Why Sales & Trading Is Dead As We Knew It

GMngmt's picture
Rank: Senior Orangutan | 489

Would you like to be a wildly successful bulge bracket (BB) trader having more money than you have time to spend? Connected to top investment bankers from shenanigans of Wall Street parties and a living testament to the high-fashion lifestyle! I know I thought like this when I was in junior college and to a lesser degree during my proprietary trading days but those are pipe dreams from yesterday's Wall Street. If you're trying to break into Sales & Trading (S&T) today, "...there's no chance because the market for traders is zero," according to Tom Sosnoff, CEO of Think or Swim. Does this mean that you should give up all together? Not at all, but you're going to need more than an undergraduate business degree even from a top-tier college to landing a start, here's the scoop...



First, I want to break-down S&T and its basic functions because one feeds the other, like clock-work and both are needed to run a department successfully.

* Sales connect buyer & sellers and execute orders from clients acting as middle-markets to generate revenue. These are specifically experienced people with a proven history of relationships & sales and usually networked their way in from adjacent fields.

* Traders use a firm's capital to further profits. In the past, hires to major desks in credit, equity, and foreign exchange might be 1 or 2 whereas in investment banking it's 2 to 5 times that. As of two year ago, the Dodd-Frank Wall Street Reform and Consumer Protection Act has caused a migration of trading activity to hedge funds and a subsequent decrease in the capital pool to

Do you have unique experience or a network in putting the odds in your favor to break-in?

Essentially, S&T is a package deal but here is where the model changes. Going back to Tom Sosnoff,

The investment banking business isn't the same as it was five or ten years ago, it means the game has changed."

Further, he says that banking is a function of spreads and when margins are razor thin for sales they simply can't afford them which effects traders but it's not that Wall Street is not making money. The industry's profitability is strong, so what's the deal?


The nature of S&T is has changed for a number of reasons, a large one is evidenced in my previous article, Why High-Frequency Trading Is Not Breaking The Markets. Concisely, it started with electronic trading, then decimalization, and increased competition from discount brokers. After that, it was Wall Street hiring graduates of the sciences which gave rise to black boxes, the popularizing of algorithmic trading, and high-frequency traders who now make markets. Let's also not forget the expansion of private markets such as credit default swaps and dark pools either as they have also cut into commissions and volumes.

I've also elaborated a bit on some other points of Sosnoff:

* Trading volume has dried up and become less selling-oriented from the brokers-commission system to more technology-oriented via algorithmic trading and outsourcing.
* Demand for Quantitative Analysts (QA) are off-the chart! People who can build financial technology solutions, program algorithms, and create strategies with complex modeling are highly sought and paid.
* Ultimately, markets will adapt, just like organic systems in their environment. Individual and retail investors will learn and adjust to these more complex strategies resulting in irreversible market structure changes.

Are you hesitant now about pursuing a career in S&T?


For today's' S&T prospective, you'll need at the very least major course work in the mathematical sciences, familarity in programming, and a great financial network where you can get insider accessibility and influence to land that open position! Otherwise, you better be well-versed in graduate mathematical sciences, master more than one programming language, and a secondary knowledge in economics or finance to land a QA position at a Hedge Fund in Connecticut.

My advice is, experience talks. Work hard, save money that you can afford to lose, and put up the numbers! Statistically, it's likely you may not end up working at a BB trading desk after all but instead could end up working for a boutique, a proprietary firm like I did, as an entrepreneur, or just individually. Realistically, I think you have better odds with those options at being financially independent and hedge your bets about Wall Street trader fame and instead maxmize their more favorable odds, like a true trader!

What else would / are you pursuing to do versify your career bets?


Sources: Mergers & Inquisitions and Minyanville, June 2012

Comments (143)

Jun 24, 2012

I agree with this post. I guess I made the right choice when I decided to double major in mathematics and finance. I'll be a junior next year and I'm planning on applying to prop shops, bulge brackets, and hedge funds alike. Ideally, I'd like to experience work on a bulge bracket desk for at least a summer, but I know it's getting pretty hard to land those jobs. Any tips besides "network, network, network" would be appreciated (I'm already feverishly networking).

    • 1
Jun 24, 2012

Great post. I agree with this. Whenever a college student asks me for advice on breaking into trading, I tell them, "Take as many CS classes as possible, especially C++, C#, Java, data structure, algorithms."

The game has changed radically in the last 5 years or so, and this trend will only accelerate. Discretionary trading is virtually dead, and almost everything is now done by algos.

Jun 24, 2012
Brady4MVP:

Great post. I agree with this. Whenever a college student asks me for advice on breaking into trading, I tell them, "Take as many CS classes as possible, especially C++, C#, Java, data structure, algorithms."

The game has changed radically in the last 5 years or so, and this trend will only accelerate. Discretionary trading is virtually dead, and almost everything is now done by algos.

At this point, I am proficient with Java, Matlab, and SQL. I've been considering picking up C++, since it is very similar to Java. Are there any other languages worth learning?

Jun 24, 2012
nontarget kid:
Brady4MVP:

Great post. I agree with this. Whenever a college student asks me for advice on breaking into trading, I tell them, "Take as many CS classes as possible, especially C++, C#, Java, data structure, algorithms."

The game has changed radically in the last 5 years or so, and this trend will only accelerate. Discretionary trading is virtually dead, and almost everything is now done by algos.

At this point, I am proficient with Java, Matlab, and SQL. I've been considering picking up C++, since it is very similar to Java. Are there any other languages worth learning?

Matlab is fine for basic quant research and data analysis, but you can't write algos with it. If you are interested in becoming a "real" quant, pick up C++ and C#. Also python is quite useful although secondary compared to the other languages. It's primarily used for scripting, but a lot of prop shops are starting to use it because it's easy to write and debug. My former fund used python almost exclusively for the algos and scripts. The disadvantage is that becuase it's an interpretive language, it's slower than C++/C#/Java, and it's a pain in the ass doing backtesting..

Jun 24, 2012

I understand how this would at least force trading to move away from the bigger banks down the road, but how would this affect sales roles? And would programming be at all necessary to become a salesman?

I'm like one of them marriage counselors. Charge by the hour to tell some fool he needa bring some flowers home. Then charge another hour telling the bitch she oughta suck some cock every little once in a while. Keep a marriage strong like that. -Prop Joe

Jun 29, 2012
Proposition Joe:

I understand how this would at least force trading to move away from the bigger banks down the road, but how would this affect sales roles? And would programming be at all necessary to become a salesman?

To answer your question, when a trader wants to execute a trade, it's the salesman that makes it happen. I'm not talking about round lots either but block trades (10k+ shares) which are executed with institutional buyers (IB). Sure, a routing system could execute the trade at an average price but they're already long-standing close relationships with IB that sales people continue profitable relationships with in turn for their commissions and the trader's profit.

However, the problem with these salesmen (i.e. brokers) is that their ability to make-markets is being diminished by high frequency trading employing these same strategies which has driven narrower spreads and reduced volumes. This is the same technology that employs the strategies of former desks and begun the fundamental shift from securities traders to quantitative analysts. Subsequently, the need for trade execution by a salesman diminishes as the algorithm can better ascribe this more efficiently with less overhead. As per my current state of knowledge, I think in the near future the increasing rate of quantification of trading methodology is eventually going to usurp sales all together...as it relates to the trading business.

