Forums  > General  > Serious: What does a quant trader/statistician/mathematician exactly do while trading?  
     
Page 1 of 1
Display using:  

behumble.


Total Posts: 3
Joined: Aug 2019
 
Posted: 2019-08-23 09:54
Hey Algoraders!

The question pretty much sums it all. I am a semi-professional futures trader (CL, ZT, ZS, some crypto), who uses basic technical indicators since two years and has been fairly profitable.

That said, I don't understand the rush towards all things maths and statistics as a basic requirement to trade - if hedge fund and prop trading positions are considered.

So here's my humble question - where exactly does complex math, statistical model making, and computer programming come into the picture? Software dev positions are explanatory, but where do the aforementioned help? Doesn't the first rule of markets say you can't predict prices and movements?

Or is everyone rushing towards complex solutions which could have been solved fairly easily?

Worse still - are talks of ML, RL, AI, algo, etc a "gimmick" to attract investor capital?

All thoughts would be very, very helpful.

PS - Don't make this a discussion about TA vs Quant. I do use indicators to trade, but definitely, don't draw discretionary trend lines or watch "heads and shoulders" or analyze moon phases to place trades. In fact, I do think TA is complete bull, but using indicators has its logical reasons.

ronin


Total Posts: 479
Joined: May 2006
 
Posted: 2019-08-23 13:03
Gawd, where does one even begin?

There are many different things that fall under the quant trading umbrella.

Do you know that scene in Wall Street when Gordon Gekko introduces a guy as "the best trader in the world"? GG decides what he wants, how much and at what price, and that guy goes off to execute it? Yeah, that was the 80s. We have robots these days. It's not the sort of thing that bothers you if you are trading a few k from your spare bedroom, but if you are shifting institutional quantities, you worry about impact, slippage, drift and a million other things. So we have algos to manage that. We could in theory have people to do it, but it would be lots and lots of people, and they probably wouldn't come close to machines in terms of quality of execution.

On the other end of the spectrum is hft where you are making less than quarter of a cent per share, and not even that if you are a milisecond late. That sort of thing you couldn't do by hand no matter what.

And there is loads in between.

Our example, our universe is 2,500 shares, we trade half a dozen long-short and market neutral strategies, and we turn over roughy a quarter of our gross exposure every day. How many people do we employ to do that? Well, there is three of us - two traders and an ops guy. By hand, we would probably need a few hundred just to monitor 2,500 stocks.

ML and AI are indeed marketing gimmicks when it comes to trading, and they largely ran themselves out. I know there are a few people who are doing well out of them, but there are always a few people who do well out of anything.

The bottom line is that, if all you do is click on a handful of contracts for a few k here and there, none of that applies to you in any way. But at institutional levels, there is literally nothing you can do without some quant ability, even if it is provided by your brokers. And regulators would be breating down your neck from day 1 if you didn't have it, and you probably couldn't satisfy them that you can be trusted with other people's money.

"There is a SIX am?" -- Arthur

behumble.


Total Posts: 3
Joined: Aug 2019
 
Posted: 2019-08-23 14:12
Thank you for the insight.

Yeah - it's probably cause I swing trade a low six-figures account focused on five different futures that high-level math or advanced stats is something I don't really utilize. My risk is hardly ever above 5-6% (less than ten contracts for the chosen instruments), and I don't trade stocks. Commodities/interest rates trend fairly well and seldom generate those sudden news/sentiment-driven price spikes that could eat me up.

I do use basic math (like percentages, addition/subtraction, and multiplication) to calculate risk, allot capital, and mentally calculate overall exposure if I am aggressively trading for the day and it seems to do the trick pretty well.

Most trading job listings I see, or articles I read, serve up the impression that Ph.D. level math is a prerequisite to trade profitably - and then there's me armed with an economics degree with a limited knowlegde of calculus and zero of programming who's making respectable profits. It makes me think I am either lucky since two years or greatly missing an essential part of trading that could, perhaps, drive my returns further.

This is not to say I call trading easy or am throwing shade on quant methods. Have shed a lot of tears before settling on something working since the better part of two years.

ronin


Total Posts: 479
Joined: May 2006
 
Posted: 2019-08-23 15:12
> Most trading job listings I see, or articles I read, serve up the impression that Ph.D. level math is a prerequisite to trade profitably

Yes, that is actualy a gripe I do have quite a bit of understanding for. It has gone downhill quite a lot in the last ten years or so. Presumably hiring by committee and applicant tracking systems.

So you end up looking for maths PhDs with 20+ years experience running Sharpe 2 strategies for the position of a trainee settlement clerk. Then you complain about how there aren't enough people qualified to be trainee settlement clerks.

"There is a SIX am?" -- Arthur

Maggette


Total Posts: 1151
Joined: Jun 2007
 
Posted: 2019-08-23 15:26
In addition to the things ronin said: I don't think that the people on the more quantitative side of things think that you need quantitative models and a server farm to be profitable.

