3 years ago a user aptly named "hft" discussed some ins and outs of high frequency trading. At that time, there appeared to be a handful of HFTers on board here. Now it's crickets. I wonder if even he could have foreseen such a dramatic shift in the landscape. Margins have continued to compress and data, infra, and regulatory costs have continued to rise. Low hanging fruit that dumb algos once fed off is drying up. Equity volumes continue to decline. Everyone has FPGA devs and microwave links. Meanwhile, some HFT have exited various asset classes and/or regions, or gone belly up altogether. CME rolled out its upgraded gateway providing "fairer" time priority. Presumably making some trades much more winner-take-all. Also not long after that post, CME stopped providing participants with their fills before they hit the public feed. Regulators have more actively pursued cases against spoofers. Manual traders really, but HFTs continue to receive scrutiny and be questioned over spoofing for legitimate order cancellations. HFTs also continue to fight off the notion that they are parasites feeding off "latency arbitrage." IEX has been approved as a U.S. equity exchange and will have a protected quote starting next month. IEX is supposed to save us from HFT yet they themselves promote and offer complex order types. Other U.S. exchanges are looking into introducing their own speed bumps. The SEC has established a Market Structure Committee to review and figure out ways to improve U.S. equity market structure. Tick pilots, access fee pilots, and more are on the way. Open issues still include the validity of Reg NMS Rule 610 (primarily as it relates to order protection), limit up limit down refinements and intermarket harmonization, and payment for order flow. Democrats have promoted the idea of a FTT which would destroy HFT, trading, and liquidity as we know it. The HFT space has matured quite a bit and it isn't a rosy picture. Yes, the strongest will survive and continue to have no or close to no losing days. Speed will continue to be important, but less so as a pure edge itself and more as a requirement for all trade execution. This isn't just applicable to HFTs, but also to institutions who are now partnering with HFTs in hopes of reducing their slippage. So ETers, how do you view HFT today and where do you see it headed? And for those of you who work for an HFT, what changes have you seen in the last few years and how are your firms adapting?
All these changes have been slowly rolled out and pretty much everyone has seen them coming. The super easy no thinking money died out but there still seems to be plenty out there for firms with solid people and ideas. The democratization of the CME has largely been a good thing and was pushed for by HFT firms. Nobody likes spending time, effort and money optimizing ilink connections. Now, also, it's easier for small teams starting from scratch to get a fair shot speed wise. Coupled with the new corporate membership rental option, I've seen several small groups get up and running profitably recently. Larger (head count) HFT firms still find pretty lucrative niches outside the US. Biggest issues facing trading firms are: Not watching costs carefully, especially tech costs Having unnecessarily unpleasant work environments Not valuing their developers FTT effectively won't happen. If there aren't enough loopholes or if it's greater than about 3c per futures contract there would quickly be no US based exchanges.
@garachen: it is very interesting what you said about "corporate membership rental". So perhaps CME is going to finally get itself together and adopt ICE-like cost structure. But I hope they keep technical documentation accessible.
Ironically, I see pendulum swinging towards riskier trades with longer holding periods and less certain edge. With all the regulatory zeal focused on "spoofing" plus over-supplied market-making, I see people going taker-only.
There is a lot of taker only. And I think you are right in that most of the new places seem to also only take liquidity. I haven't seen much with longer holding periods. Personally, I really like the several minutes timeframe because it's extremely hard to model so there seems to be less competition. But I haven't heard of it becoming popular.
I would say "I wish" but, since I have it all already I don't. Yeah. A week or so ago they publicly announced that you can pay a monthly fee instead of buying/renting the CME shares. You still have to buy/rent the seats. So you still have to outlay a bit over $1m to buy the seats that you have to and then maybe 30k/ month for about $5m of rented seats/fee. Maybe a bit less.
@temnik I suspect a similar migration. Maybe riskier, maybe not, but does appear that holding times are being pushed out and more aggressive trading is being done by HFTs. @garachen Always appreciate your insights. I am surprised smaller teams have a fairer shot speed wise. I thought port jitter helped those with worse/cheaper tech still get a piece of the action. I suppose the counterargument to that would be the big shops could just stockpile ports until they hit the jackpot.
Is that all - monthly rental of CME shares? Like a perpetual put option? I need to catch up on cmegroup.com advisories. In any case, in the past people could hedge CME shares requirement by being short in a different account (there's some accounting mechanics there) - but it'd only really work for broker-dealers, I believe.
@algonoise: HFT's can no longer just "quote for a tick" - so they HAVE to explore less statistically certain relationships. Even more traditional market-making strategies that involve carrying significant inventory have become riskier. Especially with MDP 3.0 and no longer getting own fills first.
I guess MDP 3.0 would mess with virtu style market making. So maybe slightly longer holding period and more risk. Probably the same profitability. I haven't noticed much of a difference.