Transcript A database of foreign exchange deals.

A database of foreign
exchange deals.
Philip Clarkson
02 April 2016
Introduction
2
02 April 2016
Outline
3
•
The Problem
•
The Data
•
A High Dimensional Problem
02 April 2016
The Problem
4
02 April 2016
Features of FX markets
•
Over the counter (OTC)
•
No central market place
•
5
•
Multiple liquidity sources
•
Aggregators and ECNs
24 hour market place
02 April 2016
Market Making in Foreign Exchange Markets
•
The role of the market maker
•
Trade channels
•
6
•
Telephone
•
Electronic
Issues
02 April 2016
•
Risk Management
•
Hedging Strategies
Customers
•
Who?
•
•
Hedge funds
•
Asset managers
•
Corporate clients
•
Insurance companies
Why?
•
Hedging
•
Speculating
•
7
02 April 2016
Different time frames
Market structure
Reuters/EBS
Corporates
Multi Bank
Portals
(FXAll)
Market
making
banks
Retail
Single Bank
Portals
Client
banks
8
02 April 2016
Prime
Brokerage
Asset
Managers
Hedge funds
Database
•
We have a record of customer order flow.
•
How can we use this data?
•
9
02 April 2016
•
Customer profiling = better service to customers
•
Better risk management
What are the questions that we want to answer?
Questions
•
•
10
Can we cluster or classify customers?
•
React quickly or slowly
•
Believe in mean-reversion or trends
Can we predict future currency movements?
02 April 2016
Only a partial picture
We will only see a fraction of each
customers trades.
11
02 April 2016
The data
12
02 April 2016
Data captured
13
•
Currencies
•
Quantities
•
Time stamp (accuracy depends on trade type)
•
Counterparty
•
Customer Account
•
Location
•
Etc.
02 April 2016
Data privacy
Data protection vs. data utility
14
02 April 2016
Data privacy
•
•
•
15
02 April 2016
Entities and location strings replaced by integer codes
Trade quantities rounded to one significant figure
Non disclosure agreement
The data provided
16
•
Currencies
•
Quantity of currency 1
•
Time stamp
•
Market counterparty code
•
Counterparty location code
•
Customer entity code
•
Market counterparty account code
•
Internal flows are removed
02 April 2016
Rounded to one
significant figure
Each replaced by
an integer code
The trade database
•
17
02 April 2016
Currencies included:
•
AUDUSD
EURCHF
•
EURDKK
EURNOK
•
EURSEK
EURUSD
•
GBPUSD
NZDUSD
•
USDCAD
USDJPY
FX price series data
18
•
Time series data for all 10 currencies
•
Sampled at 1-minute frequency
•
Last traded price
02 April 2016
Data analysis
19
02 April 2016
What are the key questions?
•
Different perspectives on the data
• Traders
• Researchers
20
02 April 2016
Customer classification
•
Simple questions
•
Not so simple to answer
• Clustered trades
• Crosses
• Trending markets
21
02 April 2016
Data mining
22
•
Many ways of partitioning the data
•
Use of machine learning?
•
Temporal vs. static data
02 April 2016
Questions
23
•
How do we define the feature vector?
•
What are we seeking to predict?
•
What techniques are most applicable to this type of problem?
02 April 2016
Next steps…
24
02 April 2016
Going forward…
•
Data should allow us to better characterise our customers
•
A narrow view which looks at one customer at a time is
less likely to yield useful results than a wider view
•
25
Automated data mining techniques seem appropriate
02 April 2016
BNP Paribas London Branch has prepared this document. Although the information in this document has been obtained from sources which BNP Paribas believes
to be reliable, BNP Paribas does not represent or warrant its accuracy and such information may be incomplete or condensed. Information in this publication is
intended as a general outline of the subjects covered and should not be regarded as comprehensive or sufficient for making decisions, nor should it be used in place
of professional advice. It is not intended to provide the sole basis for any evaluation of a transaction or securities and you agree that the merits or suitability of any
such transaction or securities to your particular situation will be independently determined by you including consideration of the legal, tax, accounting, regulatory,
financial and other related aspects thereof. BNP Paribas does not accept responsibility for any loss arising from any action taken by anyone using this material. All
estimates and opinions included in this document constitute the judgement of BNP Paribas as of the date of the document and may be subject to change without
notice. BNP Paribas and its affiliates may, from time to time, effect or have effected an own account transaction in, or make a market or deal as principal in or for,
the instruments mentioned herein, or in options, futures and other derivative instruments based thereon and may, to the extent permitted by law, have acted upon or
used the information herein contained, or the analysis upon which it is based. This document is confidential and is being submitted to selected recipients only. It
may not be reproduced (in whole or in part) or delivered to any other person without the prior written permission of BNP Paribas.
This document is prepared for Professional Clients and is not intended for Retail Clients in the United Kingdom as defined in the FSA Rules and should not be
passed on to any such persons. By accepting this document you agree to be bound by the foregoing limitations.
BNP PARIBAS LONDON BRANCH, 10 HAREWOOD AVENUE, LONDON NW1 6AA TEL: (44 20) 7595 2000
FAX: (44 20) 7595 2555 WEBSITE: www.bnpparibas.com
BNP PARIBAS IS INCORPORATED IN FRANCE WITH LIMITED LIABILITY. REGISTERED OFFICE 16 BOULEVARD
DES ITALIENS, 75009 PARIS
BNP PARIBAS IS REGULATED BY THE FSA FOR THE CONDUCT OF INVESTMENT BUSINESS IN THE UK AND IS
A MEMBER OF THE LONDON STOCK EXCHANGE
BNP PARIBAS LONDON BRANCH IS REGISTERED IN ENGLAND AND WALES UNDER NO. FC13447.
REGISTERED OFFICE AT THE ABOVE ADDRESS
26
02 April 2016