In my opinion, programming would not be at all useful if you're going to pursue a sales position. It doesn't employ communication or marketing skills which are keys to building successful relationships, instead focus on communications / marketing / persuasion tactics. These are matters that technology cannot do, human relationships, and that's fundamentally what drives markets. If you're good at it, you'll always have a job, even if not in finance as the industry continues to change these are very marketable skills.

Who Am I? | See what GMngmt is all about at About.Me

Jun 29, 2012
GMngmt:
Proposition Joe:

I understand how this would at least force trading to move away from the bigger banks down the road, but how would this affect sales roles? And would programming be at all necessary to become a salesman?

To answer your question, when a trader wants to execute a trade, it's the salesman that makes it happen. I'm not talking about round lots either but block trades (10k+ shares) which are executed with institutional buyers (IB). Sure, a routing system could execute the trade at an average price but they're already long-standing close relationships with IB that sales people continue profitable relationships with in turn for their commissions and the trader's profit.

However, the problem with these salesmen (i.e. brokers) is that their ability to make-markets is being diminished by high frequency trading employing these same strategies which has driven narrower spreads and reduced volumes. This is the same technology that employs the strategies of former desks and begun the fundamental shift from securities traders to quantitative analysts. Subsequently, the need for trade execution by a salesman diminishes as the algorithm can better ascribe this more efficiently with less overhead. As per my current state of knowledge, I think in the near future the increasing rate of quantification of trading methodology is eventually going to usurp sales all together...as it relates to the trading business.

In my opinion, programming would not be at all useful if you're going to pursue a sales position. It doesn't employ communication or marketing skills which are keys to building successful relationships, instead focus on communications / marketing / persuasion tactics. These are matters that technology cannot do, human relationships, and that's fundamentally what drives markets. If you're good at it, you'll always have a job, even if not in finance as the industry continues to change these are very marketable skills.

Awesome thanks. When you say they are marketable skills, do you mean within finance? If so would it make more sense to pursue something like wealth management or private banking?

I'm like one of them marriage counselors. Charge by the hour to tell some fool he needa bring some flowers home. Then charge another hour telling the bitch she oughta suck some cock every little once in a while. Keep a marriage strong like that. -Prop Joe

Jan 18, 2013

yep, financial markets will cease to exist in the next 10 years, as trading becomes unlawful and traders become obsolete

"Every man should lose a battle in his youth, so he does not lose a war when he is old"

Jan 18, 2013
RichardPennybags:

yep, financial markets will cease to exist in the next 10 years, as trading becomes unlawful and traders become obsolete

I've been saying this for the last 50 years and no one believes me it's like i'm talking to myself or something

Jan 18, 2013

you'll still need traders...the industry has just been scaling back alot

Jan 18, 2013
Crane:

you'll still need traders...the industry has just been scaling back alot

The question is: will those traders be human beings?

Jan 18, 2013

Learn programming

Jan 18, 2013

What about Sales for bbs?

Jan 18, 2013

automation isn't only for dummies

If the glove don't fit, you must acquit!

Jan 18, 2013

Physical trading will always need human beings for any and all commodities. Just because you can automate processes does not mean you can forgo the human element. We need marketers, schedulers, traders,brokers etc...if we become too heavily dependent on computers in the S&T world, we will fall victim to them one day. For yourself, it is a matter of preference. Chances are, you will have to learn programming at some point in your life -- whether it be scripting (Python, Pearl), C++, C#, or something as basic as VBA. For instance, I have a finance background, but building out our crude processes which require automation and some element of coding -- even at the basic VBA level (which I would not consider coding, but it's Excel's language).

Jan 18, 2013

yes discretionary trading is dead just like it was 10 years ago when i graduated college and everyone said the same thing. Soon the only opportunities will be for robots who will do everything from analyzing the news to sexually harassing the secretaries.

    • 3
Jan 18, 2013
Bondarb:

yes discretionary trading is dead just like it was 10 years ago when i graduated college and everyone said the same thing. Soon the only opportunities will be for robots who will do everything from analyzing the news to sexually harassing the robot secretaries.

fixed that for ya ;)

Jan 18, 2013

If you are world class, they'll hire you on intellect and potential alone. Otherwise you basically come off as a lazy ass who is not willing to do any work. What are you gonna do your trading on? excel spreadsheets? Point and click trade? You might find the latter at one of those prop-firm arcades...

thestockwalker:

Seeing these threads on WSO and the comments basically saying that the trading industry is contradicting, and will be pretty much "nowhere" in ten years time (compared to other industries) with these kind of regulations I am getting more and more unsure if I really want to pursue a career in trading.

Also several people advised me to get into programming, tho that's something I would really not keen on doing/learning, not to mention that I am about to start an Economics degree (no minors here in the UK).

So what do you guys think about the S&T industry, and also if you could elaborate on which desk would be appropriate for someone who got basically no knowledge/passion for programming that would be great.

Thanks!

Jan 18, 2013

I'm amazed at the ignorance in this thread. Yes, parts of trading are more automated such as vanilla equities and rates... but there's a lot more out there than that and it's not changing anytime soon. I bet people were having this same discussion 20 years ago and the chicken littles were still yelling that the sky was falling.

Jan 18, 2013
Revsly:

I'm amazed at the ignorance in this thread. Yes, parts of trading are more automated such as vanilla equities and rates... but there's a lot more out there than that and it's not changing anytime soon. I bet people were having this same discussion 20 years ago and the chicken littles were still yelling that the sky was falling.

I'm equally amazed at your ignorance. Sure, most people in this thread are exaggerating the near-term threat to S&T jobs. But you're absolutely kidding yourself if you believe that we'll be having the same discussion in 20 years.

More likely, kids in born in 2013 will be asking their parents how trading ever worked before computers were able to automate the entire process. In the same way that banking and hedge fund analysts born in the early 90s are mystified by the prospect of researching companies without digital filings or building models without Excel.

The amount of processing power in the average computer doubles every 18 months. Surely, you are familiar with the power of compounding interest. How long is it before computers are capable of understanding trends and patterns in the market that we aren't even aware of?

Jan 18, 2013

Computers already are and do that, that is the nature of HFT trading.

It is very difficult to have this argument with someone who has never stepped foot on a trade floor, even as an intern time spent on a desk will show what the job is really about. The outside view is that the job is entirely execution and short-term patterns, while truly a lot more encompasses what a good trade shop does. While yes I agree that the execution side is at risk and will continue to do so, then again you could say that for many areas of finance that are just execution.

Next, to assume simple that in the last 20 years we have made such gains in automation etc that the same gains are still possible in the next 20 is totally silly. You are basically assuming the learning curve has not plateau'd or decreased over time and this is a silly argument. The fact of the matter is traders main job is to deal with how to best hedge and price "risk" in a marketplace, it is very difficult for a computer to do that. You could go on and build HFT supermachines that just go to war with each other everyday, but you can never get rid of people looking after the machines, unless the matrix comes alive. Also there is a human emotion ("gut") that comes to how people hedge and why, otherwise why would you ever hedge?

Jan 18, 2013

The issue with having this discussion, which seems to crop up here every single week, is that it has become apparent that very few on this forum actually know what a sell side trader does.

