2019 28.08.2019 Analysis lysis of Intraday day Trading ing in - - PowerPoint PPT Presentation

28.08.2019

2019

Timo Kern

The value of intrad raday ay electr trici city ty trad ading ng – Evalua uatin ting situa uati tion-depend ndent nt

• pportunity

tunity costs ts of flexible assets ts

Analysis lysis of Intraday day Trading ing in 2018

Absolut ute price ce deviat ation of

• f conti

tinuo nuous us intrad raday ay trad ading ng and intrad aday ay auction

2

• Representation of the

absolute deviation from mean continuous intraday price (ID3) and intraday auction price in 2018

• Hardly any seasonal or

hourly dependency can be recognized Are the average revenu nues in continuo inuous us intraday market situa uatio ion-depend ndent nt?

0:00 8:00 12:00 16:00 20:00 24:00 Time of day 20 15 10 5 >25 Absolute price deviation of continuous intraday trading and intraday auction in €/MWh 4:00 Month in 2018 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Methodolo

• dology

y for dete termining mining average ge reve venu nues in the conti tinuou

• us intrada

day y market Factor

• rs influe

fluenci cing g the leve vel l of

• f intrada

day y reve venue opportu tuniti ties Concl clusion ion and use of

• f the findings

dings

Outline ine

3

4

When is an electricit ricity y offering ing decision ion made in which ch market?

 Research topic: What revenue opportunities arise from continuous intraday trading? 12:00 Day-Ahead- Auction 15:00 0:00 Intraday- Auction 16:00 Continuous Intraday-Trading Zeit

5

Methodolog logy for deter termining ining reve venue opportunit unitie ies s in the he continu nuou

• us

s intraday day market

Approach according to Weber et al.*: Prices in the continuous intraday market develop according to a stochastic process and price changes are normally distributed.

Relative Häufigkeitsdichte Preis

Weber et al.*: Berücksichtigung von Intraday-Optionalitäten im Rahmen der Redispatch-Vergütung

Approach ch accor cording ding to to Webe ber et al.*: *:

Determina inatio ion n of the value lue of the intraday revenue nues by five parameters: 1. Distribution function of price changes (in Weber et al. Normal distribution) 2. The standard deviation 𝜏 the distribution 3. The expected value of the intraday price p at the time of evaluation 4. The marginal costs X 5. The amount of electricity offered

6

Methodolog logy for deter termining ining reve venue opportunit unitie ies s in the he continu nuou

• us

s intraday day market

Relative frequency density

Expected price Marginal costs Value of call option

Price

Weber et al.*: Berücksichtigung von Intraday-Optionalitäten im Rahmen der Redispatch-Vergütung

Approach ch accor cording ding to to Webe ber et al.*: *:

Determina inatio ion n of the value lue of the intraday revenue nues by five parameters: 1. Distribution function of price changes (in Weber et al. Normal distribution) 2. The standard deviation 𝜏 the distribution 3. The expected value of the intraday price p at the time of evaluation 4. The marginal costs X 5. The amount of electricity offered

7

Methodolog logy for deter termining ining reve venue opportunit unitie ies s in the he continu nuou

• us

s intraday day market

Value of put option

Weber et al.*: Berücksichtigung von Intraday-Optionalitäten im Rahmen der Redispatch-Vergütung

Approach ch accor cording ding to to Webe ber et al.*: *:

Determina inatio ion n of the value lue of the intraday revenue nues by five parameters: 1. Distribution function of price changes (in Weber et al. Normal distribution) 2. The standard deviation 𝜏 the distribution 3. The expected value of the intraday price p at the time of evaluation 4. The marginal costs X 5. The amount of electricity offered Relative frequency density

Expected price Marginal costs

Price

8

Assump mptio tion n Weber et al.: : Price changes es are distribu ibuted normall lly

9

Distrib ribut ution ion of

• f price changes from intraday

aday auctio ion to to continu nuous

• us intraday

day tradin ing

Consid nsideratio ion n of all l quarter- hours of the year 2018

• Price differences show normal

distributed characteristics only in strongly simplified assumption

• Distribution function from two

superimposed normal distributions represents empirical distribution much better  Use of the superim imposed normal l distrib ibut utio ion n functio ion ID_Auction – ID3

ID_Auction = Price intraday auction ID3 = volume weighted average price of the last three hours in continuous intraday trading

Relative frequency

Empirical distribution Normal distribution Superimposed normal distribution

10

Assump mptio tion n Weber et al.: : The expected ted value lue of continuous uous intraday aday trading ing corresponds nds to to the intrad raday ay auctio tion n value lue*. *.

