Forums  > Careers  > quant career advice  
Page 1 of 1
Display using:  


Total Posts: 20
Joined: Feb 2018
Posted: 2018-03-10 10:42
I'm in my middle thirties: now researcher, experienced in programming, applied mathematician, stem master's degree. For the last several years occupied with the things close to what Quantopian and Websim do, managed to produce a set of portfolio construction techniques allowing to extract statistically significant abnormal returns, built the mechanism to mix output of those alpha generators and showed this delivers statistically significant alpha improvement relative to what equal weighting does, handled risk forecasting and portfolio optimization thus allowing to keep portfolio volatility under control resulting in the uniformly improved information ratio of the portfolio. So far silence from the US, European hedge funds industry leaders. Want to change that situation but with no feedback it's unclear what to do so I'm asking for a good advice here. Many thanks


Total Posts: 362
Joined: Jan 2015
Posted: 2018-03-10 16:24
Let's pretend that you invented a perpetual motion machine. How hard do you think it would be to get people to pay attention? Think of how many thousands of cranks are trying to sell perpetual motion devices every single year. If you're in establishment physics or engineering, it's just an endless of deluge of failed promises. If you unmute the noise, even a little, you'll waste countless amounts of time, effort and money.

When it comes to novel systematic trading strategies that deliver actual alpha in real trading, the ratio of real opportunities to quacks maybe isn't quite as small as perpetual motions. But it's close. Let's say that you actually have a gem. How do potential investors not know that you didn't just overfit the hell out of it? How can they be sure that any out-of-sample walk-forward results aren't just made up? How do they know that your backtest assumptions accurately reflect reality?

So, if you did have a legitimate perpetual motion machine, how would you go about getting it out into the world? The first step is to distinguish your device from the endless masses of charlatans. You need to signal, in the econ sense of the world. Meaning show something that would be hard for a charlatan to fake.

Option one: Some real-world demonstration of the perpetual motion machine. Designs on paper are easy. Building and running something is not. If you can start by powering a car or your house with the device, it doesn't prove it with 100% certainty. Charlatans could be employ some Penn & Teller trick. But at the very least it helps you rise above the noise. Maybe if you can get the attention of some people, then you can acquire the resources to run a bigger demonstration and so on.

In quant trading a demonstration is some track record involving real life money. The more money involved the more legit the demonstration. Question number one, have you put any real money into your strategy? Personal money, friends and family, business associates? Number one that shows you at least have personal confidence. And number two live track records are more costly to fake than backtests.

Option two: Acquire some sort of social standing in the physics community, such that people will pay attention to you. If a Nobel prize winner claims to invent a perpetual machine, people will pay a lot more attention than Joe Schmoe. If you really believe in it, that maybe means getting a physics PhD, working in a physics labs, publishing papers on more conventional topics, building up a citation rank, getting a professorship. All of those things confer personal legitimacy. When you've acquired more status in the community, then you can start making extraordinary claims without being dismissed out of hand. You might even have to let a more established physicist take the lion's share of the credit just to boost the idea's credibility.

What does this look like for quant trading? It would mean becoming involved in industry or academia in some way. A PhD in quant trading adjacent field would help. Getting a job at a major firm would help. They're not going to hire you to build perpetual motion machines, but they'll probably hire you to do backend programming. That still builds up your resume and network. After enough years and with enough reputation, people will pay more attention to your backtests. It might involve just handing over the strategy to a portfolio manager, without any guarantees for yourself. If you're a great employee, you may even convince your firm to start by putting a little bit of real-life money behind your ideas. If they do as well as you believe, they'll eventually clamor to put more.

Good questions outrank easy answers. -Paul Samuelson


Total Posts: 20
Joined: Feb 2018
Posted: 2018-03-10 18:36
Thanks ))) I'm not making claim about perpetual motion machine, just plain active equity e.g. Grinold Kahn, Qian-style and the others which are about abnormal return things. My interest is to land a job in the decent quantitative place. Right now it's a quantitative research position but due to the fact that hedge funds industry actually does not exists at my place of living I'm looking at the places/countries where this industry exists.