    • 1
Jan 18, 2013

All these threads end up being the same. People who have never set foot on a trading floor saying it's dead and traders won't exist in 10 years and the people who actually trade for a living saying that it's not. You only have to read some comments (I particularly loved the algo one) to realize that most of the people talking have absolutely no idea what a trader does It's fucking ridiculous. I don't understand why you guys even ask...

    • 1
Jan 18, 2013

The human brain is basically a computer, but with more primitive processing power. But the human brain has two main inherent advantages: creativity and sensory analysis. The creativity aspect of course involves devising new ways of thinking, and the sensory aspect involves analysis of people among other things. In this way the human brain is superior. In all other ways computers are potentially superior, but maybe not yet in some areas, such as with things that require more abstract thinking or analysis of language and text as opposed to numbers and hard data.

Jan 18, 2013
Going Concern:

The human brain is basically a computer, but with more primitive processing power. But the human brain has two main inherent advantages: creativity and sensory analysis.

Excellent, and they tie together in something the brain is overwhelmingly superior at: parallel processing. Computers are good at linear algorithmic processing....perform a function, do the next one, many times a second. The brain references all sorts of data, concious or otherwise....simultaneously, and references those inputs in reference to each other. There are computers that can do this, sort of, but the sheer diversity of the spectrum of stimuli that the brain integrates into its processing is staggering. Meanwhile, computers actually only process 0's and 1's very quickly.

Check out this article:
http://library.thinkquest.org/C001501/the_saga/com...

Bottom line is that you'd need a LOT of computers and a ton of hardware for artificial intelligence to process the world the way the brain does. A computer can add 1 + 1 faster, but it has no idea what that actually means or what to do with it. Trading requires understanding reality, and unlike chess, markets outside of the low hanging fruit of HFT and a few plain vanilla products are far too complex for a computer to understand.

Execution and broker/trader type roles (see: floor trader) are becomming extinct as the process of trading becomes more automated. But actual decision making still rests with humans. If you up the ante to investing, computers don't stand a chance at all.

Jan 18, 2013

I agree with Northsider on this one. I do not think that trader on the sell side will be the same job 20years from now.
You won't be making market anymore, algo will do that. You will just make sure that there isn't price abnormality or maybe looking for opportunities offered by "irrational exuberance" but you won't quote price anymore.
If you want a glimpse of sell side trading in 20 years look no further than HFT prop shop (Getco, Optiver, IMC and the like).

Jan 18, 2013

I am a sell side trader/derivatives market maket with a few years experience, and I think trading is dying, mainly the issue of margin compression and increasing regulation, not automation. There will always be the need for a trader - spot/cash traders are effectively prop traders now anyway with how sophisticated algos are getting. What skills a trader requires will most likely be geared towards more technical knowledge going forward, which is why I advise all new/young aspiring traders to have multiple skills that can open doors into other areas. Not because they will get fired one day but the lucrativeness of trading has been waning for years in some products.

One thing I will say is I doubt a computer will ever really be able to run a derivatives book, and the more exotic/bespoke the products within that book the less likely that is to happen.You're talking about sophisticated AI in that scenario - if such technology does come in to place, we are either looking at a Terminator-esque situation or none of us will most likely have to work again while robots take over so it's a non-issue.

Jan 18, 2013

To echo ya boy above... It's important to refine "trading" a bit. "Market making", especially in liquid markets, is a very different beastie to prop trading, where the number of degrees of freedom is much higher.

    • 1
Best Response
Jan 18, 2013

I think I didn't explain myself well. My point is that certain ares of sell-side trading will be more automated. Sure. No doubt. But to give you an idea, the only desks that execute everything via algos are cash equities and spot FX, and they are not both fully automated in every BB, at least in London. There are certain products that are impossible to automate because of their own nature, and some that are impossible because of certain market conditions, the main one being illiquidity. This last factor can obviously change, which is where the room to grow is, but the first one isn't which is what you guys can't understand. On top of continuing with the automation in cash eq and spot fx, in my opinion, there are few areas that have the potential to be automated in the near future, maybe one or two desks apart from cash eq and sport fx, and that's a very small percentage of a sell side trading floor anyway.

The biggest flaw in your argument is that it lacks dimensionality. You are only considering one dynamic component, which is my second and most important point. Yes, products that already exist will be automated, but not that many. Fine. But you are failing to consider the fact that new products are created every day. And these products can't start trading in auto mode, they need to develop from a bespoke solution to 1 client to a market, and that takes decades, no matter how good the product is (it took fx and cash eq centuries to get to this point). And computing power doesn't have anything to do with it anymore (of course it did in the past, I'm not denying that), the reason swaps trading is not automatic has nothing to do with it not being technologically possible. For developing a market you need a sell-side desk literally making the marke. Manually. I mean, take CDSs as an example, they were created roughly 20 years ago. JPM decided it was an interesting product, so they decided to pitch it to other clients, it kept growing, they created a market making desk and now cds mm employs what? hundreds of people at least?. It's still far from possible to automate a cds mm process. Swaps have been around for 30 years and you still can't automate them, and in their development they have become one of the most (if not the most) liquid instrument in finance and created thousands of jobs.

So my point is that yes, I agree that you can't deny that some products will be fully automated in the future (not that many on the short term) and therefore some traders will be out of that job. But you also can't deny that some new products will be created, and these products will develop markets that will need to employ traders, and the first ones will need to come from other desks. It's no coincidence that there are more traders now than 20 years ago. There are more products and more markets. And that's been a strong and clear trend if I've ever seen one.

    • 3
Jan 18, 2013
Maximus Decimus Meridius:

Illiquid, exotic or bespoke products cannot / will not be automated, due primarily to all of the issues present today.

Revsly:

See what Maximus wrote. But essentially:

Automated market making requires liquidity and transparency in the market.

You're welcome to call me crazy, but I think these represent very narrow views of the future of technology. Sure, highly structured, bespoke trades and some sales roles will likely remain human responsibilities for negotiation and interaction purposes, but that misses the entire point of this argument. A great number of traders make markets in products that aren't automated because they are too illiquid for our familiar programmatic trading interfaces to deal with. It is in these markets where people are suggesting spreads will tighten and jobs will be lost.

But you're assuming virtually zero improvement in the ability of computers to recognize and understand patterns from limited data when you say that "automated market making requires liquidity and transparency in the market." Applying today's rules of thumb to the conditions of 20 years from now is dubious, at best.

Even today, we have developed machines that operate on levels far beyond the von Neumann architecture (your traditional arithmetic logic unit) and programming. Content-addressable memory and other recent innovations have allowed giant leaps in neuromorphic computing.

In 2011, IBM was able to simulate 4.5% of the human brain using 150,000 interconnected processors. In the same tests, it completely simulated the brain of a mouse with 500 processors, that of a rat with 2,000 processors, that of a cat with 25,000 processors. In even the most conservative estimates of growth of processing power, that means that we will be able to completely simulate the neurological interactions of the brain by 2020. Within the next 20-30 years, that processing power could conceivably fit in the palm of your hands. A consortium of biologists, neurologists, computer programmers, mathematicians and engineers are currently working on brain machinization.

In the meantime, incredible advances computerized "reasoning" have been made in the past several years.

Computers exist today that can pass the infamous Turing Test - that is, human beings who have conversations with a computer and another human being are unable to determine which was the computer and which was the human. Certainly you wouldn't say that conversation and language is programmatic; a computer programmer cannot know ahead of time what topics will be brought up, what questions will be asked, and in what form they will be asked. Nevertheless, today's computers are able to adapt and intuit the answers to highly complex, philosophical or even self-referential questions and statements that occur in the course of normal conversation.

