Q&A: I worked as a CTO at prop-shop doing high frequency trading (HFT)
Hello everybody! I've been very lucky to end up doing what I'm doing My expertise is building low latency trading systems for HFT strategies. I led the software development to do high-frequency trading, executing several different strategies. Nowadays, I started my own consulting firm and working on opening my own quant shop. Ask me anything!
How can one best try and design their own HFT algorithm? I know with mid to low frequency strategies there are ample backtesting platforms, but for HTF it seems to rely quite a bit on order book mechanics which their isn't a ton of readily accessible data for.
Also for a math heavy undergrad looking to get into the field, what is the best route to entry?
True. There is no much public data available for this unless you are already working in an institution. However, you can download pure tick data from NASDAQ, and play around. The strategies we applied are related to just market microstructure and order flow analysis.
Forget about backtesting, this won't make any sense. Once you get an initial idea, you test it live. Having this kind of trading frequency you can collect enough data within minutes.
Thanks for doing this. Which firms are leading in the HFT space? Which firm has the best low latency network?
Leading hft are the usual suspects: citadel, jane street, etc However, I've seen tons of startups trying to successfully do the same. Mostly on the institutional forex side. Entry cost is getting cheaper and cheaper every day.
How do you see the industry changing in the future? Do you think it'll continue to become more and more of a latency game, and eventually the top handful of firms with the fastest tech will dominate and push the others out, or do you see it going in a different direction?
The industry has certainly changed already. Lately, the big boys were leading the space because of their big pockets, but that's also changing.
I think that latency race is over. Now they are adding more flavors to their strategies: AI, ML, etc
But, latency always is a key component. Every algo depends on it
Where exactly does AI/ML fit into HFT?
There aren't that many features that are coming in at a high frequency relating to a particular security (equity fundamentals change once per quarter). I'm very curious what exactly HFT guys would be using machine learning for given the seeming dearth of features/factors that are changing intraday.
What kind of strategies are you leaning towards upon setting up shop? I know your expertise is in latency arb, but are you considering other strategies (alternative data, ai etc) or longer term trading ?
Anything long term (higher than an hour) is too risky for my taste. The shorter the time frames, your exposure to market fluctuations is less In terms of strategies, besides the well kknown latency arb, you have market making (one of my favorites), mean revertion and also momentum. Of course these are strategies that try to catch a VERY small profit and get out.
To give you an idea, we trade blocks of $1M (as minimum), and we aim to take $15 Do your math!! Now, multiply that same thing 10,000 times per day https://media1.giphy.com/media/AVBo5eqFXd3SU/giphy-downsized.gif" alt="amazing" />
What other roles are available in a HFT/prop shop that are non quant and non software related?
You will find technical people, mostly I would say 60% Software Engineer (low level), 20% researchers and 10% networking and infrastructure
I've seen Accounting people from Big 4 at prop shops, is that the least quant intensive role at a prop shop?
Thanks for the AMA hft1111 - several questions.
Any thoughts on DRW as a firm?
Nothing to comment on this firm. :)
Are you looking into opening a HFT shop? Where do you think new entrants can compete, given they don't have any infrastructure / "petabytes" of data? Cryptos maybe?
Yes, I am. That's going to happen in late 2019 For now, I'm just doing consultancy.
Nowadays, everything is easier than ever for new entrants. Infrastructure is getting cheaper and cheaper, and the petabytes of data are extremely accessible. Maybe not for an individual, but for a fund it is.
I do not think cryptos is an option yet in this space (HFT). Fees are extremely high and exchanges' technology too immature. Not to mention the lack of liquidity. I think that needs more time until the HFT guys enter into this space
Thanks a lot. What asset classes would you consider? Do you think there are still opportunities in US cash equity in public exchanges for new guys (given how crowded it is), or are you looking into dark pool or less liquid assets?
Wow, I had to read your post twice to make sure I am not dreaming ;) I am running a pilot project with my algorithms by trading Dow e-minis. Currently, because it is still a pilot, my time-in-position ranges from several seconds to 1+ min. As I have been considering switching to S&P e-minis, can you please comment on how much more S&P e-minis market is "infested" with HFT's compared to Dow e-mini market? Thank you. Cheers!
HFT is everywhere nowadays. On futures, they were for many years. I've found these numbers on internet, which will give you a better idea.
Quia corporis quia itaque magni. Blanditiis repudiandae inventore vel est accusantium illum neque. Ab qui unde quo. Commodi enim dolorum impedit rem id.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...
Sint accusantium omnis est aut. Iste repellendus repellat et est asperiores labore. Iste rerum et commodi nihil libero autem. Tempore maxime nostrum autem.
Et voluptatem qui labore possimus et est beatae et. Dolorum recusandae qui magnam optio velit est debitis. Officiis neque dolor similique quasi. Voluptas labore quae ex consequuntur quia autem.
Dicta sed ut id ut. Consequatur et laboriosam laboriosam dicta officiis necessitatibus sed. Fuga hic veritatis blanditiis saepe quasi est vel nobis.
In eum aperiam nihil voluptas voluptatem corporis ipsum. Qui est officiis eum nisi. Sequi itaque beatae aut ipsum.