-
Notifications
You must be signed in to change notification settings - Fork 156
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Does setting min_exposure per player no longer work? #432
Comments
For some reason it doesn't work
…On Mon, Sep 23, 2024, 1:38 PM EdCarGoes ***@***.***> wrote:
I swear I've done this successfully many times before, but for whatever
reason setting min_exposure does not seem to work at all now. I've tested
with Site.DRAFTKINGS and Site.FANDUEL with Sport.FOOTBALL
Here's a snippet of my code:
player = optimizer.player_pool.get_player_by_name("Josh Allen")
print(player)
player.min_expossure = 0.5
lineups = list(optimizer.optimize(n=10,max_exposure=0.8))
It builds 10 lineups but Josh Allen is only in one of them.
The print(player) line returns "Josh Allen QB (BUF)", showing that it is
finding the player successfully.
The max_exposure for the whole thing is working as no players are in
lineups more than 8 times.
Am I not using this right or is there an issue somewhere?
—
Reply to this email directly, view it on GitHub
<#432>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ANBWO7U6K7HEUBBLRVGWT4LZYBGX3AVCNFSM6AAAAABOWSGKP6VHI2DSMVQWIX3LMV43ASLTON2WKOZSGU2DGMRVG43TKNA>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
In the rule.py
Let’s break down potential issues in your MinExposureRule class and go over
the logic step by step to understand why it might not be working as
expected.
Possible Issues and Potential Fixes
1. Tracking of min_exposure_players:
You are using a dictionary self.min_exposure_players to track how many
lineups each player should be forced into based on their minimum exposure
(player.min_exposure). However, after each lineup result, you're
decrementing this count and deleting the player when the count reaches 0.
Issue: Deleting players from self.min_exposure_players can make the list of
constraints incomplete, especially if you need to apply the rule over
multiple iterations.
Fix: You might want to store the original exposure counts in a separate
structure and avoid deleting players. Instead, you can simply skip players
whose minimum exposure has already been met.
Example Fix:
def apply_for_iteration(self, solver, result):
if not self.min_exposure_players:
return
if result:
for player in result:
if player not in self.min_exposure_players:
continue
self.min_exposure_players[player] -= 1
# Skip players with zero remaining exposure instead of deleting
them
self._create_constraints(solver)
2. Inconsistent positions Handling:
In the constructor, you initialize self.positions using
get_positions_for_optimizer(). You are passing None as the second argument,
which may not be intended.
Issue: Depending on how get_positions_for_optimizer() behaves, passing None
might result in incomplete or incorrect positions being returned.
Fix: Ensure that the second argument is correctly passed or handled within
the function.
Example:
self.positions = get_positions_for_optimizer(optimizer.settings.positions,
context)
3. Incorrect Calculation of total_force:
The line that calculates total_force is complex, and there could be a
logical issue.
Issue: The expression total_force = sum(self.min_exposure_players[p] for p
in min_exposure_players) - total_for_positions * (remaining_lineups - 1)
might be leading to incorrect results. Specifically, the subtraction of
total_for_positions * (remaining_lineups - 1) may not always yield a valid
constraint value.
Fix: Simplify the calculation to ensure that it reflects the correct
remaining exposures relative to the number of remaining lineups.
Example Fix:
total_force = sum(self.min_exposure_players[p] for p in
min_exposure_players)
total_force = min(total_force, total_for_positions) # Simplified to ensure
force is bounded
4. Handling of remaining_lineups:
Issue: The division ceil(total_force / remaining_lineups) can result in
unexpected constraints if the total_force is close to zero or if
remaining_lineups is incorrectly managed.
Fix: Make sure that remaining_lineups is always positive and that
total_force is correctly normalized for the remaining iterations.
5. Solver Constraints:
The way you're adding constraints to the solver could be problematic if the
calculation of total_force or variables is wrong.
Fix: Verify that the variables being passed to solver.add_constraint() are
correct and that the total_force is appropriately bounded for the solver.