Computers exist today that can drive cars and fly planes far more adeptly than human beings. Certainly you wouldn't say that the obstacles and unexpected events experienced in everyday driving are programmatic. A programmer can't write the C++ version of "don't hit things" and be done with it. Driving requires thousands of calculated, even instinctive reactions - split-second reasoning skills that we develop with experience. Yet computers are capable of making those decisions.

These are just a couple examples of what computers are capable of within the bounds of current processing power. There is absolutely no telling what will be possible as computing power continues to double and redouble in the coming years.

Almost by definition, human beings making markets in illiquid products are using experience, recognizing patterns and reacting to events in order to determine pricing. You're kidding yourself if you don't think a machine is capable of replicating this. Observing that the current rules of computer programming don't allow such innovations is not an intellectually rigorous way to think about technology.

    • 1
Jan 18, 2013

That's exactly the point I'm making. Automated market making requires liquidity and transparency in the market. That's fine in very liquid markets where there is two way interest (take cash equities or spot fx). That is not really true in the derivatives. That's the key thing you seem to keep missing... we're not debating cash market movement to algos, but discussing how far/difficult that model would be in OTC/bespoke. And to your "markets more liquid now and it resembles nothing like 10 years ago," you'd be surprised. The market I trade is wider and less liquid than 10 years ago really (pre-fin crisis) and doesn't look a hell of a lot different. Models have improved, and strategies have changed... but not in the algo direction.

If you can find a computer to profitably market make and risk manage an options portfolio in an OTC market, let me know. I'll boot that up and watch TV all day.

    • 1
Mar 20, 2018
Revsly:

If you can find a computer to profitably market make and risk manage an options portfolio in an OTC market, let me know. I'll boot that up and watch TV all day.

I swear I was told that this exists now? No?

Jan 18, 2013

I am surprised that Illiniprogrammer has yet to respond.

Jan 18, 2013

Singularity is coming soon. I just hope I'm a cyborg by then.

Jan 18, 2013
Going Concern:

Singularity is coming soon. I just hope I'm a cyborg by then.

hope is not a strategy. Plan now.

Jan 18, 2013

I'm looking forward to waging war on these self-aware trading programs

Jan 18, 2013

This is taking on larger philosophical questions outside of the trading debate. I think some of you are talking past one another.

Anyway, I'm looking for John Conner.

Jan 18, 2013

Haha NorthSider, it came even faster than you'd predicted, see this: http://www.kurzweilai.net/ibm-simulates-530-billon....

That being said, there is a huge difference between mapping the architecture of the human brain and recreating its functionality. One of the comments in the article above mentions how scientists can't even simulate a worm's brain, the architecture of which has been known for many years (302 neurons and 7000 synapses) - never mind a human's.

From that, I think it's clear that there will be two distinct kinds of brain for a long time - that of a human, and that of a machine. Computer brains (if you will) make money largely by preying on human brains' weaknesses. If humans leave the market, how do you expect a computer to turn a profit?

Nevertheless, some excellent points there. +1 SB

Jan 18, 2013
hmm2:

Haha NorthSider, it came even faster than you'd predicted, see this: http://www.kurzweilai.net/ibm-simulates-530-billon....

They haven't achieved what I predicted. Nevertheless, this article demonstrates the milestones being achieved in rapid succession. The velocity of innovation is only increasing.

That being said, there is a huge difference between mapping the architecture of the human brain and recreating its functionality. One of the comments in the article above mentions how scientists can't even simulate a worm's brain, the architecture of which has been known for many years (302 neurons and 7000 synapses) - never mind a human's.

The point of IBM's SyNAPSE project is not to recreate the brain's functionality (the Watson project is much closer to what you're describing). It is to create a disruptive architecture for computer activity. What the IBM scientists are creating doesn't (and isn't meant to) resemble human brain activity. It doesn't operate in real time and it isn't a biologically realistic representation of brain activity. What it does accomplish is a total reinvention of von Neumann machines, it's a machine that is capable of parallel processing, capable of adapting itself. More than anything else, it's evidence that the power of computing is already starting to extend beyond the realm of arithmetic logic and drawing from the power of creative reasoning.

From that, I think it's clear that there will be two distinct kinds of brain for a long time - that of a human, and that of a machine. Computer brains (if you will) make money largely by preying on human brains' weaknesses. If humans leave the market, how do you expect a computer to turn a profit?

Suffice to say, I don't agree with any of this. You're assuming that the human brain is capable of some higher level computation that cannot be replicated inorganically. Creative thought is thoroughly within the realm of computational power. The brain is nothing but an incredibly sophisticated biological computer, I see no compelling reason why its capabilities cannot be replicated.

Re: humans leaving the market, I'm not clear on what you're referring to. Why would humans leave the market? They just decide they don't want ROI anymore?

I appreciate the SB though! :)

Jan 18, 2013

My god, this debate goes on like once a week here. Truth is...over the long-term technological unemployment is going to soar.

Before anyone goes..."oh that's what they've been saying since the industrial revolution...blah blah"...it is happening for real this time. It may sound strange, but we will be slowly transitioning to a Post-Scarcity society.

IBM's Watson is now a better diagnostician of lung cancer than the all-star team of oncologists at Memorial Sloan Kettering that helped train it....MUCH better. New programs are already performing E-Discovery... a labor intensive and high billable activity of Big Law. Lawyers and Doctors are the stereotypical examples of white-collar work and just because your job can't be outsourced to Hal9000 yet...it is naive to think that it either won't be or that organizations won't be running super lean when it comes to head count. Adapt or die, I guess.

PS I'm not some quant nerd with a hard-on for tech either, C++ to me sounds like one of my grades in Calc II.

Please don't quote Patrick Bateman.

Jan 18, 2013

There is those who believe in EMH and all that fancy academic stuff we read and learn in books. Then there are those who believe Buffet is not the luckiest man alive, Paul Tudor Jones does not just take crazy bets and so on.

You keep talking about duplicating the human brain and a rational person. The truth most traders believe and work in a world with "irrational beings" that is what makes a marketplace that is why one person will hedge at "X" and another at "Y". If we had a marketplace made of perfectly sane, rational beings then you would never have anything as a "il-liquid market" or "transparency issues".

For instance many commodities most people know the exact level a majority of the market will hedge, most producers even state we can make positive ROI at "XYZ" price. So the question becomes why do we ever have a upwards sloping forward curve? Should not a rational computer style model/brain tell us at some price the market can find infinite supply and producers could make $$$? Actually if you ask most engineers they say just that and any trader will say ok sell me supply 5years from now, cuz the forward curve is disagreeing with you. Why because the market is made of irrational people working in not 100% certain solution. This is why we
have supply shocks, trends, bubbles, wave theory etc... A rational brain would not work in such a market.

This is also why computerized trading works best with HFT. Where do we have the most certain conditions, the speed of light, now move that model to 5 years out and let's see what you could do.

    • 1
Jan 18, 2013
marcellus_wallace:

There is those who believe in EMH and all that fancy academic stuff we read and learn in books. Then there are those who believe Buffet is not the luckiest man alive, Paul Tudor Jones does not just take crazy bets and so on.

