Deutsche Bank Qp1 Internships

  • Uploaded by: zubezub
  • 0
  • 0
  • June 2020
  • PDF

This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA


Overview

Download & View Deutsche Bank Qp1 Internships as PDF for free.

More details

  • Words: 1,554
  • Pages: 20
Quantitative Products Alexander Gerko NES, Thursday, February 22th , 2007

1

Deutsche Bank Global Markets Equity

Equity Quantitative Products    

Quantitative Products Analytics (equity derivatives quants) Quantitative Products Engineering (pricing) Quantitative Products Laboratory (research lab in Berlin) Quantitative Products One (equity quants)

2

Deutsche Bank Global Markets Equity

The team  





Head of all Quantitative Products: Dr.2 Marcus Overhaus (differential geometry + theoretical physics) „Analytics“ quant team consists currently of 10 members, to grow to 13 by end of March –

12 in London



1 senior US quant



Headed by Hans Buehler, PhD in Finance

Research institute „QP Laboratory“ in cooperation with two Berlin universities (three professors with MS, PhD and PostDoc students) –

Research on fundamental questions



PhD projects together with the quant teams

„Engineering“ team is 36 people

3

Deutsche Bank Global Markets Equity

The team, continued 

 

„One“ (=delta one) quant team consists currently of 6 members –

4 in London (3 of them NES graduates)



2 in US



Headed by Andy Ferraris, PhD in Physics

10 quant developers across regions 3 more are joining in 2 months

4

Deutsche Bank Global Markets Equity

What exactly is QP One doing? 

Algorithmic Trading – automated execution of single stock and portfolio transactions –



New and fast-paced field ◆

Driven by shrinking margins in cash and program trading (commissions are in single digits bps, 10mln euro trade can be only 5K euro in commissions)



Expensive traders are being substituted by quants + computers



Pioneered in the late 90s by ITG and CSFB, picked up by all major players



Currently majority of flow in large caps goes through automated execution

No fundamental starting points like “Black-Scholes”, yet ◆

Few research papers (notably, a series of papers by R. Almgren et al. (market impact and optimal execution); Obizhaeva-Wang (market impact and optimal execution); Lo-MacKinlay-Zhang (time-to-fill modeling); Nevmyvaka (optimal order placement))



Industry generally well ahead of academic research

5

Deutsche Bank Global Markets Equity

What exactly is QP One doing?, continued 

Primary research topics –

Transaction costs analysis, pre- and post- trade analysis



Optimal execution of equity transactions









Macro level: equity basket pricing and optimal liquidation



Intermediate level: “trajectory follower”



Micro level: optimal single order placement

Modeling of intraday dynamics of literally everything (volumes, spreads, volatilities, durations, correlations, etc)!

Data specifics –

Terabytes of market and proprietary high frequency (transaction level) data



Need specialized DBMS to handle such volumes

Development environment –

Perl + Matlab for research



C++ for production

6

Deutsche Bank Global Markets Equity

Transaction cost modeling  

 

Some banks have heuristic “models” provided by traders R. Almgren et al. estimated transaction cost model on Citigroup data –

Separated permanent (due to information revelation) and temporary (due to liquidity demand) components



Made some statistical mistakes along the way…



…yet became the industry standard... sort of

We can do better than that (after Econometrics 3, 4)! We used the model with exponential decay of temporary impact –



independently appeared in Obizhaeva-Wang, and justified theoretically

Also include spread effect

7

Deutsche Bank Global Markets Equity

Algorithmic Trading, macro level Providing best execution of a portfolio transaction with respect to a certain benchmark, minimizing risk and transaction costs  Single stock case pioneered by R. Almgren, yet little practical value 



Optimal execution horizon does not depend on the order size, go tell that to a trader



Selection of a proper optimization target is a big theoretical and practical issue, mean-variance optimization produces dynamically inconsistent trajectories  Very computationally intensive (sizes of involved [dense, quadratic] optimization problems are up to 50000 variables) 

Exogenously (by ) constrained in terms of computation time  Lots of other constraints (trading speed, factor exposures, self financing, etc)  Yet possible to do, although only approximate solutions are available 



6 seconds for 500 stocks



60 seconds for 2000 stocks

8

Deutsche Bank Global Markets Equity

Algorithmic Trading, “intermediate” level 



First generation single stock strategies –

VWAP [value-weighted average price as a benchmark]



PctVol [constant “participation” rate as a % of market volume]

Can be improved indefinitely, but marginal benefit is decreasing –



VWAP - 2bps is a market standard

Most general form: “Follow a trading schedule defined in terms of physical and business time, taking into account market conditions, limits, volume caps, etc”

9

Deutsche Bank Global Markets Equity

Algorithmic Trading, micro level: Execution of a single small order through limit order book

Rule-based systems (non-portable, hard to support/modify and prone to errors) vs.  Optimal order placement 



Simulation - based estimation



Stochastic dynamic programming



Stock by stock



Must be robust to misspecification of micro-structural parameters



Computationally intensive...