Example of Simplified Constraint Creation:
def _create_constraints(self, solver: Solver) -> None:
remaining_lineups = self.context.remaining_lineups
for positions, total_for_positions in self.positions.items():
min_exposure_players = [p for p in self.min_exposure_players if
list_intersection(p.positions, positions)]
if not min_exposure_players:
continue
# Calculate the remaining force to apply the constraint
total_force = sum(self.min_exposure_players[p] for p in
min_exposure_players) - total_for_positions * (remaining_lineups - 1)
if total_force > 0:
variables = [self.players_dict[p] for p in min_exposure_players]
total_force = min(total_force, ceil(total_force /
remaining_lineups)) # Normalizing the force
solver.add_constraint(variables, None, SolverSign.GTE,
total_force)
for player, total_lineups in self.min_exposure_players.items():
if total_lineups >= remaining_lineups:
solver.add_constraint([self.players_dict[player]], None,
SolverSign.EQ, 1)
Summary of the Main Fixes:
1. Avoid deleting players from self.min_exposure_players. Instead, keep
track of players' exposures but skip players whose minimum exposure is
already met.
2. Ensure correct handling of positions in the constructor. Make sure the
get_positions_for_optimizer function receives the correct arguments.
3. Simplify and validate the calculation of total_force. Make sure the
logic of this constraint is sound and that the force is being normalized
correctly.
4. Check solver constraints. Verify that the constraints being passed are
consistent with the logic and the solver's expectations.
Test the updated logic to see if it addresses the issue you’re facing. If
you need further assistance, feel free to share more details about the
specific problem you're encountering, such as error messages or unexpected
results.
…On Tue, Sep 24, 2024, 12:50 AM joseph offen ***@***.***> wrote:
For some reason it doesn't work
On Mon, Sep 23, 2024, 1:38 PM EdCarGoes ***@***.***> wrote:
> I swear I've done this successfully many times before, but for whatever
> reason setting min_exposure does not seem to work at all now. I've tested
> with Site.DRAFTKINGS and Site.FANDUEL with Sport.FOOTBALL
>
> Here's a snippet of my code:
>
> player = optimizer.player_pool.get_player_by_name("Josh Allen")
> print(player)
> player.min_expossure = 0.5
> lineups = list(optimizer.optimize(n=10,max_exposure=0.8))
>
> It builds 10 lineups but Josh Allen is only in one of them.
> The print(player) line returns "Josh Allen QB (BUF)", showing that it is
> finding the player successfully.
> The max_exposure for the whole thing is working as no players are in
> lineups more than 8 times.
>
> Am I not using this right or is there an issue somewhere?
>
> —
> Reply to this email directly, view it on GitHub
> <#432>, or
> unsubscribe
> <https://github.com/notifications/unsubscribe-auth/ANBWO7U6K7HEUBBLRVGWT4LZYBGX3AVCNFSM6AAAAABOWSGKP6VHI2DSMVQWIX3LMV43ASLTON2WKOZSGU2DGMRVG43TKNA>
> .
> You are receiving this because you are subscribed to this thread.Message
> ID: ***@***.***>
>
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I swear I've done this successfully many times before, but for whatever reason setting min_exposure does not seem to work at all now. I've tested with Site.DRAFTKINGS and Site.FANDUEL with Sport.FOOTBALL
Here's a snippet of my code:
player = optimizer.player_pool.get_player_by_name("Josh Allen")
print(player)
player.min_expossure = 0.5
lineups = list(optimizer.optimize(n=10,max_exposure=0.8))
It builds 10 lineups but Josh Allen is only in one of them.
The print(player) line returns "Josh Allen QB (BUF)", showing that it is finding the player successfully.
The max_exposure for the whole thing is working as no players are in lineups more than 8 times.
Am I not using this right or is there an issue somewhere?
The text was updated successfully, but these errors were encountered: