Advanced

Prediktioner av allsvenska fotbollsmatcher - Modellering och testande av strategier för satsning av kapital på spelbolagsmarknaden

Christiansen, André LU (2019) STAH11 20182
Department of Statistics
Abstract
This paper investigates if there is a quantitative relationship between Expected Goals (xG) and results in football games for the Swedish top-ranked league, Allsvenskan. Multicategory logit models are fitted with different types of explanatory xG-variables to see if these models can give better predictions than the bookmakers, and if there’s a possibility of generating a positive return using different betting strategies. An indication of a quantitative relationship was found, and a positive return was generated. However, the models couldn’t give better predictions in comparison with the bookmakers.
Abstract (Swedish)
Denna uppsats redogör för om det finns ett kvantitativt samband mellan Expected Goals (xG) och matchresultat för fotbollsmatcher i Sveriges högst rankade fotbollsserie, Allsvenskan. Modeller baserade på multikategorisk logistisk regression anpassas med olika varianter av förklarande xG-variabler för att kunna se om dessa modeller kan ge bättre prediktioner än spelbolagen och generera en positiv avkastning på spelmarknaden med hjälp av spelstrategier. Det fanns en antydan till ett kvantitativt samband mellan xG och matchresultat och positiv avkastning kunde genereras. Modellerna kunde dock inte ge bättre prediktioner vid jämförelser med spelbolagen.
Please use this url to cite or link to this publication:
author
Christiansen, André LU
supervisor
organization
course
STAH11 20182
year
type
M2 - Bachelor Degree
subject
keywords
Expected Goals, xG, predictions, logistic regression, football, soccer
language
Swedish
id
8967642
date added to LUP
2019-03-29 09:02:31
date last changed
2019-03-29 09:02:31
@misc{8967642,
  abstract     = {This paper investigates if there is a quantitative relationship between Expected Goals (xG) and results in football games for the Swedish top-ranked league, Allsvenskan. Multicategory logit models are fitted with different types of explanatory xG-variables to see if these models can give better predictions than the bookmakers, and if there’s a possibility of generating a positive return using different betting strategies. An indication of a quantitative relationship was found, and a positive return was generated. However, the models couldn’t give better predictions in comparison with the bookmakers.},
  author       = {Christiansen, André},
  keyword      = {Expected Goals,xG,predictions,logistic regression,football,soccer},
  language     = {swe},
  note         = {Student Paper},
  title        = {Prediktioner av allsvenska fotbollsmatcher - Modellering och testande av strategier för satsning av kapital på spelbolagsmarknaden},
  year         = {2019},
}