Big Data: A crucial challenge for energy players July 2014 www.chappuishalder.com Twitter : @ch_retail
Jan 27, 2015
Big Data: A crucial challenge for energy players
July 2014
www.chappuishalder.com Twitter : @ch_retail
2
Big Data revolution is in motion in the energy sector
Given recent evolutions on energy markets, remaining competitive implies to be able to process a growing flow of data.
We strongly believe that the keys to success are two-fold when it comes to designing a successful Big Data strategy:
• A structured and robust framework
• A continuous upgrade of hardware and infrastructure to stick to volume of data and complexity of analyses
Big Data represents a source of business opportunities for energy players…
… from rethinking client relationship to crafting tomorrow’s risk management.
Quality
Depth
Exhausti-veness
Centrali-sation
« Real time » data
« Data storage »
Exogen-ous data
Big Data
3
Evolutions within energy markets imply a growing flow of data to be processed…
Energy players strengthen their international footprint as more regions reach crucial liquidity to generate significant trading volumes …
… and multi-commodity players are becoming the rule
Extension to 24 hours/7 days opened derivatives and spot exchanges • Pan-European Gas Market Exchange (PEGAS)
will be one, as of July (still no exchange in Spain)
• New products emerge that can modify supply and hedging options and strategy
Numerous parameters required to optimize portfolio management (power plants)
Providers increase real-time flow on meteorological data
Development of smart metering increases the volume and frequency of data collection: • Giving suppliers/ shippers a better view on
how energy is consumed (at grid or user level)…
• And enabling operators (transmission and distribution systems, LNG Terminal & Storage) to publish real time data and forecasts
Supply problems or political and
regulatory decisions must be known rapidly and taken into account in order to be exploited on several time horizons
… as business conduct gets more complex … and more information becomes available
Implementation of a successful Big Data strategy requires a structured and robust framework, from acquisition of info to decision-making process
Acquire Analyze Organize Decide 1 3 4 6
Control 2 Report 5
Structured data: • Ex. market data gathered
both on supply and demand side
Unstructured data: • Non-parametric statistics
and unconventional data • Ex.: Twitter content
A dataset clean of errors
• Wrong data can cause you to make wrong decisions
• Warning mechanism and auto correction replacing wrong data with substitutes
Process collected data:
• Reduce -> Aggregate -> Enrich -> Structure
Create datasets:
• Use relational technology to assemble and match data
Data modeling: • Fast read and write speeds • Build statistical models
comparing data sets’ variables and datasets
Find correlations: • Bring additional value to
classical market analysis • Machine learning
approach to capitalize on past events knowledge
Sort through to extract the useful and stay synthetic
• Give it meaning, and make it exploitable,
• Show the big picture and focus on a few key points
Anticipate • Predict market trends / price • Estimate likelihood of external
factors (weather, media impact …)
Make predictive decisions • New market positions to
enhance margins • Ex: new capacity planning,
capacity expansions
5
Big Data requires investing in new forms of data processing…
Integrated framework
Dynamic framework
Static framework
3
4
5
6
2
1 0
Data Collection Data warehouse IT implementation & Infrastructure
Data Storage Historical data
Data Cleaning Data quality management Data transformation
Data Statistical Description Mean, Median, standard deviation Histogram…| VaR 95,99%
Data Analysis Clustering & segmentation Automatic classification | Factorial analysis
“X factor” Mining Web mining (behavior…) Image mining (face recognition) |
Text mining
Data Mining & Big Data Prediction Ranking / discrimination Anticipation & simulation
Today’s average position
Past
Future
Present
7 Artificial Intelligence Self-Learning models (auto
efficient) Multi crossing data set
Integrated framework
Dynamic framework
Static framework
6
… that will necessitate continuous hardware and infrastructure upgrades to adapt to the volume and complexity of data to be managed
Flexibility & Adaptability
Change the way you treat the data according to your continuous experience
You always need to handle more data, more frequently and with agility, implying great storage capacities
Power & Speed
Windows of opportunity may close up quickly therefore your calculation speed should be optimized ASAP
As long as you keep manual steps, you will not reach optimum -> automation must be your motto
Modeling Computation Storage
Self-detection, auto upgrading and efficient models
Optimized computation capacity
Centralized, unlimited data storage capacity
Homogenisation of modeling practices, business incentives for Data modelling technique development
Good computation capacity
Aggregation of data sources (Finance, Risk, Marketing, Sales…)
R&D Development Limited computation capacity
Limited storage capacity
Integrated
Dynamic
Static
7
Numerical transformation induced by Big Data has root causes and objectives that transcend sectors, and the energy industry is no exception
Source : Ventana Research | 2013
Source : Analytics | IBM Institute | 2012 Refocus on customer (CRM)
55%
Process optimisation incl. cost optimisation
4% New business
model 15%
Risk management/ Financial reporting
23%
Collaborative working mode
2%
Cost reductions
Reduce and limit manual steps
Produce daily results evermore precisely
Increase the compute speed
Store and analyze even more data
But in addition to these ‘standard’ objectives, players on energy markets will focus on specific issues…
8
On-demand data mining to dig into meta-data Risk and P&L indicators (calculation and reporting) P&L explain – detection of abnormal variations - VaR back-testing
Data preparation for EMIR, Basel II/III, MIFID, REMIT, audits …
To support KYC, rogue trading, AML or anti-fraud process …
Pre-trade decision support (ex. locational/ geographical spreads, time spreads on storage etc.)
To help identify trades from various systems to avoid missed or duplicated trades
… with business or process orientations
Automatically executed quantitative processes or High-Frequency-Trading
Real-time optimization of day-ahead and intra-day position coverage
Optimize client consumption forecasts (short – medium – long-term) Optimize production forecasts (generation)
Bu
sin
ess
-ori
en
ted
is
sue
s P
roce
ss-o
rien
ted
issu
es
Data Tagging
Trading Analytics
Watch Tower
Regulation
Financial Data Management
Forecasting
Hedging Strategy
Systematic Trading
9
Don’t let yourself be overrun by competitors …
Hardware and infrastructure upgrades induced by Big Data (storage, calculation capacity, etc.) must not overshadow the necessity of investing in human capital.
Players on energy markets that will best ride the Big Data wave will not only get their heads above water in a harsh competition context…
Controlling the three V’s of data (Variety, Volume and Velocity) creates an alternative information edge, which is:
• A potential new source of uncorrelated excess returns
• Advanced techniques for valuing clients and deals
• Most helpful in risk and performance management
• Easing data management for internal purposes
10
CH&Cie at a glance
Management Consultancy ... … for Financial Services & Commodities
Retail Banking
Private Banking Corporate & Investment Banking
Insurance Commodities Customer
Experience
Risk & Finance
IT & Operations
Business Development
11
8 offices around the world, in major trading and financial places…
… 100+ consultants, with strong academic backgrounds and experience
Business school 60%
Engineering school 30%
Others 10% In average, CH&Cie consultants have 7 years of
experience within consulting firms, Financial Services and Commodities.
12
Your contacts for this offer
Director - Head of CH&Cie Commodities + 33 6 40 56 21 71 [email protected] Paris Office
CEO & Partner + 44 78 34 55 03 98 + 33 6 12 41 64 06 [email protected] London Office
Partner + 44 203 427 3559 + 33 7 87 68 81 77 [email protected] London Office
Manager + 33 6 65 02 80 07 [email protected] Paris Office
Geneva Office
Rue de Lausanne 80
CH 1202 Genève, Suisse
London Office
50 Great Portland Street
London W1W 7ND
Paris Office
25 rue Alphonse de Neuville
75017 PARIS