* at the time of the auction (gate closure)

11

Time of

• f day

0:00 6:00 12:00 18:00 24:00 2017 2018 Average age deviat iatio ion ID3 - ID_Auct uctio ion in in €/MWh 3 2

• 1
• 2
• 3

1

Impact ct of

• f daytim

ime on expectatio ation of

• f intraday

aday prices

11

Consid nsideratio ion n of all l quarter- hours of the year 2018

• Slight day/night price

dependence can be assumed

• When looking at individual

quarter-hours, it is difficult to identify dependencies  Average, expected price deviation of zero seems to be reasonable

12

Average age deviat iatio ion ID3 - ID_Auct uctio ion in €/MWh 3 2

• 1
• 2
• 3

1 2017 2018 <20 20-30 30-40 40-50 >50 Day ahead ad forecas ast resiudal iudal loa

• ad in GW

Impact ct of

• f day ahead residual

ual load forecast st on expectation ation

• f
• f intraday

aday prices

12

Consid nsideratio ion n of all l quarter- hours of the year 2018

• Higher expected continuous

intraday prices for low residual load forecast

• Lower expected continuous

intraday prices for high residual load forecast  Residual load dependent average price deviation has impact on expected revenues

• f continuous intraday

trading

13

Assump mptio tion n Weber et al.: : The standar ard devia iation* ion* for each quarter ter-hour

• ur product

should ld be calcu cula late ted once a a month

* Of the deviation between the volume-weighted prices of the last three hours before close of trading and the prices of the intraday opening auction

Time of

• f day

0:00 6:00 12:00 18:00 24:00 2017 2018 30 25 15 10 5 20 Stand andar ard d deviat iatio ion ID3 - ID_Auc uction ion in €/MWh

Impact ct of

• f daytim

ime on standar ard deviatio ation of

• f intraday

aday prices

14

Consid nsideratio ion n of all l quarter- hours of the year 2018

• Slight day/night price

uncertainty can be assumed

• When looking at individual

quarter-hours, it is difficult to identify specific uncertainties

• f prices

 From empirical data it is not reasonable to use specific standard deviations for specific day times

2017 2018 <20 20-30 30-40 40-50 >50 Day ahead ad forecas ast resiudal iudal loa

• ad in GW

30 25 15 10 5 20 Stand andar ard d deviat iatio ion ID3 - ID_Auc uction ion in €/MWh

Impact ct of

• f day ahead residual

ual load forecast st on standar dard devia iation ion of

• f intrad

aday ay prices

15

Consid nsideratio ion n of all l quarter- hours of the year 2018

• High price uncertainty for

very low and high residual load forecasts

• Low price uncertainty for

moderate residual load forecasts  Residual load forecast has huge impact on price uncertainty of continuous intraday prices

16

Evaluati aluation

• n of
• f situ

tuation ation-de depen ende dent reven enues es at conti tinu nuous

• us intrad

raday tradi ding ng

Expected revenues at continuous intraday trading vary strongly in dependence of residual load forecast!

Exe xempla mplary inve vesti tiga gati tion

• n
• Biogas plant
• Marginal costs: 60 €/MWh
• Electricity price in the intraday auction: 59 €/MWh

Expe pecta ctati tion

• n va

value lue: : 59 €/MWh Standa dard d devi viati tion

• n:

18 €/MWh

1

Average ge expe pect cted reve venues in conti tinuou

• us intrada

day tradin ding 6,7 €/MWh Expe pecta ctati tion

• n va

value lue: : 61 €/MWh Standa dard d devi viati tion

• n:

23 23 €/MWh

2

10,2 €/MWh Expe pecta ctati tion

• n va

value lue: : 59 €/MWh Standa dard d devi viati tion

• n:

9 9 €/MWh

3

3,1 €/MWh High residual load forecast Low residual load forecast Moderate residual load forecast

Forschungsgesellschaft für Energiewirtschaft mbH Am Blütenanger 71 80995 München Tel.: +49(0)89 15 81 21 – 0 Email: info@ffe.de Internet: www.ffegmbh.de Twitter: @FfE_Muenchen

Timo

• Kern

Forschungsgesellschaft für Energiewirtschaft mbH Tel.: +49(0)89 15 81 21 – 35 Email: tkern@ffe.de

