Greater Care with Salaries
in progress
jayk123
The current algorithm does a good job when it has a decent sample to work with, but since the takeover it's been drastically mispricing players who have played 0-1 game this season. e.g., tomorrow, anthony miles and jasper pittard (both veteran players and proven fantasy scorers) are both significantly underpriced because they've played 0 and 1 game this year. at the other end of the scale, paul ahern has jumped from 5k to 12k based on one good game. Greater subtlety in the algo or some proofreading style editing is needed in salary changes for players with a small number of games.
Draftstars
in progress
L
Lampy
Draftstars Your AFL salaries are a fucking joke. One look at the ownership %s yesterday and it's clear this needs to be made an immediate priority. There's going to be a few blokes today over 50% owned again, which is the sort of careless attitude that we've come to expect from MoneyBall, not DS. Do what Jay suggested and at least review your database before pushing your salaries live. Heck, I'll do it for you. Any player who has played 0 games in 2018 needs to use 2017 average when determining starting salary. Ignore games where players have played <60% gametime (injured).
Draftstars
Lampy: sorry you feel let down with this. It's an area we are particular keen to improve and will be investing considerable time into. We have some ideas we will start to implement and gauge how we are tracking.
I
Instorgata
Lampy: yerp sick of choosing players as a block..im trying to make teams then 20 in I'm like oh shit this team dosent have this 5k player. its border line impossible to fade a player at 5k who Ave 60-70 in the past. even with the 5k players who have never played a game..fuck you only choose them in hope..some pump out 90's some get 30. yeah i know all about research, but 90% of players just plug them blind so he ends up 40% and when they jag an 80-90 jeez your a bit behind. if they all come in atleast at 6500, its not so much a kick in the stomach.
L
Liam Fariss
Draftstars: just after some clarification with the algorithm. How can someone like Rory Sloane go down in price after a 110? Same goes for Scott Selwood after an 85 drops to $9.6k
Draftstars
Liam Fariss: we have made some adjustments to the AFL algorithm and it will take a few rounds to balance out. We believe the new salaries will be more consistent with this adjustment in the long term.
Draftstars
under review
H
Hedge
100% agree jayk. Ahern and Miles are perfect examples of how the current system can be improved on both ends. Miles is no rookie and has been an emergency for a few weeks now. You dont want an established afl player (who only is not getting games because he cant crack the premiership midfield) to come in at 5k and be 75% owned (or higher!). no one wants that. that ideally would have been picked up by someone and he would have started at a 7.5k or 8k price point.
Ahern is the other end. he hasnt played for 3 years and came in and junked up against a Lions side who were down on runners for a 92 at 5k. you just cant put him to 12.5k the week after. someone there needs to make the call on these players second and third up, look at the opponent and game situation and adjust manually if need be. at 12.5k almost no one will touch him. at 10 or 10.5k every single coach would have looked athim, and thats bwhat you want - you want every coach to consider every player. thats the sign of a good cap.
I really do hope that we can move away from any ownership being part of salaries. i dont know that they are but it is the only way to explain some of these salary flucs. tim smith, spargo, mihoceck, guelfi - every rookie who has come in and gone half well has skyrocketed to be almost unplayable the week after.
Build them up to tempting range again and let coaches make the call on the risk/reward for second up games. then once they have 3 games played they shouldnt fluctuate as much.
so to recap this stream of consciousness type comment;
- established players returning and debutantes or kids who have barely played any games would be treated differently in terms of their base price first up.
- both groups second and third up should be manually adjusted so as not to be priced out of selection.
- player ownsership should not be a factor in determining salary. ever.
jayk hit the nail on the head with one word. subtlety. it needs to be someones job to be onto this each week prior to the contests going live. controlling the quality of this one thing has such a trickle down effect on team selection, live game experience, everything.
cheers for the feedback site DS, im loving the chance to contribute to a better product.
Knibb High Football Rules!
jayk123
Hedge: i reckon the issue is that it's pricing players solely based a combination of their avg from this year/last x games from this year, no matter how many games they've played this year (or any other year). so ahern is priced straight up as a 92 avg guy, the same as if he'd played 10 games. miles is priced as a guy that's never played before. i don't think ownership is part of the formula atm.
back in the day the DS algo used to look
too much
at past history, so prices didn't move quickly enough. they did a good job of correcting that, and there's been better movement, but now they have an issue where the formula is too simple, which is causing the massive issues you mentioned in your points 1 and 2. all you'd really need to do would be to 1) flag all players that have played 0 games this year, and check if they have past history to factor in (clearly you should price experienced players that haven't been playing based on last year's avg rather than pricing them as a rookie)
and 2) modify the formula for players that have played less than 3 games in their career so that the prices change more gradually. from 3 games on, it's fine for them to be priced the same way other players are.