Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I've seen several myths (mostly on another forum, thankfully) about quantitative/"high-frequency" trading lately and decided to start this thread for the younger guys on the forum who are considering to pursue a career in buy-side finance.
I'll reply whatever's within my means: I did purely quantitative trading at two large hedge funds and an investment bank before starting my own firm. I have a BS in physics and math and left academia right before I finished my PhD. My firm has more than twice the number of "HYPSM" degrees as it has employees. I'll prefer not to discuss low level details about strategy implementation or what a particular firm does. We're also not looking for funds or employees, so I can't take any inquiries about those. My goal is more to dispel unfair accusations about Wall Street/HFT in popular media, share what's meaningful about a career in finance, and encourage the younger guys to enter this field for more purposeful motivations besides money.
I was involved in quantitative research but on higher time frames. Some of my friends back in Israel are experts in AI etc. On of them was actually doing a funded research in looking at historical data of poker games and trying to identify "inefficiencies "
Are HFT companies still willing to pay a few millions to be 1 foot closer to the exchange ?
I see that you are in Cambridge, MA - I'm in BackBay Boston - almost neighbors
What kind of Sharpe ratios are HFT funds achieving now? Isn't the HFT space getting more competitive now? I know a couple of funds that were minting money back in the day ( 2007 ) are now closed.
Great to know others in the area. I hope you were safe during the ordeal.
1 foot of propagation latency is not enough to justify a few million, but you're in the right ballpark for infrastructural costs. It will also depend on the strategies deployed, but most of this is paid for the low-latency interconnect between venues.
I wouldn't know the average, but the maximum I've seen is low hundred, delivered over a year.
1. Every field and industry gets more competitive with time and undergoes consolidation, so it is not unexpected.
2. The firms that had always dominated the field are well-and-alive.
3. 2012-2013 has seen a growth in hiring (although this has been associated with a decay in sell-side employment).
4. In a broad sense, strategies in this space are buy-high-sell-low (no typo), which means there are simply less opportunities in the traditional space of listed and displayed markets in today's environment.
5. Of course, many firms have both low-latency and long duration strategies. I'm just scrapping the surface but in '08 we had the stock market and interest rates move down together and that was enough to toast many of these funds. (Their strategy re-allocations effectively resembled total return swaps which had 0.2 sigma.)
6. There are plenty of opportunities in the low-latency space that are yet unexplored. Some of these require regulatory clearance and joint effort (e.g. messaging in non-deliverable forwards).
7. The attrition rate in any kind of fund, not just those dealing with low-latency strategies, is very high. Obviously if the media is doing a good job of their statistics, they should detrend this to remove extraneous variables.
(: I've been to a few conferences in the past 2 years and from my angle, I wouldn't put Stanford on the list.
I prefer the academic view: The first order relationship would be that information drives risk-neutral pricing from market makers (whether designated or functional) who have the choice to widen their spreads, which then drives price discovery. It makes sense from an agent-based or behavioral view of the market that there are higher-order relationships which drive price discovery, but these would be indistinguishable from a stochastic process.
The practitioner's view is easier to grasp: The execution algorithms that the majority of the institutional traders use, by design, do not attempt to capture opportunity costs of informed trading. It's considerably easier to solve this program. Also, if you're designing an execution algorithm, chances are that you were educated in a way that ingrained EMH in you. So you'd probably think it makes little sense to have a forecast on the price in your execution algorithm anyway.
Another good question, and this actually requires some detail because I realize people don't think a lot about it: what exactly is a "hold time"?
Firstly, an educated answer is that the $80 trillion or so in pension funds, mutual funds and insurance greatly dwarves the $2.3 trillion hedge fund industry. If you are investing an insurance company's portfolio, the major risk factor that drives your portfolio is the duration in long-tail products such as variable annuity, so the view is really in the precision of days and above. If you are managing a mutual fund, the goal is all about minimizing tracking error against a particular index, and effects such as dilution, delisting, M&A do not take place intraday, so the trading horizons are really in the precision of days and you leave it all to an execution desk or algo to deal with the nitty gritty of large-order transaction costs.
I'm guessing your question is more geared towards hedge funds and their consensus is on an "optimal trading horizon". Large hedge funds trade long horizons again in the order of days and above - it's not that they prefer to, but rather they have to because of the size and the small number of distinct strategies you can run before you eventually run into institutional frictions. (Most I've seen is like 20.)
Another issue is that average hold time is a very uninteresting thing to measure when you have a large portfolio. An institutional trading platform like Tethys aggregates firmwide trades so if manager A shorts Q of S and manager B goes long Q of S, the net position is zero. Then a while later manager B offloads S so now the engine has to short Q - I don't really know how you'd put a finger on "holding period" because I can see many different ways to measure it for different parties. Or if your position size changes with a sequence like: +3, +7, -2, -2, +4. You can see how quickly this gets out of hand. In practice, Manager A, B; risk department and management should have different measures because they have different interests.
While trade holding time is an ambiguous metric, order duration is not (the full equities feed lets you track each order by a unique ID). And of course, about 30% of orders are cancelled in under a second.
Some companies especially in Chicago came up on market making, that is providing liquidity. Recently a CFTC ( or SEC , I cannot rmemeber ) report said that HFT firms take liquidity 57% of the time. Do you see more liquidity takers or liquidity makers nowadays ?