Fantasy Scout - A necessary premise
milivella
explain the fundamental principle of my behavior in this reform.
Why should you be interested in my principles? Just because I'm one head,
i.e. a vote, like you all. I suppose that you don't want me to vote against
a proposal you like, and so that you are interested in my principles. Of
course, for the same reason I've to consider the opinions of all the other
scouts, to make the changes I desire happen.
Here I'm stating the point in the clearest form I can:
The reform will have to change how the *future* actions of the scouts will
be evaluated (otherwise, why should we reform?). But it *cannot* change the
way the past actions are evaluated.
Practically:
We all want (if I'm not wrong: anyway, I want for sure) a system that
reward scouts that are not only good, but also constant. But this apply to
the future only: scouts that have been good but not constant should not be
penalized.
How to know whether someone is penalized? Use the following rule of thumb:
A and B have picked some players so far. They won't pick players any more,
'til the end of the times. The winner should be the same *both* under the
old rules *and* after the reform.
If it's not so, that reform is IMHO unfair, so I'll vote against it. Stop.
...
I think that there is another point that I wasn't able to explain.
1. *It doesn't matter* why a scout chose to pick 3 or 4 or 5 players. The
only way to evaluate his actions are the rules under which he chose to do
so. Their eventual agreement with the "spirit" of the game can't be used to
evaluate them.
2. In any case, nobody stated what the spirit of the game is. The only
person who was interested enough to write something about the features of
the game was me:
http://fantasyscout.altervista.org/features.txt
Of course this was a personal opinion, but it's the only opinion so far, so
let's read it:
"If you have picked a very good player, you should keep watching for
players, but you should be able to not call good players, if they are not
as good as your first player."
So the spirit of the game is "know when it's better to abstain", not "pick
every good player you see".
3. More: you aren't inside other people's minds, so you can't know why they
picked few or many players. So, even if it mattered, nobody could know the
truth.
4. But, even if you would be able to know the truth, the truth is probably
that the scouts who picked few players differed from the players who picked
a lot of players only in strategy. Let me use an example that I know well,
and about which I can be sure: Alberto (but he is only an example) and me.
Once Alberto phoned me and told me: "I saw a young player. His name is
Balotelli. I think he's good, very good. But not as good as Banega, Bojan
and Lavezzi. So I don't call him in Fantasy Scout, because, under the
current rules, he would lower my score." This happened before anyone picked
Balotelli. The same is true of Giovinco, Acquafresca, Maggio, Walcott,
Montolivo, Santacroce, Clichy and others: he identified them as good
players before anyone picked them. He just thought that they were not as
good as the players he had already picked, so he did what is sensible under
the current rules: he didn't pick them. [Under the current rules, if he was
right about these relative judgment (Lavezzi better than Balotelli, etc.),
he'll win; if he was wrong, he'll lose.]
Now me: I've picked 17 players just because I wasn't sure (at the moment of
the pick) about any of them. If I was sure that Gago was a crack, I would
have picked no more players. I would keep watching around for prospects,
but I would have not picked them, just like Alberto did.
So: you have the scout (Alberto) who picked less players (3), and the one
(me) who picked more players (17). And the difference between them is just
that the former was always sure (being right or wrong, it doesn't matter)
about the value of the players he picked, while the latter wasn't sure
about any single player he picked (I just knew that some of them were good,
but not sure about who was good and who wasn't). There is no other
difference: one was sure about his picks, the other wasn't. They play with
same constancy and with the same spirit: still, they have picked 3 vs. 17
players.
...
I hope not to offend anyone.
If my effort to be clear wasn't enough, please ask me to rephrase my
thoughts. Thanks.
--
Cheers
milivella
Futbolmetrix
news:19lmi7ys2c67e.e...@40tude.net...
>
> 4. But, even if you would be able to know the truth, the truth is probably
> that the scouts who picked few players differed from the players who
> picked
> a lot of players only in strategy. Let me use an example that I know well,
> and about which I can be sure: Alberto (but he is only an example) and me.
>
> Once Alberto phoned me and told me: "I saw a young player. His name is
> Balotelli. I think he's good, very good. But not as good as Banega, Bojan
> and Lavezzi. So I don't call him in Fantasy Scout, because, under the
> current rules, he would lower my score."
But this example illustrates exactly why any change in the rules is in some
sense unfair. Even if you reweight Alberto's *actual* choices so that his
score (after the reform) is the same of that an equivalent scout who picked
more players, you can never go back and correct the fact that Alberto chose
not to pick certain players, even though he had identified them as very good
(but not as good as the ones he already had).
D
milivella
No. Alberto picked 3 players because he thought they have a very good
expected score, let's say 100. Then he didn't picked 10 other players from
which he expected 90 (still good, but not good enough to make his average
higher). Now, let's assume that these players were all picked by
Antialberto.
If Alberto is right (i.e. the players he picked were better than the
players he haven't picked) what will happen is:
- Average: Alberto 100-90 Antialberto
- Simple switch to sum: Alberto 300-900 Antialberto
- Switch to sum through a multiplier 10/n: Alberto 1000-900 Antialberto
So, through the use of the multiplier, he can switch to sum, and still has
the same position vs. Antialberto: Alberto will win if correct about
players' expected score, will lose if mistaken.
So *there is no need* to "correct the fact that Alberto chose not to pick
certain players, even though he had identified them as very good (but not
as good as the ones he already had)".
Am I wrong?
--
Cheers
milivella
Futbolmetrix
>
> Am I wrong?
We're just arguing in parallel. I think your reweighting system has flaws,
which doesn't necessarily mean that I'll vote against it, since any system
that we adopt to make the transition will have some flaws.
Example 1:
There are three players, who all started playing under the old rules. These
are the players they picked.
Player A: 100, 100, 100
Player B: 0, 90
Player C: 10 players, all 0.
Now you reweight their scores, and the ranking is:
A: 1000
B: 450
C: 0
But suppose instead that the 90-point player had been initially spotted by
A, but he chose not to pick him because it would have lowered his score
under the old rules, therefore leaving him available to player B.
If A had known about the upcoming change of the rules and picked that
player, things would have looked as follows:
A: 100, 100, 100, 90
B: 0
C: 10 players all with zero
After reweighting:
A: 975
B: 0
C: 0
Unfair to player A.
But here's a more damning example (which was my original point about the
scale issue).
Player A: 100
Player B: 90
Player C: 10 players with 0
You do the reweighting, and the scores are
Player A: 1000
Player B: 900
Player C: 0
But suppose instead that the night before the reweighting, player C decides
to add 10 players to his roster. So the multiplier is 20 instead of 10.
Scores after reweighting:
Player A: 2000
Player B: 1800
Player C: 0
So now player B has to make up a bigger gap with respect to A, even though
absolutely nothing has changed in their behavior.
D
milivella
> "milivella" <mili...@gmail.com> wrote in message
> news:ss5nrd5ja0qc$.x8mk5s3dcb6p$.dlg@40tude.net...
>>
>> Am I wrong?
>
> We're just arguing in parallel. I think your reweighting system has flaws,
> which doesn't necessarily mean that I'll vote against it, since any system
> that we adopt to make the transition will have some flaws.
Ok, and thanks for such a sensible reply. But I think that "my" system has
no flaw, so I'm trying to persuade you - but also, and in the same breath,
I'm trying to be persuaded by you, in the case that your statements are
true in my opinion (you know that it has happened before!) -.
This said, maybe we won't reach perfect agreement ("there are flaws!", "no
there aren't!"), and you are right: we all should vote "yes" to any
proposal that is not really bad and/or worse than the current rule, and
then express our preference among these "not bad" proposals. I for one will
behave so: I will reject only proposals that are IMHO unfair (for my
definition + rule of thumb of unfairness, see my original post). Indeed, I
see two very good proposals (restart+top 10 and multiplier+sum), but each
of them has pros and cons: there is not IMHO a perfect proposals (at least,
not so far), there are just some proposals that are better than others.
> If A had known about the upcoming change of the rules and picked that
> player
Very special case: for years, they played with given rules, then they'll
play with new rules for years again; what you are suggesting can happen
only in the few weeks of the reform process.
Anyway, A can't be sure about the upcoming change: one vote against the
change by anyone, and they're back to average, and he has lost points. Of
course, it could be worth the risk. But this is one of the reasons because,
IMHO, proposals should be submitted as soon as it's possible. Remember: you
and Benny asked to not vote before some debate!
Easy solution: ban picks when a reform is considered.
(Anyway, I think that we all assume fair play in all the other scouts. But,
I agree: a system where you can't play in a unfair way is better.)
> But here's a more damning example (which was my original point about the
> scale issue).
>
> Player A: 100
> Player B: 90
> Player C: 10 players with 0
>
>
> You do the reweighting, and the scores are
> Player A: 1000
> Player B: 900
> Player C: 0
>
>
> But suppose instead that the night before the reweighting, player C decides
> to add 10 players to his roster. So the multiplier is 20 instead of 10.
>
> Scores after reweighting:
> Player A: 2000
> Player B: 1800
> Player C: 0
>
> So now player B has to make up a bigger gap with respect to A, even though
> absolutely nothing has changed in their behavior.
Now I understand what you meant. But let me ask you: do you see an overall
fairness in both the scenarios (because C has to pick more good players
after the reform, in the second case), even if not everything is perfect?
Or is your criticism stronger?
Anyway, let me reply to your example. Even if A's and B's behavior has not
changed, the context has. I.e. now they know that "pick 20 players" is a
strategy that someone find sensible under the old (score=average) rule.
Hadn't C picked those players, they could have only known that the bravest
scout can pick 10 players (maybe more, but they can't be sure). So they
have different knowledge in the two scenario. But, in different contexts,
the same behavior can have different meanings: so it's fair to me that the
gap is bigger in the second scenario.
In the first scenario, A and B were less "greedy" than in the second. Yes,
they've done exactly the same thing, but the context is different. Just
like tipping a waiter 10$ can be lavish in a context, but greedy in another
one.
--
Cheers
milivella
milivella
> Futbolmetrix:
More fitting example: A tips 10$, B tips 1 cent. In your ordinary pub, A is
lavish, B is greedy. At the Millionaires Club, they both are greedy.
Same behavior, but in different contexts -> A and B should be treated in
different ways in the two scenarios.
--
Cheers
milivella
milivella
> But here's a more damning example (which was my original point about the
> scale issue).
>
> Player A: 100
> Player B: 90
> Player C: 10 players with 0
>
>
> You do the reweighting, and the scores are
> Player A: 1000
> Player B: 900
> Player C: 0
>
>
> But suppose instead that the night before the reweighting, player C decides
> to add 10 players to his roster. So the multiplier is 20 instead of 10.
>
> Scores after reweighting:
> Player A: 2000
> Player B: 1800
> Player C: 0
>
> So now player B has to make up a bigger gap with respect to A, even though
> absolutely nothing has changed in their behavior.
>
> D
Solution: forget division
max_actual_number_of_players / your_number_of_players
and choose instead:
max_possible_number_of players / your_number_of_players
I.e., in our case: no scout could have more than 32 players (16 picked in
2007 + 16 picked in 2008); so multiply players' score for 32/n, where n is
the number of the players his scout has picked so far.
E.g. A player owned by Alberto will have his score multiplied by 32/3,
while a player owned by me will have his score multiplied by 32/17.
In the example you provided, both max_possible_number_of players and
your_number_of_players will be the same in both scenarios, so the result of
the reweighting too will be the same.
I bet that you prefer this reply to the previous one, don't you? ;) (Still,
the former flows from the latter: if everything is relative to context,
just create an absolute context.) It's like adding Agnostic, a virtual
scout who picks the max number of players he is allowed to pick, starting
from 100 years old people.
Of course, I'm taking note of the new form of the proposal (32/n), that
supersedes the previous one (17/n).
--
Cheers
milivella