Total Posts: 1509
Joined: Jun 2004
Posted: 2018-03-10 20:29
To add to what EL said, some thoughts/questions:
* experienced PMs/traders will have a good enough sense if your alpha is real by simply having a vague idea about the strategy, so it’s possible that your strategy did not pass a smell test
* in most cases, people like to see ideas that are off the beaten path but do not fall into the perpetual engine space
* you will find that there is a huge disconnect between the types of alpha being discussed on Quantopian and the types of things done at hedge funds
* besides your insight into alpha, what unique knowledge do you bring to the table?

I don't interest myself in 'why?'. I think more often in terms of 'when?'...sometimes 'where?'. And always how much?'


Total Posts: 20
Joined: Feb 2018
Posted: 2018-03-10 21:46
*If only I could talk to experienced PMs/traders)))
*regarding the huge difference between Quantopian and the types of things done at hedge funds: that is actually very interesting which kind of things are employed))) though may be some funds are Quantopian like at least taking into account the things that are available in open publications: e.g. numerous articles about sources of alpha in equity + framework developed and published by already mentioned Grinold Kahn (Barra) strongly remind Quantopian/Websim, and of course we can look at the books by Zhou (Point72), Tulchinksy (Worldquant).
*I'm trying to get a quantitative researcher job and even not obligatory in equity space. It's just happened that way given popularity of the already mentioned projects seeking alpha in equity that I also started in that area and decided to cover the technology described in the mentioned books rigorously. And it happened to be a fascinating area for me. And it looks like my skills are very in line with the tasks I dealt so far. Really don't think my methods of generating alpha bring somethings new. But I guess that my professional skills could be applied to the tasks the funds are dealing with.


Total Posts: 1509
Joined: Jun 2004
Posted: 2018-03-11 02:21
Send me your CV if you feel like it, I’ll see if anyone cares at the place where I work (large US fund, but we do have offices in all regions).

I don't interest myself in 'why?'. I think more often in terms of 'when?'...sometimes 'where?'. And always how much?'


Total Posts: 20
Joined: Feb 2018
Posted: 2018-03-11 09:52
thank you, Strange


Total Posts: 237
Joined: Oct 2006
Posted: 2018-03-11 17:21
Feel free to send me your cv, I will try to help.

The dark is light enough.


Total Posts: 20
Joined: Feb 2018
Posted: 2018-03-11 18:31
thank you, signalseeker


Total Posts: 1056
Joined: Nov 2004
Posted: 2018-03-11 18:35
To me, that particular extract is a proof you are probably just like any other overfit out there and miss the basics that everyone in this business should at least master.

> statistically significant abnormal returns

May I suggest the reading of a very well written paper that should be basic reading for any graduate student... and too often for their statistics teacher too?

Attached File: Null_Hypothesis_Significance_Tests_A_Mix_Up_of_Two_Different_Theories_The_Basis_for_Widespread_Confusion_and_Numerous_Misinterpretations-Schneider.pdf

Best of luck, and remember: sometimes you have to eat your own cooking!

If you are not living on the edge you are taking up too much space.


Total Posts: 458
Joined: Jul 2008
Posted: 2018-03-11 21:44
Thanks for the paper! One learns something very useful every day.


Total Posts: 1017
Joined: May 2004
Posted: 2018-03-12 06:51
Hi Zoho,

I am happy to have a look at your CV and connect you if possible.

"Earth: some bacteria and basic life forms, no sign of intelligent life" (Message from a type III civilization probe sent to the solar system circa 2016)


Total Posts: 2
Joined: Mar 2018
Posted: 2018-03-12 09:31
Heyyy Zoho ,

Would love to connect with you , here's my email cryptoarnold1gmailcom
I've been researching on what quantopian and similars have as well . Let me know !


Total Posts: 20
Joined: Feb 2018
Posted: 2018-03-12 19:20
>To me, that particular extract is a proof you are probably just like any other goldorak out there and miss the basics that everyone in this business should at least master


Total Posts: 20
Joined: Feb 2018
Posted: 2018-03-12 20:30
Actually, I didn't make any statement that my results are free from overfitting. It's inevitable to some degree. I tried to follow systematic approach inspired by something like for example this:
there is not much space for possibility of overfitting
Previous Thread :: Next Thread 
Page 1 of 1