$100B hedge fund

When I was a student in the 1980s, all top science students wanted to break into the engineering or medical schools. Those who chose to major math/physics were laughed by their peers. Now, everyone want to be a quant. Many engineers become idiots!:D
 
Quote from bkk:

In summary, hedge funds do provide a very useful function in society - in fact probably one of the most useful. They can distinguish between bullsh*t companies vs good ones and market bubbles vs reality, while mutual funds, the other major institutional investor, cannot. If the people there could, would they really still be at a mutual fund?

Wow, pardon me, but that is pretty d*mn stupid.

How could you not notice the strong similiarities between the hedge funds produced by the hype of the last few years (easily most hedge funds today) and mutual funds. It's the same scam, leech off investor money with fees no matter whether you win or lose.

Don't confuse Medallion fund with most hedge funds. Just like with quality closed-end mutual funds, regular joe shmoe can't get into any fund that is great. Most mutual & hedge funds will take anyone, however.
 
Quote from Hydroblunt:

Wow, pardon me, but that is pretty d*mn stupid.

How could you not notice the strong similiarities between the hedge funds produced by the hype of the last few years (easily most hedge funds today) and mutual funds. It's the same scam, leech off investor money with fees no matter whether you win or lose.

Don't confuse Medallion fund with most hedge funds. Just like with quality closed-end mutual funds, regular joe shmoe can't get into any fund that is great. Most mutual & hedge funds will take anyone, however.

However, the big differences are:
1) if a hedge fund loses money, they shut down, and
2) since they invest their own personal money, they lose even more.

If a mutual fund loses money, they simply blame the stock market. And try to lobby to get hedge funds regulated.
 
Quote from rufus_4000:



* Order entry, I haven't thought about this through yet, but I have a vague
idea that having a simulated market view can give you a better idea of
how the book is moving, so you can choose your order entry time a bit
better.

There's a growing body of (academic) literature on the modelling of limit order books. One of the interesting and often referenced papers is "Econometric Models of Limit-Order Executions" authored by Andrew Lo of MIT (see paper nr. 31 at http://web.mit.edu/alo/www/articles.html). In this paper Lo et al. model the conditional distribution of limit-order execution times as a function of economic variables such as limit price, order size, and current market conditions, something that can be very useful for (high frequency) intraday trading.
 
Quote from quant:

There's a growing body of (academic) literature on the modelling of limit order books. One of the interesting and often referenced papers is "Econometric Models of Limit-Order Executions" authored by Andrew Lo of MIT (see paper nr. 31 at http://web.mit.edu/alo/www/articles.html). In this paper Lo et al. model the conditional distribution of limit-order execution times as a function of economic variables such as limit price, order size, and current market conditions, something that can be very useful for (high frequency) intraday trading.
I am skeptical about published papers/books. If it really can help you make money, they would not publish it. Jim Simons doesn't reveal anything about his method. There may be a few gems among those papers, but most of them are worthless in the real world.
 
IC.

Thanks for the explanation.

nitro
Quote from rufus_4000:

Definitely measurable, naturally both systems are using the historical sims
for say, correlation calculations. Let's say my scenario DB has 50000 5-10
minute scenarios, and we run through 10% of them for relevance (which is
about norm), the net calculation will be about 500 ms for me, plus I/O time, and
around 30-40 ms for him, plus I/O time.

So conceivably, in a pure stat arb (hypothetical, there isn't that many
pure scenarios these days) situation, his engine would enter his trades at
least .4 sec earlier than me, very significant in terms of catching an
anomaly.

Obviously there are significant hardware differences, I have a small 8
node compute server that cost around 50k, he as a 128 node Egenera that
cost around $1.3M for hardware alone.

This is just a scenario, obviously people have difference models, this is just an
illustration of how model tuning and hardware investment can impact
greatly in a standard type of trade. The game is to use the scenario
analytical power to find different anomalies, and like Jim Simons said,
this is somewhat dependent on luck.
 
Quote from CoolTrader:

I am skeptical about published papers/books. If it really can help you make money, they would not publish it. Jim Simons doesn't reveal anything about his method. There may be a few gems among those papers, but most of them are worthless in the real world.

I generally agree that only very few among the published papers contain descriptions of complete trading systems that are highly profitable. Even among the large crowd of academics, who only strive for (academic) fame and live for no other reason than conducting (groundbreaking) research and publishing their work (personally, I know a lot of these type of people among my friends), you'll be hard pressed to find somebody who is foolish enough to publish such kind of papers.

However, many of the papers or books are devoted to more basic (sub) problems/topics and are part of research fields such as econometrics and statistics providing those among us who are focussed on developing the ultimate killer trading system/algorithm with valuable ideas, methods and directions to achieve our goals, in other words, they may provide us with (small) pieces of the puzzle. This is also my experience from working as a quant in which I must have read hundreds of (mostly academic) papers.
 
Quote from quant:


However, many of the papers or books are devoted to more basic (sub) problems/topics and are part of research fields such as econometrics and statistics providing those among us who are focussed on developing the ultimate killer trading system/algorithm with valuable ideas, methods and directions to achieve our goals, in other words, they may provide us with (small) pieces of the puzzle. This is also my experience from working as a quant in which I must have read hundreds of (mostly academic) papers.

Have you developed any trading system derived from math models introduced in those papers? My strategy is based on traditional TA, which doesn't use high math. Although I'm a math major, I haven't touch math since I left school. It's a intimidating task for me to brush up my old math and dig into those papers.
 
Quote from CoolTrader:

Have you developed any trading system derived from math models introduced in those papers? My strategy is based on traditional TA, which doesn't use high math. Although I'm a math major, I haven't touch math since I left school. It's a intimidating task for me to brush up my old math and dig into those papers.

I have used bits and pieces from a number of, lets say, more mathematically inclined papers, marrying them with elements from conventional technical analysis to develop trading systems. Usually, I stay away from papers that seem too esoteric, i.e. those containing theorems after theorems, including their proofs, and targeting a very limited crowd/cult of specialists.
 
Back
Top