You keep talking about duplicating the human brain and a rational person. The truth most traders believe and work in a world with "irrational beings" that is what makes a marketplace that is why one person will hedge at "X" and another at "Y". If we had a marketplace made of perfectly sane, rational beings then you would never have anything as a "il-liquid market" or "transparency issues".

For instance many commodities most people know the exact level a majority of the market will hedge, most producers even state we can make positive ROI at "XYZ" price. So the question becomes why do we ever have a upwards sloping forward curve? Should not a rational computer style model/brain tell us at some price the market can find infinite supply and producers could make $$$? Actually if you ask most engineers they say just that and any trader will say ok sell me supply 5years from now, cuz the forward curve is disagreeing with you. Why because the market is made of irrational people working in not 100% certain solution. This is why we
have supply shocks, trends, bubbles, wave theory etc... A rational brain would not work in such a market.

This is also why computerized trading works best with HFT. Where do we have the most certain conditions, the speed of light, now move that model to 5 years out and let's see what you could do.

Ugh. Clearly the idea of computers operating in any architecture other than arithmetic logic just doesn't click with most people. When you say "irrational", you mean "using non-arithmetic logic"; do you realize that the posts you're bemoaning here are arguing that computers are becoming more and more capable of that "irrational" part?

No, S&T is not dead. No, modern von Neumann computers cannot effectively make markets in exotic and illiquid markets. You're entirely right there. Where you go wrong is in making the - remarkably human - assumption that future computers will act in the same way that the Dell you bought the other day does. Using the past to predict the future... when has that ever gone wrong?

Jan 18, 2013

I don't want to write another long post because it's quite clear the discussion has gone off-topic, but I want to say a couple of things.

- The problem is not that people don't understand how non arithmetic logic works, it's that many people here don't understand how the market making and trading process works.

- The argument with liquidity and transparency is not about rationality and irrationality and computer architectures. Very simplified, it's about inputs. No matter what kind of computer architecture you use, a computer needs inputs. And the fact is that inputs don't exist in certain markets. That's what makes it tremendously different from diagnosing a patient or driving a car to use your examples. There is no such thing as a computer working in a vacuum of information.

- You would also need to assume very developed markets from a geographical perspective, otherwise interaction would fail.

-Certain parts of trading, typically commodities have an actual physical component to them, which I don't think you can resolve unless we are talking about very perfected robots.

- You dismissed biology as a silly component when it's proven (unlike the future of tech) that there is a huge biological component to trading.

- NorthSider, your last argument is easily reversible. You are also using the past evolution of technology to predict how technology is going to evolve in the future. Maybe no one will develop new architectures or increase in computing power will be put to a stop. Huge change in direction has happened very often in technology. And I can't see how you can claim that people can't assume that computers will work in the same fashion as today (I want to say again that I don't, I agree with you on that part) but you can assume that the evolving trend can be predicted based on the past.

Jan 18, 2013
Maximus Decimus Meridius:

- The problem is not that people don't understand how non arithmetic logic works, it's that many people here don't understand how the market making and trading process works.

It most certainly was (and obviously still is, see below) a problem with understanding how non-arithmetic logic works.

- The argument with liquidity and transparency is not about rationality and irrationality and computer architectures. Very simplified, it's about inputs. No matter what kind of computer architecture you use, a computer needs inputs. And the fact is that inputs don't exist in certain markets. That's what makes it tremendously different from diagnosing a patient or driving a car to use your examples. There is no such thing as a computer working in a vacuum of information.

Case in point. This entire argument revolves around the notion of computers using linear, arithmetic logic, while humans have - apparently exclusive - access to creative thought. Quickly:

The human brain doesn't spontaneously and inexplicably generate outputs. Taking your choice of the word "inputs" at face value, neurological reactions are inextricably linked to inputs: sensory data is consumed and interpreted, leading to chemical and physical reactions. Equally obvious is the fact that trading in illiquid and opaque products / markets requires input to generate decisions (unless you're arguing that trading decisions are elicited by pure randomness). Therefore, the only coherent interpretation of your argument is that you assume some distinction between the types of inputs that a human and a machine can process.

Axiomatically, then, your argument flows roughly as follows:

P1: There is a distinction between "hard data" inputs (read: arithmetic / quantitative) and "soft data" inputs (read: qualitative / sensory / experiential).

P2: Trading illiquid / opaque products requires interpretation of "soft data", due to a dearth of "hard data".

C1: A computer is incapable of trading illiquid / opaque products.

C2: Human beings are capable of trading in illiquid / opaque products.

You'll note the embedded links / premises:

L1: Computers are incapable of interpreting "soft data".

L2: Humans are capable of interpreting "soft data".

The only possible warrant for L1 is that non-von Neumann computers are only in their infancy and that the overwhelming majority of modern machines operate using arithmetic logic. That's precisely the point in contention, and the inability to understand non-arithmetic logic appears to be the confusion.

FWIW, I find P1 (although seemingly reasonable at first glance) to be fairly dubious as well. The chemical reactions in the brain that give way to qualitative interpretation are quite readily quantified. In that sense, the primary achievement of the brain is the ability to convert qualitative / sensory / experiential information into "quantitative" (I use that term for clarity's sake, though there are certainly better words here) reactions. That basically boils down to the presumably tautological observation that neurology / chemistry / biology are sciences that study empirical information (brain reactions, in this case). But that's a debate for another day.

- You dismissed biology as a silly component when it's proven (unlike the future of tech) that there is a huge biological component to trading.

You obviously didn't follow the train of thought with that comment (probably my fault for poor diction). Insofar as you would agree that the organic material (the physical stuff) of which neurons and brain tissue are comprised is not involved in trading, you agree with me on this point.

- NorthSider, your last argument is easily reversible. You are also using the past evolution of technology to predict how technology is going to evolve in the future. Maybe no one will develop new architectures or increase in computing power will be put to a stop. Huge change in direction has happened very often in technology. And I can't see how you can claim that people can't assume that computers will work in the same fashion as today (I want to say again that I don't, I agree with you on that part) but you can assume that the evolving trend can be predicted based on the past.

I'm merely using ongoing revolutions in technology and (what I find to be) fairly straightforward scientific observations to suggest that: 1) "sentient" (I hate that word, but I use it in the technical sense here) computing is possible, and 2) we are well on the way to developing the technology necessary in said computing. Your argument ignores widely reported advances in computing technology and focuses on the familiar, everyday von Neumann machines to project the future of systems architecture (and, by proxy, the bounds of machinization).

Whether consciously or unconsciously, you (and several others here) are stuck on the movement from arithmetic logic units to neuromorphic computers.

Jan 18, 2013

Ignoring the sensationalized bullshit, I feel the article is valid to a larger extent than one may be willing to accept: http://techcrunch.com/2013/06/01/after-your-job-is...

And due to the tools of natural language processing, computers can draw inferences from the news more efficiently than any human and more effectively than a majority of humans,

Jan 18, 2013

Why is everyone focusing on automation? It's just mind boggling.

Flows are down significantly on previous years in almost, or dare I say, all asset classes and even if this is not the case, the amount of pigeons vying for the crumbs remaining have multiplied. If I recall correctly (which I don't normally) spot FX flows are actually at all time highs but assuming profitability has remained the same (razor thin as is), it's split between more market participants than ever before.