For example many global macro funds are to my knowledge more or less semi quantitative. Sure, they might look at some or even lots of data while generating trading ideas ....but more like an Bussiness Intelligence person would do. But, like ronins said, I guess when it comes to portfolio management, risk management and execution of these trading idea, the better shops have to get more quantitative.

Same holds true for many fundamental driven stuff in equity and of course private equity or VC stuff.

Regarding your profitability: I don't want to doubt your "respectable profits". I heard of a couple of semi professional retail swing traders. But if two sucessfull years are enough to sort "skill from luck" depends on a lot of parameters...

You have to understand, if you are really profitable, you seem to do something different! Lots of retail traders enter the market and use "swing trading" and indicators. There is a huge!!!! industry to give people like you access to markets. Some retail traders claim they do 15% a year over ten years. But the vast majority of them loose money. So it dosen't seem to be that straight forward.

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

sharpe_machine


Total Posts: 25
Joined: Feb 2018
 
Posted: 2019-08-23 15:39
> Most trading job listings I see, or articles I read, serve up the impression that Ph.D. level math is a prerequisite to trade profitably

There is definitely a cohort of quants who really need differential equations and other really sophisticated math.
But also, there are quants (and very successful) who apply "just linear regression" to carefully designed features and with careful regularization. Obviously, there are many people who understand linear regression and do not have a PhD.

But why we want to hire PhDs? I would say, that PhD has a proven track record of achieving meaningful results in a very competitive and uncertain field. Therefore, you do not need a title "PhD" to be on a PhD level of research&development skills. For example, if you have spent 5 years working alongside G. Hinton, but do not have a "PhD diploma paper", I would still consider you a PhD-level applicant (unless your job was to serve him a coffee every morning).

(applies mostly to pure quant business)

Jurassic


Total Posts: 255
Joined: Mar 2018
 
Posted: 2019-08-23 15:54
"ML and AI are indeed marketing gimmicks when it comes to trading"

@ronin do you honestly think renaissance just uses linear regression?

behumble.


Total Posts: 3
Joined: Aug 2019
 
Posted: 2019-08-23 16:10
@Maggette

Here's an outline about how I trade.

As an economics graduate, I understand the most widely used macro-economic models and can construct common models used by mutual funds or I-banks - goes to show I am comfortable in some level of math. In a way, my education and bachelors thesis has kind of helped in shaping how I trade or the markets I choose.

For example - while trading the soybean oil futures (ZL) market, I look at any directly affecting news variables or major import/export law changes or drastic macro climate changes that would impact long-term ZL prices. I then form a hypothesis of which side (long/short) to take more exposure in, check basic moving averages, check soybean futures (they kinda lead oil prices indicator-wise) and depending on my screen availability, either trade aggressively or place a couple of contracts and walk away. Each contract costs me $490, each hundredth of a tick movement is worth $6, so a brief move nets me $300-$500 if I trade 6-7 contracts. Some days the market goes crazy and it equates to over a thousand dollars in profits, and it rarely does crazy reversals (like in stocks or options) that I end up taking on a huge loss.

For futures like NQ that are absolute beasts in terms of volatility and movement, a few contracts can easily net a few grand if the market goes ballistic (trump tweets, trade war updates, etc). My risk is protected depending on the number of contracts and ongoing volatility in the shorter time frames - latter helps to calculate an appropriate stop loss level.

I guess my point is, this basic idea of how I trade gives me "respectable" profits - leaving me pondering for the need for a math or physics Ph.D. to trade markets profitably or the so called "prerequisite" for such an education to be successful.

However, generally knowing how complex math works in the markets would help me appreciate the quant trading sector, and leaves a good impression of how I must not get comfortable with my strategy and look towards including more advanced concepts to ensure I don't end up crying my way to the morgue.

ronin


Total Posts: 479
Joined: May 2006
 
Posted: 2019-08-27 08:06
@behumble,

I'm not sure how much feedback you want on this. At this stage, if you do get an interview for a trading role, I dont think this approach will get you very far.

Scalping indices is the most basic trade there is. And it doesn't have the risk profile that you describe. If you had been doing what you describe through 2018, your best case scenario would be double digit loss and your wife telling you to drop trading and get a real job. The most likely scenario would be blowing through your account, owing money to the brokers, and the wife moving in with her parents.

As for this "soybean crack spread" or what ever it is called, it is very niche. I have never heard of anybody doing it outside of specialist softs houses. You will have an uphill struggle convincing somebody there is money to be made there, and even more so that you are the right person to do it. Your description mixes three different strategies that operate on three different price and time scales, which is where most people would lose interest.

Anyway, don't let this discourage you. Rome wasn't built in a day. But it does look like you still have a lot to learn.

"There is a SIX am?" -- Arthur
Previous Thread :: Next Thread 
Page 1 of 1