely in real time, such that someone playing the game would actually be trading, without knowing it. Then they could pay people to play the game with some small fraction of the profits they were making.

How's that for an evil scheme? BWAHAHAHAHA


Then someone realizes what you're doing and "plays" your game to play you. If you outsource your brain, don't expect your bodies sanctity to stay intact. How's that for an evil scheme.


It's a marketing gimmick. Along with a ridiculous algorithms heavy whiteboard interview process (think Manacher's, graph algos)... all for average compensation and you'll probably end up babysitting a database at most. Try talking to someone who actually works there before believing these fluff pieces.


Those first two paragraphs make me feel like there's a good TV show about this industry that hasn't been written yet.


Billions on Showtime


Good the first season, with the tension building up to a peak in the season finale, with the two main characters revving their engines to clash in battle. Perfect for a second season to erupt with unrestrained fighting, and eventually resolve and end the series. Instead, the makers began the "let's draw this out" procedure. :-|


I agree. They even included a charity poker tournament.


Thanks, I'll check it out


Amazing show.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: