This forum is archived and no longer active. You can visit us in our Discord Server here!

Topic: Filter based on projection range

1 Topic by Reboshua
For GPP purposes, I'd like to be able to generate lineups based on a certain range of expected outcomes. Maybe I want the Top 20 projected lineups, or perhaps sometimes I'll want lineups projected to be in the range of 230-250 to see what they show as possible contrary plays.
To further illustrate why this is important, see the attached image. No Top team has had a projected value greater than 258.1 in the small sample I grabbed. It would be nice to be able to force a range of lineups generated that were lower than "top projected." That combined with some other things (locking in core guys, limiting exposure, etc) will help get closer to GPP success.

I may go ahead and add this into the advanced options, since it wouldn't be that difficult to do. You would simply set the max score, so if for example the top projected lineup was 280.5, you could set a value of max score - 260 and you would get lineups that were projected worse. I am not sure I personally see the value in this, but if you guys want it I am willing to make it happen.
The issue is that when you get toward the lower projected scores, there are SO MANY COMBINATIONS (my guess is in the thousands) to make that score. My bet on a decent sized slate is that if you set your max score to 250 or 260, and ask for 20 lineups, you will see 20 lineups, all extremely different that will get to 260 exactly, then to choose from those, you're going to need to lock several players, by that time, you might as well be entering lineups manually.
This is very interesting Data however I cant see how this could be useful due to the amount of lineups and combos that this would generate. On any given night there is no way that cruncher can determine DONT start Chris Paul start some random backup who goes off due to foul trouble or injury.
Doing it allows you to then start to see how many lineups are possible for that range. FantasyCruncher projections already limit the universe of guys to crunch, and I'm willing to bet there are far fewer permutations between 235-250 than you think. Tonight's best lineup was $234.90, BTW.

Once the major sights allow API access to easily upload lineups (and I guarantee it's coming), I think a lot of people might try and grab a handful of lineups in this range.

This, combined with my other suggestion of allowing me to limit my exposure to a guy would further limit the number of unique linueps. Unclear if this helps or hurts the goal of the "perfect lineup."
Reboshua, you really think FD and/or DK will implement auto-upload/update of lineups? I'd be really surprised, and I'm not sure it would be good for the market. Might be just like online poker, where some of these tools eventually just make it harder to find an edge.

Still, I would LOVE to see more features in the lineup-generator. At the top of my own wish list is the ability to "like" multiple players that will be preferred by the lineup generator. Sometimes, especially on a large-slate night, I like a half dozen PGs and several Cs, what I'd really like is to be able to "like" 20 players and know that none of my "liked" players are being overlooked. Ideally this would incorporate "locked" players as well, so for example I could lock one PG and have FC distribute the other PGs that I've "liked" throughout the lineups.

I realize that I can lock one C, generate a few lineups, then lock another C, etc. But when I have 4-6 preferred players PER POSITION it's very difficult to generate lineups that don't exclude players. I realize this might result in a lineup that's NOT generated by projections, floor, ceiling, or average ... but instead is generated by primarily by simply distributing "liked" and "locked" players given salary and/or points restrictions of the user. But for those of us who use FC to generate GPP lineups, this would be perfectly fine.

Please, please, consider adding this feature! :)
THis is a good idea. The best lineups are not at the top or the bottom. They are somewhere in the middle Being able to eliminate the extremely high projections and extremely low projections will give you a pool of lineups with more winning combinations.

Of course uploading custom projections or tweaking them as I do helps alot :)
"Might be just like online poker, where some of these tools eventually just make it harder to find an edge."

This is exactly why they'll do it. One of the biggest problems people have is getting all of the lineups they want to put into a GPP into it. The sites have an incentive to allow as many lineups to make it into their GPPs as possible, as quickly as possible. More money, bigger prize pools, etc.

They also have an incentive to help the people who are currently losing to win more often. At the Fantasy Sports trade show in Vegas, a presentation was given by the Draft Kings CEO where he shared that only 20% of the players are profitable on a weekly basis. This isn't good for the industry. Allowing an API to upload lineups from sites that have better analytical tools will achieve two things:

1. More lineups
2. Hopefully better performance by the 80% not performing well

Here is the link to the DraftKings presentation:
https://www.youtube.com/watch?v=0nuaKXoiqf0
"Still, I would LOVE to see more features in the lineup-generator. At the top of my own wish list is the ability to "like" multiple players that will be preferred by the lineup generator."

This is a recurring theme and I suspect the Cruncher guys are listening. RotoGrinders has a way to click a heart and say that you "Like" a player. It's just unclear to me how that affects the algo. My suggestion on several other boards is to allow for the user to specify an "exposure %", both at the global level and for individual players. I may like 6 cheap point guards tonight. If I can't limit my exposure to say, 30 % per player, then all of the lineups generated will default to the two with the highest projected value number, never incorporating the other 4.

Allowing for an exposure field would solve this problem. I'd keep the default experience the way it is now, to not be confusing to folks, but adding this will definitely improve the way lineups are constructed.
I love the global exposure % idea with an override on a player-level. Don't get me wrong, I love all 4 current projection models, but an additional model that would rely more on "preferred" players to ensure a desired distribution would be a HUGELY beneficial feature to us GPP players.

Regarding the automation of lineup entry, let me ask you this: (before Black Friday) did the various poker tools benefit the 80% of losing players? No. No way. I'm guessing it only increased the % of losing players. And similarly, there is NO WAY that automated lineup entry would benefit the 80% of losing players. In fact, I could see the reporting of this feature driving away your more casual players. "Pros using bots to submit all of their winning lineups!!!" I agree that the 80% figure is bad for the industry. Although it's not terrible. Hell, it's more fun than a scratch-off ticket and better odds, too. Probably better odds than a casino as well, dollar for dollar in GPPs anyway. But IMO automation will be even worse for the industry. If they really want to improve that number, they'll have to look at the rake. They probably won't, but hell at least give us better rewards to bring down the effective rake.

I'm not a high roller, but if/when this becomes a reality let's have a fun wager on which direction that 80% figure moves over the course of one year. :)
Speed, aka "time," is one of the major sources of productivity and growth in any industry. Population increase is the other. Condia does so well because he/it can put in more time than the average player. Tools that save us time are negative EV for Condia. Better analytics are negative EV for Condia. Now, when it gets to that point, someone like Condia exits because they won't play if they don't possess an advantage. Unclear what happens to the liquidity in the marketplace when that happens, but I suspect it will happen in the next few years.