Regulation has increased the cost of doing business - forcing markets to become more transparent, provide more information to clients (you want the 'mid price'?), funding costs between a trading desk and internal treasury are going to weigh down on profitability/ increase opportunity cost as a whole due to Basel III requirements, whatever happens to prop trading etc. Only banks with an already existing critical mass of infrastructure, flow and talented people will truly be able to ride this out by absorbing the flow of banks unable to remain competitive with these emerging dynamics.

I am obv referring to S&T - there will always be a healthy appetite by humans from all walks of life to leverage their savings into further $$$$, so the hedge fund/prop trader space will always have a heart beat. I'd like to think a successful trader in today's market has the drive and intelligence to succeed down another path if things don't work out.

Automation should be the least of anyone's concerns.

    • 2
Jan 18, 2013

I am w/Maximus on this... In fact, generally speaking, what is wrong with using the past to predict the future? Come to think of it, what is there, other than one's past experiences, that can help one predict the future? Imagination which is completely divorced from past or present experience is rather worthless, IMHO.

Jan 18, 2013
thestockwalker:

the trading industry is contradicting

Ahem, I'm sure you meant "contracting"

Jan 18, 2013

You know the nice thing about robot analysts? They don't need a bonus or any other pay, and they give out infinite face time. Now that's what I call top bucket.

Jan 18, 2013

regulation will hurt trading more than HFT/Algos. There are even Algo platforms that don't require programming and are more or less developed by traders/analysts but im not sure how many big institutions use them. I know of day traders being profitable as click traders so i dont see why prop shops and funds can't do it..

alpha currency trader wanna-be

Jan 18, 2013

Ultimately I think UFOinsider is right on this one. It might happen eventually but not soon.

Riddle me this broskis: how come there's no computer program that has mastered poker, like the Deep Blue computer which mastered chess over 15 yrs ago? The answer is pretty obvious.

Jan 18, 2013
Going Concern:

Ultimately I think UFOinsider is right on this one. It will happen eventually just not soon.

Riddle me this broskis: how come there's no computer program that has mastered poker, like the Deep Blue computer which mastered chess 15 yrs ago? The answer is pretty obvious.

There are dozens of computer programs that employ near-"perfect" strategies at poker. It was one of the catalysts that led to the destruction of the online poker industry over the past 10 years in the US. The difference is that "success" in poker is defined quite differently than it is in chess. An incredibly talented tournament professional might end in the money in 1/8 of the tournaments he/she enters. In order to achieve success in poker, you need to have a sufficiently large sample of hands to demonstrate a consistent edge. You can't just "beat" someone at poker every time. You can in chess.

Jan 18, 2013

I'm beginning to think that NorthSider is a computer. No human types out that lengthy of responses.

    • 2
Jan 18, 2013
SirTradesaLot:

I'm beginning to think that NorthSider is a computer. No human types out that lengthy of responses.

LOL. That would explain a couple things.

Jan 18, 2013
SirTradesaLot:

I'm beginning to think that NorthSider is a computer. No human types out that lengthy of responses.

http://narrativescience.com/

Please don't quote Patrick Bateman.

Jan 18, 2013

NorthSider, while I agree with many of your points, there are two things which I'm skeptical about. One is the pace at which you imply technological innovation is going to progress. The second (which is related to the first) is the significance of the numerous examples you provide where computers have become more effective than virtually all humans.

Regarding the speed of technological innovation - you claim (and I agree) that over the last 40 years, we have consistently overestimated the pace of scientific discovery while underestimating the pace of technological innovation. This is because the vast majority of technological innovation comes in the form of better or more varied implementation of already known scientific concepts. Take for example Moore's Law, which states that the number of transistors in an integrated circuit doubles every 2 years, has held for the last 50 years because the theoretical concepts underlying the engineering largely have not changed. For a long time it has been known that computation could be performed with single atoms (and possibly smaller), yet today's transistors are still comprised of millions of atoms. So engineers have more or less been squeezing the juice out of the same idea for half a century (and we probably still have another half century to go), but they have been operating off the same concepts. When was the last time the world operated under a different concept of computing? Well...that was the age of mechanical computing, with things like the mechanical calculator (invented in the early to mid-1600s)...that age was around for at least 300 years before a conceptual shift occurred. Technology is more or less applied science, and there are certain breakthroughs in technology which cannot be made without corresponding breakthroughs in science. As you said, the pace of scientific discovery is slowing. And I'm going to bet that developing computers which will truly threaten the vast majority of white-collar jobs is going to require at least one more conceptual shift, if not more.

Regarding the examples you provide, I refer you to a quote by Paul Allen (co-founder of Microsoft): "While we have learned a great deal about how to build individual AI systems that do seemingly intelligent things, our systems have always remained brittle--their performance boundaries are rigidly set by their internal assumptions and defining algorithms, they cannot generalize, and they frequently give nonsensical answers outside of their specific focus areas." Full article: http://www.technologyreview.com/view/425733/paul-a....

So while computers have become very good at playing chess/poker, driving cars, flying planes, diagnosing diseases, sifting through documents, etc. they have until now become good at these things precisely because we designed them to be. But the human mind has a way of making intelligent insights and connections even when we are not consciously trying to do so...though now this is veering off toward subconsciousness land so I'm going to stop now. :)

Jan 18, 2013
hmm2:

NorthSider, while I agree with many of your points, there are two things which I'm skeptical about. One is the pace at which you imply technological innovation is going to progress. The second (which is related to the first) is the significance of the numerous examples you provide where computers have become more effective than virtually all humans.

Regarding the speed of technological innovation - you claim (and I agree) that over the last 40 years, we have consistently overestimated the pace of scientific discovery while underestimating the pace of technological innovation. This is because the vast majority of technological innovation comes in the form of better or more varied implementation of already known scientific concepts. Take for example Moore's Law, which states that the number of transistors in an integrated circuit doubles every 2 years, has held for the last 50 years because the theoretical concepts underlying the engineering largely have not changed. For a long time it has been known that computation could be performed with single atoms (and possibly smaller), yet today's transistors are still comprised of millions of atoms. So engineers have more or less been squeezing the juice out of the same idea for half a century (and we probably still have another half century to go), but they have been operating off the same concepts. When was the last time the world operated under a different concept of computing? Well...that was the age of mechanical computing, with things like the mechanical calculator (invented in the early to mid-1600s)...that age was around for at least 300 years before a conceptual shift occurred. Technology is more or less applied science, and there are certain breakthroughs in technology which cannot be made without corresponding breakthroughs in science. As you said, the pace of scientific discovery is slowing. And I'm going to bet that developing computers which will truly threaten the vast majority of white-collar jobs is going to require at least one more conceptual shift, if not more.

Regarding the examples you provide, I refer you to a quote by Paul Allen (co-founder of Microsoft): "While we have learned a great deal about how to build individual AI systems that do seemingly intelligent things, our systems have always remained brittle--their performance boundaries are rigidly set by their internal assumptions and defining algorithms, they cannot generalize, and they frequently give nonsensical answers outside of their specific focus areas." Full article: http://www.technologyreview.com/view/425733/paul-a....

