Prediction Analysis

Football Buddy can try to predict scores based on prediction models that you create.  This can be achieved by blending the pre-configured algorithms in the way that you think will work best.

Simulator%20Screen%20Shot%20 %20iPad%20Pro%20%2812.9-inch%29%20%284th%20generation%29%20-%202020-07-21%20at%2020.40.05

Create your own Prediction Models

Create prediction models that fit your expections. Different models may work better at different times of the year.

Simulator%20Screen%20Shot%20 %20iPad%20Pro%20%2812.9-inch%29%20%284th%20generation%29%20-%202020-07-21%20at%2020.41.54

Blend the algorithms that you think will work best

The graph tracks performance of the prediction model over time.  You will find some models work better at certain points in a season.

There are various algorithms available that can be blended together to make your perfect prediction.
See below for a complete list and explanation of each.

Simulator%20Screen%20Shot%20 %20iPad%20Pro%20%2812.9-inch%29%20%284th%20generation%29%20-%202020-07-21%20at%2020.42.22

League Table

Score Predictions happen automatically, but its possible to customise how many fixtures are predicted and for which divisions.  

Prediction models which blend multiple algorithms over large numbers of fixtures will require intensive calculations which may take time to perform.  To reduce the calculation time, reduce the number of fixtures that are predicted, and remove divisions that you have no interest in.

Prediction Algorithms

1. Result performance trend analysis

This is a new algorithm introduced in FB2, and is only available with the purchase of FB+.

It attempts to look at recent performance of the two teams compared with what would be expected.  For example, if Team A was at the top of the league and Team B was at the bottom, it would be expected that Team A would win with a significant margin.  Anything less than that may be an indicator of sub-optimal performance.  By trending these statistics for both teams, we can try to establish an upward or downward trend in relative performance.  This performance ratio is then applied to the anticipated result given the league position delta of both teams.

Assessing performance over a large number of matches will flatten the trend, and recent changes in performance will be less evident.

Using this algorithm at the start of the season (or for cup matches) may be less reliable.

2. Scoreline using same match rating

Match Rating is a comparison of the recent results of two teams.  A positive value indicates a probable outcome in favour of the home team, and a negative value reflects an outcome in favour of the away team.  The bigger the number, the higher the liklihood.  

Match Rating is calculated over the last 5 games as;
((HomeGoalsFor) - (HomeGoalsAgainst)) - 
((AwayGoalsFor) - (AwayGoalsAgainst))

This algorithm then looks for previous matches with the same Match Rating and averages their scorelines to produce an expected outcome.

Match Rating does not consider the strength of the  opposition of each team.  So a team with easy matches prior to the current one, may have an artificially high Goal Difference which will skew the match rating.  When this is applied to many games to achieve an average, the scorelines may vary significantly making the resultant score unpredictable.

3. Scoreline for the same fixture

This algorithm simply looks at the previous encounters of the two teams and takes an average of the scorelines.

If you select the “Prioritise the most recent” option, then the algorithm will give more weight to the most recent results.

4. Similar league position delta

League position is a good indicator of relative performance after a season has been underway for a while.  This algorithm compares the league positions of each team and looks for previous results when there was a similar delta - it also tries to check for results at a similar point in the year (therefore at a similar point in the season).

Optionally, the algorithm will also look for the same Match Rating, for the same league position delta. As noted above, Match Rating can be erratic so adding this to the league position comparison does not always give a better accuracy.

Be wary about using this algorithm at the start of the season.  It will also provide erratic results for cup games, where there is no league table.

5. Scoring record of each team

This simple routine simply takes the average (or trend) of the goals scored for for the home team to give the home goals scored, and the goals scored by the away team to calculate the goals scored by the away team.

6. Conceding record of each team

This is the inverse of the previous rule whereby the goals conceded by the home team makes the goals scored by the away team, and the goals conceded by the Away team makes the goals scored by the home team.   

7. Bookmakers’ Favourite

This new algorithm is only available with purchase of FB+.

It aggregates odds data collected from multiple Bookmakers to ascertain the bookmakers’ overall favourite for a match.  It then uses that information to select their most likely scoreline for that result outcome.

In other words, this algorithm is trying to leverage the bookmakers’ knowledge to determine what the score will be.  

Get the app



© Difference Solutions Ltd 2021.  All rights reserved.