18

0,00 20,00 40,00 60,00 80,00 100,00 120,00 06.08. 15:00 06.08. 21:00 07.08. 03:00 07.08. 09:00 07.08. 15:00

Preis in €/MWh Handelszeitpunkt 15qh4 15qh3 15qh2 15qh1

Analysis lysis of price structur ures s of the cont. Intrad aday ay trading ing

Cha haracteris istic ics of continu inuous us intraday prices

• Trading takes place mainly in the

three hours before delivery

• Partly high volatility

Time of trading Transaction price in €/MWh

19

0,00 20,00 40,00 60,00 80,00 100,00 120,00 07.08. 09:00 07.08. 15:00

Preis in €/MWh Handelszeitpunkt 15qh4 15qh3 15qh2 15qh1

Analysis lysis of price structur ures s of the cont. Intraday y trading ing

Cha haracteris istic ics of continu inuous us intraday prices

• Trading takes place mainly in the

three hours before delivery

• Partly high volatility
• For the analysis of continuous

intraday trading, product ID3 is evaluated (volume weighted average price of the last three hours in continuous intraday trading).

ID3 qh1

Time of trading Transaction price in €/MWh

Wie gut we werden reale e Erlöse se auf dem kontinuie uierli rliche hen n Intraday day-Markt Markt durch Vertei teilu lung ngsf sfun unktion tionen n repräs äsen entie tiert?

21

Reale ale und simuli lier erte e Erlöse löse im kont ntin inuie ierlich rlichen en Intra traday day Handel del in 2018 18

Modellierte Erlöse der überlagerten, situationsabhängigen Verteilungsfunktion trifft reale Erlöse wesentlich besser als die der Normalverteilung

1,000 2,000 3,000 4,000 5,000 6,000 <=20 20-30 30-40 40-50 >50

Intraday Erlöse in €/a Residuallast in GW

Exempla laris ische he Analy lyse:

• Leistung: 1 MW
• Variablen Grenzkosten: 30 €/MWh
• Erstellung Verteilungsfunktionen an

Daten aus 2017

• Simulation der Erlöse im Jahr 2018

Reale Erlöse Normalverteilung Situationsabhängige, überlagerte Normalverteilung

22

Faz azit it und Nutzung ng der Erken ennt ntni nisse

Welche lche Parame meter sind für eine Intrada day-Opti Option

• nalit

ität entsch cheide idend? d? 1) Verteilungsfunktion der Preisänderungen (in Weber et al. Normalverteilung) 2) Den Erwartungswert des Intraday-Preises p bei Fälligkeit der Option 3) Die Standardabweichung 𝜏 der noch unbekannten Intraday-Preise 4) Die variablen Kosten X bei einem thermischen Kraftwerk 5) Die zu vergütende Leistung Wie charakteris isie ieren sich diese Parame meter?

• Klare Abhängigkeit der Standardabweichung von Windenergieprognosen/Residuallastprognosen
• Auch Erwartungswert des kontinuierlichen Intradayhandels mit residuallastsabhängigen Schwankungen

 Preisänderungen gut repräsentiert durch situationsabhängige, überlagerte Verteilungsfunktion

Durch rch das Verfahr ahren n können n situa uati tions nsab abhängig hängige Opportuni tunität täten n für den konti tinui nuierli rlich chen n Intrad raday ay-Mark arkt t bestimm mmt t werd rden, n, um die Einsatz atzents ntsch cheidung ng von Flexibilität täten n auf anderen n Märk rkten n zu optimieren

23

Analys lyse e der r Sprea eads ds im Day-Ahea ead-Ma Mark rkt

23

2014

30 €/MWh

Jahr Monatliche Mittelwerte 60 40 20 >80 Day-Ahead-Spread in €/MWh Tägliche Spreads 2015

29 €/MWh

2016

23 €/MWh

2017

30 €/MWh

2018

32 €/MWh

Methodik

• dik

& Erkenntnis tnisse

• Darstellung der maximalen, täglichen Spreads der letzten 5 Jahre
• Variation der Spreads zwischen 7 und 150 €/MWh
• Saisonales Profil mit höheren Preisunterschieden in den Wintermonaten
• Schwankungen der durchschnittlichen, jährlichen Spreads wesentlich geringer als Schwankungen

des Spotpreises (2017: 34,2 €/MWh; 2018: 44,5 €/MWh)