PDA

View Full Version : Math help needed



Hall3
10th December 2016, 03:59 AM
i thought i was working something out right but apparently not

I was looking at the % of both teams to score in a soccer game

Brisbane scored in 83% of their home games
Newcastle scored in 73% of their away games
so added them both up and divided by 2 to get 78 which I thought would be the odds that the teams would both score newcastle v brisbane would be 78%??
but got told WRONG something to do with the mean then turn succes rate into odds? any idea on this one?

admin
12th December 2016, 11:03 AM
Hi Hall3,

Welcome to the forum!

For both teams to score, bookmakers take into account both offensive and defensive stats. For example a team that scores in 83% of their home games will have a higher than 83% chance of scoring against a team with a weaker defensive record on average, and a lower than 83% chance of scoring against a team with a stronger than average defence. Most likely the calculations involve the defensive strengths of Brisbane and Newcastle in addition to their scoring strengths.

The calculation probably isn't based on an average of the two team's scoring statistics because both teams need to score for the wager to win. For this reason it may rely more heavily on the team with the lower likelihood of scoring. If Newcastle score in 73% of their away games they may actually have some clean sheets during those fixtures, so the probability of both teams scoring is probably less than 73%.

One model (out of many) for determining stats such as these is the Poisson distribution. An online calculator (on our sister site) that uses this model can be used here:
http://www.aussportstipping.com/sports/a_league/poisson/
Enter Brisbane as the home team and Newcastle as the away team. You can also choose the range of historical data to base the calculations on. Based on the last 12 months of data (12 Dec 2015 - 12 Dec 2016), the model predicts there's a 49.38% chance of both teams scoring. When I use all available data (back to 2008) the model predicts there's a 49.02% chance.

Just note that the above model doesn't incorporate situational details such as player injuries, derbies or the weather. It also doesn't account for head-to-head stats, which bookmakers might also look at when making their calculations.

As for turning probabilities into odds, for decimal odds simply take the reciprocal of the probability. For example, 78% equates to 1/0.78 =1.282.