Career: Trader vs. Quant - Pros and Cons for Physics Graduates

[Onamor]

Hey, thanks for your replies, particularly this part;

Let me tell you what a trader does.

I (and I'm probably not alone in this) have read books that simply don't state this as clearly as yourself, so thank you for that.

But could I throw you another question?

Now that we have computers, can you program a computer to automatically look for people wanting to sell stock and then match them with people that want to buy stock. Sure.

Why do we have have human market makers at all? Surely they must inject some kind of (human) prejudice into the process? Not just buying at the lowest, selling at the highest - a computer can do that as well...

 

[twofish-quant]

Why do we have have human market makers at all? Surely they must inject some kind of (human) prejudice into the process?

If that human prejudice includes wisdom and intuition, that's a good thing. Suppose you have a $1M position on Exxon stock. It's now down to $950,000. Then it goes to $900,000, and $850,000. If it hits $800,000, and you haven't closed the position, the system will automatically close the position since you've exceeded your trading limits, and bad things will happen to you.

So what do you do? If you've been trading Exxon stock for years, you'll have some idea as to whether or not to take the loss.

It should be noted that there is a human bias not to take losses and keep to a losing position rather than admit defeat. That will get you killed as a trader which is why most people don't make good traders. What happens is that traders have a trading limit, and if you exceed those, it's like maxing out on your credit card. Bad things happen to you.

As far as how those limits are set, there's another group of people that deal with that.

Not just buying at the lowest, selling at the highest - a computer can do that as well...

Computers can't think. What they *can* do (and do in fact do) is process vast data at millisecond frequencies.

Sure if you have a bunch of computer programmers programming the system and a bunch of traders that know the markets well enough to input trading strategies, but at the end of the day the computer is just a tool, and you need people to tell it what to do.

For example, one thing that the traders do tell me is not to try buying at the lowest and selling at the highest. If you try to time the peak, you'll always miss, and lose your shirt trying.

 

[Onamor]

Sorry, I don't think we're on the same page here. I was asking about why we have market makers - the guy that sits at the computer to match buyers and sellers. He just provides liquidity, and makes profit from the spread, if he sells higher than he buys, or vice versa.

The guy I think you were referring to here;

If a trader has just bought 100 shares of Exxon stock, their job is to find someone that wants to buy 100 shares of Exxon stock as fast as possible so that they can make money off the difference. Traders do not buy and hold because the longer they hold on to stock, the higher the chances that "something bad" will happen.

As far as I know, in broad terms, there's proprietary traders and market makers. One trades when he thinks he knows a profit opportunity and the other provides liquidity.
Why can't a computer do the job of the market maker? Do they do something else?

Sorry if it seems I'm hammering on here, but sometimes you have to ask the simple questions... So thanks again for any replies.

 

[snowjoke]

There are different types of exchanges for different instruments. For very liquid instruments, you have an automated order book system, where buy and sell orders are placed (which can be at limit, where maximum/minimum price is specified, at best, where you fill the order based on what's already on the book) and automatically matched. For less frequently traded instruments markets tend to be quote-based, i.e. the market maket provides a bid and ask price at all times. It's his job to decide what these prices are. For even more exotic and illiquid instruments, they might not even be listed on an exchange, and selling it might involve phoning up clients and negotiating prices.

 

[twofish-quant]

As far as I know, in broad terms, there's proprietary traders and market makers.

There are also other permutations of this. For example, there are portfolio managers that look for profits, but for pension funds, insurance companies, mutual funds, etc. That's another set of players.

One trades when he thinks he knows a profit opportunity and the other provides liquidity.
Why can't a computer do the job of the market maker?

They often do, but then someone has to program the computer to recognize market signals. In some markets, computers do badly either because there aren't enough trades to justify the expense of a computer, or because there isn't a single market, or because computers just aren't as good as human beings for recognizing market signals. Other markets are now dominated by computer trading.

Just to give one example in which computers aren't that good. If you want to trade Exxon stock, then there is only one real type of Exxon stock. If you want to trade a contract on the price of oil, there are a huge number of ways you can structure that, and so to place the order you have to be on the phone for a while getting the exact details of the contract.

Finally, you need a human being to recognize when the machine is going haywire and to shut it off. Most markets have a rule that say that if the index moves more than X% in Y minutes, then the computers get shut off and people have to execute orders by hand so that people can figure out what happened.

The trend over time has been for trading to become more and more automated. At one end you have market makers that just put in an algorithm and let the computer run. At the other end, you have people sitting in from of computer screens, but even they use a lot of compute power so that they can display the data in ways that they can find what they are looking for. And you have various combinations of man and machine in between.

But curiously more automation means more humans are involved. Computers don't program themselves, so what often happens is that you have programmers create a trading platform that takes canned strategies from traders and executes them. Even in the area in which people are making the trading decisions, they rely very heavily on computers to display information in some way that they find comprehensible.

The other thing is that computers just break down sometimes, and when you have computers making a lot of the decisions, it's a very, very bad thing when the computer just stops working suddenly. At that point you bring in a team of tactical programmers to fix the system, and that sort of work is very interesting.

Also, in some markets pretty much all trades are automated, so that point it's a fight between who has the better computer programmers.

One analogy is that traders are like race car drivers, but to get anywhere as a race car driver, you have to have a good pit crew and great engineers to design your car.