Proposed new scoring system for Football Predictor games
We have been reviewing our Football Predictor games, and soliciting some feedback on the scoring system in particular, and are now proposing a change to the scoring ahead of the World Cup in June.
Why fix what isn't broken?
The main feedback is that our current scoring system is mostly fair and that our Closeness Index formula generally works and offers more sophistication than other Predictor games. However, many people have said that it’s not immediately clear how the scoring works, nor is it easy to explain to a friend when you invite them to play. Moreover, whilst watching a game, it’s not always obvious how many points you will win if the scoreline stays as it is, which takes away some of the excitement.
The other contention with the scoring system is derived from the Bonus Point, and how to determine whether one prediction is closer than another if neither are exact. The BP means that your score is not just dependent on your own prediction - it's dependent on what your pool rivals have picked as well, and it's not easy to calculate.
With that challenge in mind, we are proposing the following, which I am nicknaming 'Sophisticated Simplicity':
* Win Point: 1pt
This is unchanged - 1 point for predicting the correct match result
PLUS
* Close pick: 0.5pts, OR
* Exact score pick: 2pts
There would be no Bonus Points for the closest pick in the pool.
Benefits
The key benefit is simplicity.
Removing the Bonus Points means we don't need to define degrees of closeness to award points - in the new world, you are either close or you are not.
Being close or not will still be defined by our Closeness Index, but essentially if you are one goal away from the exact score, you are close, and the points you win do not depend on how close anybody else was.
It should add greater excitement during a game, as you’ll know what effect one more goal at either end would have. For example, I picked Man City to beat Stoke 0-2 on Monday night, and I knew that several of my colleagues had picked 0-3 - but I didn’t really know what effect a third goal would have, because they would have split the bonus point, and I wasn’t entirely sure what level of margin point I would have won for a 0-2 pick in a 0-3 result.
In the new world, I would have known that my exact pick would yield a total of 3 points, and my rivals just 1.5 points. But a third goal would give them all 3 points, and put me back to 1.5, and so I would have been willing Stoke’s defence to hold firm much more than I was already just by knowing the exact impact of one goal.
This proposed model is also in line with other football predictor games, where the standard scoring system rewards correct result picks, plus exact score picks. This is important for new people that you invite to play the game, or that discover Superbru another way, as they come with certain expectations of how the scoring should work.
Adding in the points for a Close pick will create greater scoring differentiation than in other football predictor games, whilst fairly rewarding pick accuracy, and adding a greater level of sophistication than most other games.
Removing the bonus points simplifies things further, since these are often shared in the current system, resulting in fractions of points being awarded, whilst also leading to further scoring confusion where people have different points totals in different pools.
Simple, yet sophisticated. Sophisticated simplicity.
Data modelling
We don’t do things by halves at Superbru HQ and we consider any potential change very carefully. To that end, we have modelled the changes on over 5 million Premier League predictions, and discovered some interesting effects of the adjustment.
The main thing to note is that the effect of the change is not huge, and those that make the best predictions will still win.
Overall, slightly more points will be awarded, mostly because an exact pick will win 3 points, rather than a maximum of 2.75 in the current system - and because bonus points are often split, an exact pick can often yield just 2 points.
We found that accurate predictions were slightly better rewarded. In the current system, those that pick the odd outlier result and collect a full bonus point will often do well, even if they predict fewer results correctly over the course of the season, and they can outperform those that consistently pick more accurately. Rewarding accuracy feels like a good outcome.
We also modelled the impact of awarding Close points even if the result is not predicted correctly - e.g. you pick 2-1 and the match finishes 1-1, which is pretty close. However, this rewards more conservative predictions - a 2-1 prediction would score points in 64% of all results, compared to just 41% without it - and those blindly picking 2-1 for every match would win.
What do you think?
Whilst we are largely in favour of this change at HQ, we would still like to hear your feedback. Please take the one-question survey in the box below, and add any further feedback in the comments below
Why fix what isn't broken?
The main feedback is that our current scoring system is mostly fair and that our Closeness Index formula generally works and offers more sophistication than other Predictor games. However, many people have said that it’s not immediately clear how the scoring works, nor is it easy to explain to a friend when you invite them to play. Moreover, whilst watching a game, it’s not always obvious how many points you will win if the scoreline stays as it is, which takes away some of the excitement.
The other contention with the scoring system is derived from the Bonus Point, and how to determine whether one prediction is closer than another if neither are exact. The BP means that your score is not just dependent on your own prediction - it's dependent on what your pool rivals have picked as well, and it's not easy to calculate.
With that challenge in mind, we are proposing the following, which I am nicknaming 'Sophisticated Simplicity':
* Win Point: 1pt
This is unchanged - 1 point for predicting the correct match result
PLUS
* Close pick: 0.5pts, OR
* Exact score pick: 2pts
There would be no Bonus Points for the closest pick in the pool.
Benefits
The key benefit is simplicity.
Removing the Bonus Points means we don't need to define degrees of closeness to award points - in the new world, you are either close or you are not.
Being close or not will still be defined by our Closeness Index, but essentially if you are one goal away from the exact score, you are close, and the points you win do not depend on how close anybody else was.
It should add greater excitement during a game, as you’ll know what effect one more goal at either end would have. For example, I picked Man City to beat Stoke 0-2 on Monday night, and I knew that several of my colleagues had picked 0-3 - but I didn’t really know what effect a third goal would have, because they would have split the bonus point, and I wasn’t entirely sure what level of margin point I would have won for a 0-2 pick in a 0-3 result.
In the new world, I would have known that my exact pick would yield a total of 3 points, and my rivals just 1.5 points. But a third goal would give them all 3 points, and put me back to 1.5, and so I would have been willing Stoke’s defence to hold firm much more than I was already just by knowing the exact impact of one goal.
This proposed model is also in line with other football predictor games, where the standard scoring system rewards correct result picks, plus exact score picks. This is important for new people that you invite to play the game, or that discover Superbru another way, as they come with certain expectations of how the scoring should work.
Adding in the points for a Close pick will create greater scoring differentiation than in other football predictor games, whilst fairly rewarding pick accuracy, and adding a greater level of sophistication than most other games.
Removing the bonus points simplifies things further, since these are often shared in the current system, resulting in fractions of points being awarded, whilst also leading to further scoring confusion where people have different points totals in different pools.
Simple, yet sophisticated. Sophisticated simplicity.
Data modelling
We don’t do things by halves at Superbru HQ and we consider any potential change very carefully. To that end, we have modelled the changes on over 5 million Premier League predictions, and discovered some interesting effects of the adjustment.
The main thing to note is that the effect of the change is not huge, and those that make the best predictions will still win.
Overall, slightly more points will be awarded, mostly because an exact pick will win 3 points, rather than a maximum of 2.75 in the current system - and because bonus points are often split, an exact pick can often yield just 2 points.
We found that accurate predictions were slightly better rewarded. In the current system, those that pick the odd outlier result and collect a full bonus point will often do well, even if they predict fewer results correctly over the course of the season, and they can outperform those that consistently pick more accurately. Rewarding accuracy feels like a good outcome.
We also modelled the impact of awarding Close points even if the result is not predicted correctly - e.g. you pick 2-1 and the match finishes 1-1, which is pretty close. However, this rewards more conservative predictions - a 2-1 prediction would score points in 64% of all results, compared to just 41% without it - and those blindly picking 2-1 for every match would win.
What do you think?
Whilst we are largely in favour of this change at HQ, we would still like to hear your feedback. Please take the one-question survey in the box below, and add any further feedback in the comments below
Superbru on X
