mardi 16 août 2016

Goldfinger (and The man with the golden gun)

If you want to quit particle physics, what after : Finance or Data Science?
Here are a few hints grasped into the comment section about the post "After the hangover" by the blogger with a golden gun to kill science news overhyped


From Flakmeister (Semi-retired, 20 years experience as a professional Higgs Boson Hunter and other beasts at CERN, BNL, SLAC and FNAL.... Worked on Wall St. 2005-08 modelling CDOs so I had a front row seat for what was coming)
Well, call me gobsmacked, a fundamental scalar exists at mass which is clearly suggesting to us that the next new physics threshold may be forever beyond our reach... 
At Snowmass 2001, you would have garnered strange looks if you asked what if SM and nothing else at the LHC. Looking up the chimney and seeing nothing but blue sky so to speak. It would seem the famous "No Lose Theorem" of a TeV scale collider paid off with the lowest possible jackpot...
20 March 2013 at 20:24



I think the field is big trouble as the feared outcome of a SM Higgs and a desert appears to be materializing. I remember discussions with Marcela and Howie H. (and other theorists) about the Higgs being all there is "all the way up the chimney" and noticing that it was anathema to them. We are victims of our own success. 
I left {high energy physics} HEP (some might say was pushed out) in 2005. I am now in charge of Quantitative Analytics for the US operations of a top 10 bank by assets. There is a dearth of people that really understand how to organize and process data in the financial world and there are opportunities out there for those willing to make the shift.
This new career is joke compared to what I used to do, however, HEP was never going to pay me $300,000 a year for working 20 hrs a week... 
By the way, I co-authored a pretty famous paper on Higgs searches that most here would have known about. It was quite a result at the time and was bourne out by data.
8 August 2016 at 17:29

An anonymous asks...
Hi Flakmeister, ... what do you mean by "This new career is joke compared to what I used to do". Can you explain? This might be very useful for many particle physicists who are looking to make the switch from physics to Wall Street/ data analysis. 
8 August 2016 at 23:10
Flakmeister answers
By joke, I refer to "what qualifies as a quality analysis" and what it takes to satisfy management. The 60 hour work week is now a distant memory except for maybe 2 or 3 weeks a year. The challenge now is explain basic statistical techniques and results to people without any quantitative background. The really advanced actuarial techniques for OpRisk such as LDA are now frowned upon by regulators. (a reflection of the dearth of skill to internally and externally review the analysis.) 
The main difference is that the politics in a large financial institution is brutal because merit plays a small role in who decides what. There are no shortage of characters who are clueless bullies trying to climb the ladder or maintain their little fiefdom, sometimes at your expense.

I think the golden age of shifting to the financial sector from HEP is over, at least for the theorists, that being said, people with good quantitative skills are always in demand. Don't underestimate the value of understanding the basics of data presentation in the real world.

If anyone is thinking of making the change, start by immersing yourself in learning fixed income and the associated concepts. Any Hep-Ex Ph.D should be able to learn the basics in 2 weeks. The data is basically simple time-series derived n-tuples (relational database entries). One simple test if you know what you are doing is to use Excel to compute your mortgage payment (interest and principle) from first principles. Also learn how to bullshit your way with SQL, R, SAS, Excel for the interview. You can figure it out on the fly as needed.
9 August 2016 at 18:00

... there is serious politics in HEP-Ex, but at the end of the day a few egos get bruised and someone whines a bit that their version of an analysis was not the headliner or they did not get a certain piece of a hardware project. It's hardball but no one gets carried out on a stretcher. 
In the corporate world, it's downright gladitorial. In HEP, a Ph.D. is the entrance fee, which for better or worse does imply a certain level of scholarship and academic acumen; in the corporate world, MBAs are a dime-a-dozen and thuggery more than accomplishment is your ticket to the C-suite.. 
People get shit-canned for the mere reason that a new hire two levels up has been brought in and feels they needs to make "changes". In a meeting, if you don't know who is to be scapegoated for some failure, it is likely that it is you in the crosshairs. 
Another difference is that compensation is not always commensurate with skill or responsibility. Overpaid mediocre HEP'sters don't exist. There are no shortage of clowns in Banks whose primary ability is to deflect blame and be a sycophant. 
10 August 2016 at 17:34


Real politics is not about junior people. If pissing off a PhysCo means you didn't get a tenure track position interview, you were a marginal candidate to begin with. I know plenty of people that pissed off lots of people that now have tenure in no small part because they had real merit.

There is huge difference being "chucked to the curb" at 30 after a post-doc vs being 50 and getting dumped in a re-organization. The former is likely to quickly find a job that will likely double their income, the latter is likely toast and will have to take a 30% paycut at a new institution.

I do agree that the corporate-like structures that HEP experiments have morphed into is similar to the financial world. One big difference is that an LHC-type experiment is really a republic consisting of institutions whereas a Bank is an autocratic bureaucracy. For example, you don't work for ATLAS, you work on ATLAS.

Bank stocks were an incredible investment at one time as they can be insanely profitable. Going forward there are serious headwinds from collapsing interest rates. Banks live off of the spread, and those spreads have all but collapsed.
11 August 2016 at 16:53

From Disputationist (another commentator)
How to get a Data Science job as an HEP PhD 
... an HEP PhD alone is not enough to get you a data-science job, but you just need a few weeks/months of additional preparation. Roughly 50% of data scientists have PhDs - it's something the industry highly prefers for some reason, although its not a sufficient condition. 
This is what you need to study and put on your resume:
  • Machine Learning - Read a couple of textbooks. Start with "Learning from Data" by Abu-Mostafa, a very concise and easy read. Next read Elements of Statistical Learning by Hastie for a deeper look into many algorithms. Concurrently, do a couple of projects to demonstrate your ML skills - Kaggle has straightforward problems to work on. You'll need to read some blog posts/tutorials to get a decent ranking. Anywhere within top 10-20% will be impressive and give you a lot to talk about during interviews.
  • Basic computer science - Learn python if you don't already. Read up on the common algorithms and data structures - sorting, search, trees, linked lists etc. Practice some coding problems and hackerrank and read a book like "Crack the coding interview"
  • Stats/probability - Read a couple of intro stats/probability books and work out some of the problems
The first several interviews will be a learning experience, but if you put some effort into the above, you'll get a DS job after a few months that will pay 100-130K at first, and if you change companies every year or so you can very quickly get to 200K. And the work can be a lot more interesting and meaningful than finance. The first company may be lame, but after that you will be in high demand and you can pick a company/product/domain that you think is meaningful and interesting. Feel free to ask more questions 
11 August 2016 at 17:44

Aucun commentaire:

Enregistrer un commentaire

Cher-ère lecteur-trice, le blogueur espère que ce billet vous a sinon interessé-e du moins interpellé-e donc, si le coeur vous en dit, osez partager avec les autres internautes comme moi vos commentaires éclairés !
Dear reader, the blogger hopes you have been interested by his post or have noticed something (ir)relevant, then if you are in the mood, do not hesitate to share with other internauts like me your enlightened opinion !