10

Deutsche Bank Global Markets Equity

Internships programs 



2-3 for QP One –

Strong math, including optimal control



Very strong econometrics High frequency time series analysis



Survival analysis



Nonparametric methods



Strong Matlab programming skills, Perl is a plus



One of the positions is in Japan, so should have strong preference to continue full-time there (please indicate in the cover letter)

1 for QP Analytics –







Strong math, in particular ◆

Stochastic calculus



PDE



Monte-Carlo methods

Strong C++

2 for QP Engineering –

Good math



Understanding of derivative products and their pricing



Strong VB/Excel

11

Deutsche Bank Global Markets Equity

Quantitative Products One – Internship projects    

Modeling of intraday dynamics of various characteristics of the order flow Time-to-fill modeling Optimal portfolio execution using conic optimization techniques Adapting current models to new markets (HK, Tokyo, Australia)

12

Deutsche Bank Global Markets Equity

Quantitative Products Analytics – Internship projects 

Project suggestion 1: Practical implementation of hedging algorithms in the presence of jumps –

Problem: In the presence of jumps, plain delta-hedging is neither theoretically not practically justified.



If traded options are available, how can the jump risk to our portfolio be mitigated.



Broad theory on the subject but few (published) workable results.

13

Deutsche Bank Global Markets Equity

Quantitative Products Analytics – Internship projects 

Project suggestion 2: Hedging of barrier risk if the market impact function is known. –

Problem: When hedging barriers, we are subject to substantial „gap“ risk when unwinding our delta-position.



How can the knowledge of a particular market impact function (cost of trading a certain quantity of stock) be used efficiently to incorporate gap risk into the pricing of barriers.



FD/MC solution.

14

Deutsche Bank Global Markets Equity

Quantitative Products Engineering 

Deals, rather than projects.

15

Strictly Private and Confidential

Click and Insert the Date

Equitech: DB Equity Proprietary Trading

16

Deutsche Bank Global Markets Equity

About Equitech  Equitech

is in its fourteenth year of operation, managing proprietary capital on behalf of Deutsche Bank.  The group has a consistent history of profitability  The culture is very tech friendly and analytically oriented. Members of our group were formerly researchers or faculty in theoretical computer science or physics at places like Bell Labs, MIT, Stanford, NYU and Berkeley.  The team is based in New York, London and Tokyo.  Currently has 28 employees  Currently managing a global portfolio of several billion dollars.  Equitech employs a number of strategies, ranging in products from its core non black-black box based trading strategy that adapts to market conditions, to fully automated strategies, including ultra high-frequency trading in equities and listed options.

17

Deutsche Bank Global Markets Equity

Typical quant work conducted at Equitech  Trade

execution



Our large daily trading volume introduces large transaction costs inherent to most of our strategies



We are constantly working to improve the algorithms we use to execute portfolio trades - even a single basis point improvement in the execution algorithm translates into millions of dollars of additional pnl.

 Development

work on automated trading systems



Continually improve current strategies



Systematically search for new ideas that can be automated into trading systems to take advantage of market inefficiencies.

 Portfolio –

return attribution

De-composing portfolio returns into factors, this allows us to quantify our exposure to market factors (various types of beta) and better isolate the alpha our strategies generate.

18

Deutsche Bank Global Markets Equity

Opportunities  We

are looking to hire several quant analysts for our London office. In addition, opportunities exist for summer internships.

 What

we look for:



Evidence of academic excellence in a highly mathematical degree



Strong programming ability.



Ability and desire to work independently as well as part of a team



Knowledge of finance/mathematical finance/statistics is of course desirable

How to apply: Send CV to [email protected]

19

Deutsche Bank Global Markets Equity

Why Finance industry is better than Finance academia      

Real problems, not artificial ones You can see which models work and which do not You can talk directly to informed agents from the papers on market microstructure You can still publish, if you want to (true for DB, not true for most other banks) You can still do teaching, if you want to (internal seminars for traders/graduates) £,¥

20

Related Documents

Deutsche Bank
May 2020 25
Heil, Deutsche Bank!
May 2020 15
Deutsche Bank Conference
October 2019 22
Qp1.pdf
December 2019 3

More Documents from "Mohit Kvs"