Predicting future results for fantasy leagues
I think we need to come up with a way for predicting future results in head-to-head fantasy leagues based only on how many points your team has scored in its games so far.
In these types of leagues you have no control over how many points your opponents may or may not score against you, so formulas like Pythagorean and Pythagenpat, which work on the difference between your score and your opponents score, don't give a realistic result regardless of what you set your exponent to.
Ideally a team that has scored way above the average points per game compared to the rest of the league should be predicted to win the majority of their remaining games. However, I'm not clever enough to come up with a formula on my own! Anyone willing to suggest something?
Steve R
shared this idea
Good, I’ll do this. Thanks.
4 comments
-
Rhys Stephens
commented
One solution could be to modify the pythagorean/pythagenpat to simply replace the against points with the league average. As Steve mentioned above the points against really have no bearing on things. If you're scoring above average each week you are more likely to win.
I calculated an exponential of 7.211 for five seasons (520 games) of my fantasy league using the formula PtsFor^x/(PtsFor^x+LeagueAv^x)
-
Rhys Stephens
commented
Hi Ken,
Has there been any code written for an alternative for pythagenpat etc? I've just uploaded my fantasy league for our new season (Australian Football) and am keen to use the fantasy formula if possible.
Cheers
-
Aryeh Levin commented
The formula I came up with that could work is (P/2A)^c, where P is the points scored (goals for) for that team, A is the average for the league, and c is a constant. I found 1 to be the best exponent, but let me know what you find.
-
Good Idea. So, I don't have a good formula, but in the mean time I'll add a way to say "use goals for" instead of goals delta or winning percentage.
