Ask an algo trader anything
Currently working at a prop firm, joined recently, and through a bit of luck and personnel fun I've managed to run the european time zone operations for my company. I'll answer anything non technical (what do your sheets do hurhurhur), that will give people a better understanding of what we do.
What's your background?
What's the balance between computer automation and human input? What's your quant/computer background? What securities are you trading (fx, options)? Any particular strategy details (technicals, market making, stat arb, HFT)? Always been a bit confused by the term, since "algo trading" doesn't really stand for any specific strategy on its own.
All sheets are fully automated but always require human supervision. We can override anything and very often have to.
My quant background: Excel. Selftaught VBA a month before joining to a basic level. I can read most codes and understand them but can't write. Lots here can though so it's no huge problem. Algo trading is just the algorithms we set doing the trading. HFT is the algo's working on the basis of speed and number of trades. Some of our algos are pretty quick making small amounts on large numbers of transactions, others can make big money a couple of times a day and some have significant tolerances for time. Black box is just number goes in, number comes out no idea what the hell is going on until its a bit late usually. Not much of that on the 10 screens i'm watching at the moment.
Strategy details (sorry, but it can only harm my PnL anything I reveal).
We trade anything that has a line that moves. The algorithms do everything for us. A lot of companies that do the really high frequency stuff are getting a bit screwed over recently, we didn't go down that route and are growing.
Physics Degree (3 years), top UK university (but average final grade - i derailed). Strong maths/analytical skills. National bridge team and built my own startup.
Went into accountancy (audit Big4) hated it and left after a year.
no offense but this is all incredibly vague....
Are you on bloomberg?
If you want to highlight anything you'd like more details on I'll provide more. I won't name the company I work for and I won't provide any strategy related information, anything else I'm fine with.
I am on bloomberg but it's a company account (obviously), I can respond to pm's here though.
any insight you can share into the research process? or what kind of infrastructure is required for research?
Have you traded before? What was recruitment like? Is this the caliber of a jane st/Optiver type shop? Did you have a training course, or are you an assistant trader?
seems like you're in operations/execution, and not a trader.
Boxers or Briefs?
Ok: The research process: It all starts with an idea/observation. you build the logic of the sheet on paper and use it to do the obvious cases from your examples/ideas. There is an element of backtesting but it's close to meaningless. The main thing its good for is to just fire through a lot of data into the sheet to find any gremlins. PnL from backtesting is meaningless. You then test the sheet out on small number of markets/small limits and increase until it either works or has confirmed it's flawed. You continually log your results and observations and make any live adjustments as required. The problem with the amount of data you get is that there will be inevitably some complete coincidental correlations which you may end up acting on without reason. Once you have a strategy that works, unless its very specific to a financial instrument you can apply it to every instrument on every market (within reason).
Most infrastructure is either made inhouse or 3rd party specific sourced. There is a minimum level of kit you need and that has a cost, but its lower than you'd expect.. So your research infrastructure is whatever you want, you can do some really absurd stuff.
Have i traded before? No, career swap from big4 audit (didnt finish training years, hated it). Recruitment: The standard brainteaser stuff and 2 long fit interviews. In a small outfit there is nowhere to hide, you have to fit in. Caliber: We're staffed/run by a lot of high up ex Optiver/IMC spec people, mainly european so not so many US types. They each founded the algo divisions of those companies, so there's a huge experience base in this. I picked this firm over offers from some of the ones in the same league you mentioned because im sat next to major veterans of the industry, not some 2nd year dude passing on a few hints.
Oh sorry i forgot you do my job and not me, i must have misread the job description. I get as close as you can to trading as an algo trader, since technically the computers do the trading.
Boxers, if you're over 18 and wearing briefs, stop. You need the support.
Tranny or Nanny?
What's the difference between desks? What kind of people go for what kind of desk/products? What are their features? By their I guess I mean desks, people and products.
what do you reccomend to an undergrad that wants to get started in algo trading?
Nanny, I dont like surprises.
Difference between desks: My outfit trades as a team, we dont have our own PnL's, so thats a factor to remember when you consider this. There isn't a huge different between the desks for the most part. We don't have one guy trading oil and the dude next to him trading equities. All the sheets really need is supervision and manual intervention if something stupid is going on. That said, we do have specialists watching the specialist sheets. The ex option market makers watch the options sheets, the main classifications for us are intstrument type. I think people really don't appreciate how much we only care about the line and liquidity. Everything else is not important to us. Personally, can only name some equities because I've seen them in the research and when you're making a case about a strategy you need to call it something, and have no idea what any index/commodity closed at within the last year. Something I have noticed is that generally, these places are geek central. Everyone is technically inclined, and that brings with it confidence in what you know, so you do see the people that grew up in options, mainly wanting to do options and are reluctant about anything outside of their comfort zone. My advice would be to not pick a market too soon.
What do I recommend to an undergrad? The only degrees worth having for this are Computer Science, Physics or Maths.Sorry to be brutal but it's what they look for. Financial interest helps but is much less important than you'd expect. The job is problem solving, attention to detail (read testing until destruction) and getting stuff done. The best thing you can do is build an automated bit of software to do a task, in excel or more advanced if you have to, and email it to yourself. Mention it in an interview and download it. Saying I made 15k in a virtual trading account is meaningless, its very difficult to simulate the type of trading we do. PROVE you can build something (can be completely unrelated to trading) from scratch, make it work, and show you've tested it to not do what its not meant to do on bug. It's easier to put on your CV if you did it for the university club you help in.
Example: It would be very easy to script an automatic email generator that reminds people of meetings, subscription payments and newsletters. You'd select who receives it, press a button and it emails everyone. VBA is all you need for that. If you don't know it, learn it. Having some knowledge of other languages is useful too, machine code runs faster so they're the best ones to know, (C# etc.)
So, if I got the "we only care about the line and liquidity" part right, you're saying that you abstract away what's behind the data, you simply look at it as a sequence of numbers, and try to predict what will happen and how to make money out of it? And also, not what will happen on the long run(or why), but rather within the next few minutes or hours? Is this right?
yes, unless its an important part of the strategy which does happen but is pretty rare. Yes we look for any indication about a future price and trade off it. The scalability comes from the fact that once it works you can apply it to every instrument in every exchange around the world. Even 1k a day per instrument type per exchange is easily a million plus a year. We never care about the long run. Stocks can have news events which will change the price to a level that can delete your pnl, positions at the end of the day are avoided like the plague/hedged on alternate exchanges.
correct about the minutes or hours. can even be seconds.
Consequatur voluptatem omnis rerum iusto blanditiis. Ducimus quibusdam sit et consequuntur blanditiis excepturi. Reiciendis est nostrum aliquam ipsum mollitia qui velit eum. Quisquam nesciunt explicabo totam eius doloribus.
Rerum animi fugit similique. Velit voluptatem veniam consequatur ex consequuntur unde recusandae dolore. Perferendis porro nemo recusandae optio sunt. Dolore velit aliquid fuga aperiam facere rerum tempora nam. Doloribus fugit nobis odio. In et repellendus in.
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...