Math nerds are taking over Wall Street


Recommended Posts

"They are not making trading decisions -- those are all made by computers."

 

When you think about the typical Wall Street trader, you probably picture a fast-talking Wolf of Wall Street type.

But guess what? They are rapidly being replaced by "quants" -- soft-spoken super nerds armed with high-tech software to help them beat the market.

 

Elie Galam is one of them.

 

Every day, the 30-year old runs 35,000 different trading strategies through software he designed to find a handful of trading ideas with a high statistical probability of making him money.

From an early age, Galam was obsessed with math. After high school, he studied at the ?cole Centrale Paris, a prestigious French engineering university. After that, came Harvard, where he enrolled in an applied mathematics doctoral program. Galam jokes that life at Harvard was like Matt Damon's character in the film Good Will Hunting.

In one class, he built a computer algorithm that successfully identified the writer of an article based on programmed characteristics such as style and voice. It was the kind of work that would lay the foundation for his career in finance.

Soon enough, Wall Street recruiters began knocking on Galam's door. Money, prestige and the chance to work on cutting edge quantitative finance systems all appealed to him. So he cut his Ph.D program short after one year and settled for a master's degree instead. At the age of 22, he accepted a job at Blue Mountain Capital, a credit trading hedge fund in New York.

After a few years, his entrepreneurial spirit kicked in, and Galam went into business with James Greenberg, a veteran Wall Street dealmaker. The duo went on to launch Panorama Partners in late 2011.

At the core of their strategy: a quantitative software program built from scratch by Galam that uses historical data and analysis to predict price movements in various assets.

"We get as much data as we can, we shock it, test it, do back tests, historical analysis," Galam explains. "That's where I come in, where the science comes in."

 But Greenberg insists that Galam isn't your average Wall Street quant, and that he possesses the rare combination of computer skills along with raw trading instinct.

"He has a rigid approach to math but he also has creativity," Greenberg says.

Potential hires don't need to know much about finance, but they should be top notch when it comes to applied mathematics. Complex brain teasers are standard interview questions.

He keeps in close contact with his old math professors in Paris, and calls them frequently to ask about promising talent in the classroom.


more & video

Link to comment
Share on other sites

All these share volatality can not be solved by some freaking software algorithm. It is same as predicting the lottery ticket number by looking at historical winning numbers.

Link to comment
Share on other sites

All these share volatality can not be solved by some freaking software algorithm. It is same as predicting the lottery ticket number by looking at historical winning numbers.

 

Exactly, Its about predicting human behavior more than anything else. 

 

Nothing is going to stop a massive sell off when the zombie apocalypse comes. 

Link to comment
Share on other sites

I think this is old news. It's certainly been the case here in the UK that Financials Services industry is starving other industries of mathematicians for many years and I'd be surprised if the same thing hadn't been happening in Wall Street over the same period.

Link to comment
Share on other sites

All these share volatality can not be solved by some freaking software algorithm. It is same as predicting the lottery ticket number by looking at historical winning numbers.

Not so - lottery is random - trading has patterns.  buying when things are good, selling when things might go bad or are going bad - the players change but the pattern hasnt changed too much.

  • Like 2
Link to comment
Share on other sites

It is same as predicting the lottery ticket number by looking at historical winning numbers.

Funny -- I have won using my own programs. :happy:

Link to comment
Share on other sites

And I have software that predicts horse races... in the end, it's all luck

of which I apparently have none -- my method of "pick the horse with the name that sounds like it would win" doesnt work for crap !

Link to comment
Share on other sites

of which I apparently have none -- my method of "pick the horse with the name that sounds like it would win" doesnt work for crap !

 

Always bet on "Shes-Commin-up-the-rear" never fails

Link to comment
Share on other sites

as a math/finance major i've seen a glimpse of some of these models in action and it is pretty bloody amazing as to what they can accomplish.

 

it's true that you'll have events that cannot be foreseen but if you're able to model their probability of occurrence and consequential impact then assess the risk accordingly you can still come out ahead.

 

it's also true that this is about understanding human behaviour but the foundation of technical analysis *is*, in fact, behavioural finance.  when you look at past events and the market movements you basically look at 'how did people react in these circumstances?'  you try to compare similar events and the market's reaction to that, you weed out as much noise as possible...it's not perfect but you assign probability (based on data/judgement call) then make your bets accordingly.

Link to comment
Share on other sites

as a math/finance major i've seen a glimpse of some of these models in action and it is pretty bloody amazing as to what they can accomplish.

 

it's true that you'll have events that cannot be foreseen but if you're able to model their probability of occurrence and consequential impact then assess the risk accordingly you can still come out ahead.

 

it's also true that this is about understanding human behaviour but the foundation of technical analysis *is*, in fact, behavioural finance.  when you look at past events and the market movements you basically look at 'how did people react in these circumstances?'  you try to compare similar events and the market's reaction to that, you weed out as much noise as possible...it's not perfect but you assign probability (based on data/judgement call) then make your bets accordingly.

You can say what you want but market, people, always behaves unexpectedly and most of the time irrationaly which is the main reason there is no significant algorithm to predict it. The only sure shot way to make money in share market is insider trading. By the time the software predict some abnormality for the gain, the market will correct itself quickly which will negate that gain.

Link to comment
Share on other sites

You can say what you want but market, people, always behaves unexpectedly and most of the time irrationaly which is the main reason there is no significant algorithm to predict it. The only sure shot way to make money in share market is insider trading. By the time the software predict some abnormality for the gain, the market will correct itself quickly which will negate that gain.

 

you're missing the point.  the objective is not to predict the irrationality, the idea is to:

- minimize the losses once something irrational occurs (by investing fewer amount of money) or selling for a little loss

- capitalizing on the market extremes (buy when oversold, sell when overbought)

 

as an example - say i have $10K in a portfolio...$5K maybe invested in some stock...some irrational event takes place and the stock drops.  depending upon the volume/% drop (and other technical conditions) the algorithm could determine that this isn't a regular blip and sell automatically, but even if it doesn't, and the investment goes down to $2K...then the algorithm could look at the technical conditions at that time and buy more stock and average down with remaining $5K...once it goes back up, you've made your old losses + more money.

 

this is a very basic example with only stocks.  when you throw stock options into the mix you can do a ton of things to make money/minimize your losses regardless of the market direction, even if you don't time it correctly every single time.

 

timing the market helps, no doubt, but it's not the only way to make money in it.  algorithms just help speed up the transactions what a person typically cannot.

and of course, judgement calls are involved.

Link to comment
Share on other sites

Hollywood is going to make a movie of this soon hopefully :)

Link to comment
Share on other sites

This topic is now closed to further replies.