So while computers have become very good at playing chess/poker, driving cars, flying planes, diagnosing diseases, sifting through documents, etc. they have until now become good at these things precisely because we designed them to be. But the human mind has a way of making intelligent insights and connections even when we are not consciously trying to do so...though now this is veering off toward subconsciousness land so I'm going to stop now. :)

I submit to you as evidence of my humanity that I am growing fatigued of going back and forth re: the future capabilities of computers. So rather than respond piecemeal to these posts and field Paul Allen's softball criticisms of Kurzweil (if you honestly desire answers to Allen's critiques, just read Kurzweil; you'll soon realize that Allen virtually ignored all of Kurzweil's arguments), I'll just leave you with this summation:

It strikes me as truly shocking that many on this thread seem so thoroughly unimpressed by the existing advancements in technology. Things that were years ago unfathomable are now routine elements of our daily technology. What would have been described as "computerized sorcery" in 1995 is being dismissed by posters in this thread as "merely algorithmic". And while the "algorithmic" part may be apt, I challenge the nonchalance with which we discuss our modern technology. The vast capabilities of computers running solely on arithmetic logic (nevermind alternative architectures being developed) is blurring the lines between quantitative and qualitative knowledge.

Specifically:

The computer I'm working on right now (or one quite similar to it) can translate books coherently between 100+ different languages. It understand and interpret spoken commands. It can beat Gary Kasparov at chess. It can recognize a song playing in the background behind a flurry of ambient noises merely by "hearing" it and instantly cue up the artist and album information. After getting the hang of my music tastes, it can reliably predict hundreds of other songs that I would like more accurately than my best friends. It knows my favorite restaurants, can tell me how long it will take me to get there, when the next subway is leaving, how to navigate through traffic, and suggest a few alternative restaurants that suit my palette if the commute looks too long. It can drive a car better than Jeff Gordon. It can fly a plane better than the captains of 777s with 20,000+ hours of cockpit time. It can make markets in products traded by a host of rational and irrational actors. It can carry on a conversation so convincingly that the average listener would be unable to distinguish between it and a human being. It can tell when I'm lying. It can diagnose cancer more accurately than the world's best oncologists. It can read and interpret encyclopedias, comprehend metaphors, puns and similes, and apply that knowledge to defeat the world's best two Jeopardy! contestants combined (not to mention a team of Congressmen). It can capture images that rival the resolution of the human eye. It can produce lifelike digital imagery and 3D simulations to bring imagination to reality on the silver screen. It can map the entire neural network of a cat's brain. It can locate any place on Earth (and elsewhere in the universe) and cue up street-level images of that location. It can read and interpret news. It can predict the weather. It can write books.

You call these "merely algorithmic". Personally, I think computers are already accomplishing tasks a hell of a lot more complicated than trading OTC derivatives. To each his own.

I understand that you believe that the progress of computer technology will conveniently stop shy of learning how to do your job. I suppose the next 20 years of computing in your version of the world, then, will be comprised mostly of slightly faster versions of the computers we have today (and maybe slightly "sicker" graphics in WoW). Programmers, mathematicians, scientists and computer architects will obviously steer clear of innovations related to the world's most expensive white collar workers in the world's most data-rich industry. You're welcome to believe that.

I do not.

Jan 18, 2013

you guys need to get a job already... hopefully not in finance.

Jan 18, 2013

Side note, there is a sci-fi film coming out in a few months that is about this very topic, namely that of computer intelligence surpassing the abilities of the human brain. It is called Transcendance (apt title I might add) and will be produced by Christopher Nolan. Needless to say I am very excited. God I love that guy.

Jan 18, 2013
Going Concern:

Side note, there is a sci-fi film coming out in a few months that is about this very topic, namely that of computer intelligence surpassing the abilities of the human brain. It is called Transcendance (apt title I might add) and will be produced by Christopher Nolan. Needless to say I am very excited. God I love that guy.

Agreed.

Jan 18, 2013

^ SB'ed, that's what I wanted to say hah.

Jan 18, 2013

only one solution guyz... protest against the machines! we want jobzzz

anyways yeah ai and computing are extremely fascinating subjects, it's very hard to tell how things will progress one way or the other in the next 20 years or even the next 10. but i think most of us will live long enough to see some great things.

Jan 18, 2013

You have way too much time on your hands to write these long-ass philosophical responses. This is how I know you are on a trade floor. Yahza,

Jan 18, 2013

Very interesting discussion on the future of computing.
Being an Economic major I don't know as much as I would like on the subject. Do you know some books that could help me get a better understanding on the future of computers ?

Many thanks,

Jan 18, 2013
TheSquale:

Very interesting discussion on the future of computing.
Being an Economic major I don't know as much as I would like on the subject. Do you know some books that could help me get a better understanding on the future of computers ?

Many thanks,

Although slightly dated at this point, the most oft-cited book on the matter is Kurzweil's The Singularity is Near, though his more recent, How to Create a Mind, might be equally applicable to the topic at hand. Kurzweil maintains a website with a lot of shorter material on the subject matter that is enormously popular: kurzweilai.com. Also, though not specifically on the topic of computing, I recommend Daniel Dennett's Consciousness Explained, which delves very deep into philosophy of mind. Wrapping your head around the most modern understanding of the operations of the human brain actually goes a long way in thinking about the future of technology. Dennett's bibliography includes most of the relevant literature on the matter like Searle, Fodor, Marr, etc.

    • 1
Jan 18, 2013

I think its safe to say this thread officially went off of the deep end

Jan 18, 2013
wallstreetballa:

I think its safe to say this thread officially went off of the deep end

Honestly, I don't think the OP could have asked for a more thorough discussion of his question. Today, regulation is the biggest issue in S&T; however, 10+ years down the road, the sophistication of technology is far more relevant.

Jan 18, 2013

i am not qualified to join the debate about the speed of technological advance generally but I can say with 100% certainty that if Northsider showed up on a fixed income trading desk he would be very very dissapointed about the level of technology. This isnt a free market business, its a protected monopoly and these type of changes happen very very slowly. In order to make this work and disintermediate the banks, especially with OTC products that create very fat margins, you will need to sidestep a multi-billion dollar lobbying effort to stop it. As a hedge fund client and not a sell-side salesman i benefit from tighter more automated markets but it took 20 years to even create something that allows me to trade US govt bonds without using the phone and it only is allowed because the dealers own the company and control the method of transaction in such a way that they are protected.

Jan 18, 2013
Bondarb:

i am not qualified to join the debate about the speed of technological advance generally but I can say with 100% certainty that if Northsider showed up on a fixed income trading desk he would be very very dissapointed about the level of technology. This isnt a free market business, its a protected monopoly and these type of changes happen very very slowly. In order to make this work and disintermediate the banks, especially with OTC products that create very fat margins, you will need to sidestep a multi-billion dollar lobbying effort to stop it. As a hedge fund client and not a sell-side salesman i benefit from tighter more automated markets but it took 20 years to even create something that allows me to trade US govt bonds without using the phone and it only is allowed because the dealers own the company and control the method of transaction in such a way that they are protected.

Agree with everything here. Having spent several days on fixed income and equities S&T desks, it's quite clear that nothing affects the industry more directly than regulatory policy. The government (unfortunately) holds the keys, and Congress / bureaucracy can artificially slow the process of machinizing trading to placate lobbyists. I like to pretend the government doesn't exist most days, so this entire discussion comes with the obvious caveat that the government may step in and prevent disintermediation to suit their own political purposes. That said, it will become more and more difficult to resist this kind of change as technology improves.

Jan 18, 2013

Totally agree with NorthSider, but Ray Kurzweil is an expert in self and mass delusions.

