Search

På grunn av den endrende likviditet til kraftderivater gjennom kontraktens livsyklus, er effektive strategier for ordreutførelse høyst relevante for strømprodusenter. Det relevante aspektet for en strømprodusent som ønsker å redusere risiko for deler av produksjonen sin ved å handle kraftderivater, handler om hvordan transaksjoner kan utføres slik at man balanserer preferansen for risikoaversjon og maksimering av inntekter. Den eksisterende litteraturen innenfor fagfeltet fokuserer på ordreutførelse i et vakuum, uten å betrakte avgjørelser utenfor handelsperioden. Samtidig er den eksisterende litteraturen, til forfatternes viten, fokusert på handel av aksjer med høy markedsverdi. Denne avhandlingen fyller en mangel i litteraturen ved å evaluere strategier for ordreutførelse i et illikvid marked. Formålet med denne avhandlingen er å lage en modell for ordreutførelse, som imøtekommer kravene til en strømprodusent. En integrert orderutførelsesmodell, som heretter vil omtales som the Integrated Postponement model, foreslås derfor. The Integrated Postponement model introduserer utsettelsesopsjonalitet, og dynamiske handelsvolumer basert på markedslikviditet, hvor begge disse aspektene er nye bidrag innenfor feltet. The Integrated Postponement model er presentert som en flerstegs stokastisk programmeringsmodel med rullende tidshorisont. Ytelsen til the Integrated Postponement model har blitt tested ved hjelp av en backtest-metodologi som tar i bruk mikrostrukturen til en ordrebok. The Integrated Postponement model økte inntektene fra handel av strømderivater med 89 bps sammenlignet med den beste referansestrategien. En kan konkludere fra resultatene at utsettelsesopsjonaliteten øker ytelse ved å kunne være mer selektiv med tanke på handelsavgjørelser. Samtidig ble det vist at dynamiske handelsvolumer økte inntektene med 39 bps. Due to the changing liquidity of power derivatives contracts during their lifecycle, efficient order execution strategies are highly relevant for power producers. The relevant aspect for a power producer aiming to hedge part of their production by trading power derivatives corresponds to how order execution decisions can be made to balance the trade-off between the preference for risk aversion and maximising revenues. The existing literature emphasises the order execution problem in a vacuum without considering decision making outside of the trading period. Additionally, the existing literature predominantly considers order placement for blue-chip stocks to the author's best knowledge. This thesis fills a gap in the literature by evaluating order execution strategies in an illiquid market context. The purpose of the thesis is to construct an order execution model that meets the requirements of a power producer. An integrated order execution model, referred to as the Integrated Postponement model, is proposed. The Integrated Postponement model introduces postponement optionality and dynamic trading volume allocation based on market liquidity, of which both aspects are novel contributions to the field. The Integrated Postponement model is composed of a multistage stochastic hedging model with a multistage stochastic order execution model. The trading performance of the Integrated Postponement model has been evaluated using a backtesting framework that incorporates the limit order book microstructure. The Integrated Postponement model was seen to increase trading revenues by 89 bps compared to the best-performing benchmark. One can conclude from the results that the postponement optionality leads to a better trading performance by being more selective. Additionally, the introduction of dynamic trading volume allocation was seen to increase trading revenues by 39 bps.

This is an author's accepted manuscript version of an article published by Springer in the Lecture Notes in Computer Science series. The final authenticated version is available online at https://doi.org/10.1007/978-3-030-14174-5_14 Argumentation theory can encode an agent’s assessment of the state of an exchange of points of view. We present a conservative model of multiple agents potentially disagreeing on the views presented during a process of deliberation. We model this process as iteratively adding points of view (arguments), or aspects of points of view. This gives rise to a modal logic, deliberative dynamic logic, which permits us to reason about the possible developments of the deliberative state. The logic we propose applies to all natural semantics of argumentation theory. Furthermore, under a very weak assumption that the consensus considered by a group of agents is faithful to their individual views, we show that model checking these models is feasible, as long as the argumentation frameworks, which may be infinite, does not have infinite branching.