Analyzing The Likelihood Of A Shot Being Targeted In Soccer

INTRODUCTION:
This project is to analyze the likelihood of a shot (an attempt on goal) being scored in soccer as well as multiple of the factors that make a team successful such as how many shots they take per match, how many shots they take overall and how many goals they score and where they’re scoring and taking shots from. With this data I will then predict the league standings for the following league season as well as how likely a team is to score from different areas of the field.
To do this I’ll be creating an Expected Goals (ExpG) model for the 2015/16 Premier League season utilizing data from the 2014/2015 PL season. For this investigation I’ll be utilizing
Premier League shot indexes provided for public use by Opta and WhoScored

Whereas the players in the higher ranked teams are converting more of their chances due to their better quality, and that’s why the high ranking teams have a smaller amount of Expected Goals for next season than actual goals for this season because their players perform above average. This is the biggest fallacy for the ExpG model, it cannot take into account individual player quality, and that’s why some teams that should be ranked lower cannot be ranked lower.
In relation to the Goals Scored and Shots Taken vs. Position on League Table graph:
There is a clear correlation between goals scored and shots taken as well as goals scored and shots taken in relation to league position. The more shots you take the more likely you are to score and the higher on the league table you will be; on the reverse, the less shots you take will likely lead to less goals which in turn will lead to a lower position on the league

The thing about soccer and specifically related to ExpG and the English first tier is that there’s no absolutes, with that being said this season in England Leicester (ranked 11th in my ExpG model) is in first place with reasonable advantage over the second place Spurs (ranked 10th in my ExpG model), both teams when considering their ExpG shouldn’t be anywhere near the top two spots but they are, so prediction strategies like ExpG aren’t absolute because they can’t predict human nature and what comes with it.
My model also does not take into consideration new players a team might buy or the fact that relegations occur in the league and next season three new teams will join the league and three will leave. So to sum it, appropriate techniques were used to collect the data as well as appropriate mathematics to analyse the data but it cannot be fully valid because human nature is inconsistent and cannot be fully predicted, any model that tries to predict human behavior has to be taken with a grain of salt and my ExpG model is no exception.
Bibliography:
Shot Indexes,