Jan 18, 2013
Macro Arbitrage:

Totally agree with NorthSider, but Ray Kurzweil is an expert in self and mass delusions.

I do think Kurzweil demonstrates an extreme viewpoint, and definitely don't agree with everything that he says/writes. However, for better or for worse, he's the most often-cited author on the topic, especially for those that disagree with him. Personally, I prefer Dennett's discussion of the status quo in philosophy of the mind. That gives a great background on the topic.

Jan 18, 2013

One relevant article in the WSJ about speed trading and machine readable news.
http://online.wsj.com/article/SB100014241278873246...

Jan 18, 2013

Just saw this BBG quote of the day and thought of this thread:
"Computers are useless. They can only give you answers."
- Pablo Picasso

    • 1
Jan 18, 2013
Martinghoul:

Just saw this BBG quote of the day and thought of this thread:

"Computers are useless. They can only give you answers."

- Pablo Picasso

I like that.

Jan 18, 2013

I read this thread and then read another thread about prestige of different groups and exit opps and wonder how such a site can present such a disparity in terms of the quality of content/maturity of posters.

Only WSO haha.

PS. I also realize that after reading NorthSider's posts that when it comes to constructive debate/reasoning I am most likely average. Either that or he's right about computer's having passed the Turing Test because I'm pretty sure IBM Watson is trolling us.

Jan 18, 2013

I am a computer science major and I have decent experience with programming. I am also working on a fixed income trading desk. Trading is like poker - there are no black and white answers, and there's many shades of grey. Computers are good with black and white but computers are really bad at dealing with grey. There's a reason why a computer can't beat a human at poker over the long run. Computers are only tools that do what they're told to do.

"The code of competence is the only system of morality that's on a gold standard." - Francisco d'Anconia

Jun 24, 2012

Yep, that's what we need, more CS grads thinking that by proving a system works via backtesting that they will be able to earn a fortune for their bank...

Jun 25, 2012
commods:

Yep, that's what we need, more CS grads thinking that by proving a system works via backtesting that they will be able to earn a fortune for their bank...

Jun 25, 2012
commods:

Yep, that's what we need, more CS grads thinking that by proving a system works via backtesting that they will be able to earn a fortune for their bank...

haha This.

Sure, being a CS wiz will only broaden your abilities, but lets not start thinking that the algo's are taking over. Spreads are definitely drying up, but the only guys getting jobs and the only names that you see on NI FINHIRE couldn't code "Hello World" in C++ or in VB for that matter. Its truly a relationship game. The HR rep and headhunter may get wet when you name every coding language under the sun, but the person who actually matters will only ask "Have you done this before and have you made money? Do I know you or someone else that can vouch for you?"

Jan 18, 2013

No, it isn't. Nothing that we see today will be gone before we're dead (MOSTLY).

However, it is increasingly important to catch up and keep up with technology and the changing ways that it interacts with and alters our roles. That is to say that the S&T you join today will not be the same S&T you or someone else will be in 20 years from now.

A more experienced monkey (who's actually in S&T) can likely describe this better than I, or provide a different opinion.

Jun 28, 2012
BillyRay05:
commods:

Yep, that's what we need, more CS grads thinking that by proving a system works via backtesting that they will be able to earn a fortune for their bank...

haha This.

Sure, being a CS wiz will only broaden your abilities, but lets not start thinking that the algo's are taking over. Spreads are definitely drying up, but the only guys getting jobs and the only names that you see on NI FINHIRE couldn't code "Hello World" in C++ or in VB for that matter. Its truly a relationship game. The HR rep and headhunter may get wet when you name every coding language under the sun, but the person who actually matters will only ask "Have you done this before and have you made money? Do I know you or someone else that can vouch for you?"

So what is the advice for an UG, have a connection inside the firm and some trading experience?
Thus focusing less on the coding and systems side.

DLJ Analyst Class '96

Jan 18, 2013

Yes. I hear constantly about sales traders (especially equity sales traders) getting fired and displaced by technology.

Jan 18, 2013

But from my standpoint, would it make sense to pursue this opportunity to get out of the "assistant" position and then play it by ear. Exit Ops for sales-traders are what? Just straight to sales or trading on the buy side, correct? Am I missing anything else here? Thanks guys.

Jan 18, 2013

Whatever, I mean Sales Trader is better than Trading Assistant, so unless you have better opportunities elsewhere, more money and a promotion generally = good.

Making the jump from cash equity sales trading to the buy-side is really tough. If you're doing like complex OTC derivatives, then it's doable because you have specialized knowledge that's immediately transferrable. One of my equity sales traders at a big bank (he's about 33, so maybe 7 years older than me) confided in me he was concerned about getting canned and asked me what he should do. I told him he should go work selling industrial drilling equipment or jets at a defense contractor.

Jan 18, 2013

Is this the same bank you are currently employed at? If you hare a sales assistant you should have pretty good idea of the role and opportunities from dealing with the salesmen. Is it straight cash equities?

Jan 18, 2013

sorry to be a noob, but what does a sales-trader actually do? I hear about the job, I wouldn't mind getting one.

Jan 18, 2013

http://www.efinancialnews.com/story/2012-08-22/wha...

"Despite the misleading title, the sales trading function does not typically involve trading. Sales traders intermediate between traders who sit at a screen and interact with the market and the firm's clients. They sell the firm's trading expertise, ideas, trading tools, internal liquidity and its risk capital."

Jan 18, 2013
Jan 18, 2013

The cash equities sales trader does appear to be moving towards a defunct role, just as 20 years ago they had chalkies updating stock prices on a big black board. Equity sales speak to the PMs, sales traders receive and/or work the orders n the market, however with the increasing sophistication of execution algos I can't help but think that eventually the sales trader will be talking to the least important person in the investment decision process (the dealers). Be wary.

Jan 18, 2013
ya boy:

The cash equities sales trader does appear to be moving towards a defunct role, just as 20 years ago they had chalkies updating stock prices on a big black board. Equity sales speak to the PMs, sales traders receive and/or work the orders n the market, however with the increasing sophistication of execution algos I can't help but think that eventually the sales trader will be talking to the least important person in the investment decision process (the dealers). Be wary.

^This.

I used to work on an algo execution desk in a BB and the sales traders there had very solid jobs that definitely weren't going anywhere. You also don't need to be that technical even though you're selling algos.

Jun 25, 2012

great post, I want to break into trading after my MBA, unfortunately I don't really have an option to take any CS classes during my program, as it is already compacted into a single year. Luckily for me I will probably look for a job in Moscow which is not as crazy competitive as NY/LDN but still I would imagine I would want to know some of the programming languages mentioned here.

Jan 18, 2013

yes it is. it was also dead 5 years ago.

Jan 18, 2013

why?

"Hold on a sec...you mean they made all this money without doing IB --> PE --> HBS --> PE --> God?
How is this possible?!?!?!!??" - TheKing

Jan 18, 2013

The machines have killed the people who were total muppets that did nothing but stand in between two people and charge both of them for being there despite having nearly 0 liquidity risk. Good riddance

Jan 18, 2013
Comment
Jan 18, 2013
Comment

"Hold on a sec...you mean they made all this money without doing IB --> PE --> HBS --> PE --> God?
How is this possible?!?!?!!??" - TheKing

Jun 25, 2012
Jun 25, 2012
Mar 27, 2018
Comment

If the glove don't fit, you must acquit!

Pages