Testing the seasonal goal competition idea – Preliminary remarks
by milivella
I would like to test the idea of switching to seasonal goal competitions. Here are some notes about said tests, if I will ever actually do them.
What tests:
- What would have happened if these were the rules since 2010? Pro: it is relatable. Con: our picks were optimized for a different aim (caps+goals in the whole career vs. only goals counted seasonly).
- Simulate what would have happened if virtual scouts had only picked forwards (among the ones who were actually picked). Who picks whom? Assign players to virtual scouts at random, or make some virtual scouts more able (proportionally to the difference in ability among real scouts). Pro: it should be more similar to what would actually happen.
Results that I would consider good: in the ten more “mature” seasons (2013-2023),
- …at least three times the winner was not the leader before the last round.
- …at least three different scouts get to win a season.
If these criteria are not satisfied, the game would not get more fun with this reform. I do not add criteria about how the game would reward ability, as I take it for granted that being able to pick players who will score many goals for the NT takes luck.
> Simulate what would have happened if virtual scouts had only picked forwards
This is clearly the better option, not only because it would be “more similar to what would actually happen”, but also because:
– it would allow me to simulate scenarios like “only 6 scouts can play simultaneously”;
– it would allow me to run many simulations to check the most frequent results, and the extreme ones.
> make some virtual scouts more able (proportionally to the difference in ability among real scouts).
Again, this is clearly the better option.
> …at least three times the winner was not the leader before the last round.
Of course, if I can simulate many different runs, this means “in the median run, at least three times [etc.]”.
> …at least three different scouts get to win a season.
This is not enough: a 8-1-1 distribution of wins would still be not fun. I would add the following condition: “no scout wins more than six seasons”.
LikeLike
Two more notes:
– Simulating players and their careers would be even more important (than simulating scouts and their picks), as it would allow to check the optimal number of national teams for the current scouts.
– Another criterion for the idea (of seasonal goal-scoring competitions) to be a success should be related to make the game fun _for everyone_. I do not what this criterion should be: every scout scores at least 1 point per international round? there are few ties in the rankings?
Talking about ties makes me think that, while I am thinking about distorting the game, I could also consider matches between the scouts…
LikeLike
Still thinking about this…
How to simulate: I think the sweet spot between easiness=speed of execution and power of results may be to simulate scouts who, starting in 2007, randomly pick the players that are in the Fantasy Scout database.
What to simulate: different combinations of number of scouts and number of eligible nations, from 2(scouts)-1(nation) to 20-8 (or even more scouts). By the way, this could allow to extrapolate the results for a higher N than we have in the date; e.g., if we see that the results of 2-1, 4-2, 8-4 and 16-8 are similar, we can extrapolate that the results of 22-11 would be similar, too.
Criteria for a good result: to the ones I have already mentioned, I would add:
– At least 1 goal scored by (the players of) each scout in each round, minimum 95% of the times.
– How different is the ranking of a season from the ranking of the previous season? This should be more general and significant than the criteria about winners I have mentioned so far. The problem is to determine what is the value I desire. The current rule are such that, given picking randomly scouts, the correlation between successive rankings should be 0. Intuitively, I would not like a correlation higher than 0.5. And I guess this could be the dealbreaker for the idea of the seasonal goal rankings.
LikeLike
Three ways to say the same thing:
– Can “checking the score is not fun” be a problem, if “you may have to wait years before starting scoring” is not a problem?
– Given that Fantasy Scout is about long-term prediction, isn’t it normal that the progress of the situation is gradual, i.e. not based on single events?
– Fantasy Scout is a series of long-term bets, and it is usual for the resolution of such bets to be gradual. (Counter-example: if I bet on “Democrats will win the 2028 US presidential election”, the bet is long-term, and before election night the bet could seem 80% won or 80% lost, but—as the 2016 election showed—only on election night I will know if I have won.)
LikeLike