Comparing fast- and slow-acting features for short-term price predictions
2017 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE credits
Student thesisAlternative title
Jämförelse av snabba och långsamma variabler för kortsiktig prisprediktion (Swedish)
Abstract [en]
This thesis compares two groups of features for short-term price predictions of futures contracts; fast- and slow-acting features. The fast-acting group are based on limit order book derived features and technical indicators that reacts to changes in price quickly. The slow-acting features constitute of technical indicators that reacts to changes in price slowly.
The comparison is done through two methods, group importance and a mean cost calculation. This is evaluated for different forecast horizons and contracts. Furthermore, two years of data was provided to do the analysis. Moreover, the comparison is modelled with an ensemble method called random forest. The response is constructed using rolling quantiles and a volume weighted price.
The finding implies that fast-acting features are superior at predicting price changes on smaller time scales, while long-acting features are better at predicting prices changes on larger time scales. Furthermore, the multivariate model results were similar to the univariate ones. However, the results are not clear-cut and more investigation ought to be done in order to confirm these results.
Abstract [sv]
Den här uppsatsen jämför två typer av variabler för kortsiktig pris prediktion av terminskontrakt; snabba och långsamma variabler. De snabba variablerna är en sammansättning av limit order boks härledda variabler och tekniska indikatorer som svarar snabbt på prisförändringar. De långsamma variablerna utgörs av tekniska indikatorer som svarar långsamt på prisförändringar.
Jämförelsen är gjord med två metoder, "group importance" och genomsnittskostnad. Detta har gjort för olika prediktionshorisonter och kontrakt. Två-års data användes för att göra analysen. Detta modellerades med en gruppmetod, kallad "random forest". Responsvariabel är konstruerad med rullande kvantiler och ett volymviktat pris
Place, publisher, year, edition, pages
2017.
Series
TRITA-MAT-E ; 2017:24
National Category
Probability Theory and Statistics
Identifiers
URN: urn:nbn:se:kth:diva-208178OAI: oai:DiVA.org:kth-208178DiVA, id: diva2:1107306
External cooperation
Lynx Asset Management
Subject / course
Mathematical Statistics
Educational program
Master of Science - Applied and Computational Mathematics
Supervisors
Examiners
2017-06-092017-06-092022-06-27Bibliographically approved