This forum is archived and no longer active.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

Topic: Blank optimizer with the ability to add your own players

1 Topic by bjgoetze86
I've read a few comments from the FC people saying that CFB/NCAAB/etc. would be difficult to make an optimizer for because of the high yearly turnover rate etc. So... take some of the load off of the shoulders of the people at FC and let the users have more power. How about a blank optimizer that initially has no players/teams associated with it. Let the user upload their own .csv file that's imported directly from DK for each particular contest, then add their own projections, run the "cruncher" to create the lineups, then export them to a csv. DK allows lineup uploads via csv (obviously) so just let the user do it. This way the optimizer can be used for all sports. I've yet to find a commercial website with an optimizer that has this type of flexibility. I personally feel that the first optimizer to do this would take a huge leap over the others and become the obvious "go-to" for most players looking to create multiple lineups. This isn't meant to be a knock on your projections, but the real value of FC in my opinion is the "cruncher" and more flexibility with it can only be a good thing.
And honestly, I'd be more than willing to do the daily data entry for lineups for the other available DK sports for free, especially the eSports ones.
This is a great idea.
I like it! We may make a section entitled "Blank Optimizers". Not sure how quickly we out it out, but I'm adding it to the to do list.
Very good idea!!!

Add the ability to set parameters, # of players, max salary, etc and you have a winner.
Awesome. I think your optimizer overall is the best out there atm, even better than the one at RotoGrinders. I think adding full flexibility with the current algo and UI that you guys use would be an amazing product for the DFS community. Even though the guys at fantasylabs have some nice modeling tools, you can't do anything besides create one lineup with your models. I think this kind of functionality would vault you guys to the top.
Also, as a side note, maybe a minimum exposure setting would be a good idea too. I find that I have to tweak the projection number and run a bunch of batches of lineups until I get the proper exposure to a player that I want. This can be a little tedious. Since a large part of your product seems to be creating multiple lineups, I would like to point at that a player's projection isn't necessarily the most pertinent factor when you're creating 200 lineups. From my experience, a lot of the players that do this will have a core of X amount of players, then they try to get certain percentages of exposure to certain players, but the way the algo decides the lineups seems to be focused mainly on projections. I don't necessarily want the highest projected players to fill out the remaining spots around my core, I want exposure to the "4k second highest targeted receiver on team X" to fill in my other spots. Just a thought. Thanks for reading!
Just to clarify, the higher projected players tend to have higher ownership %'s, and to win a GPP you typically want a number of players with low ownership %.
This is a great idea
Allow me to step in and ruin the fun a bit. Just a bit. While it is a very powerful and flexible tool, the downside is that it will be very difficult for us to troubleshoot your issues or debug an issue if it's on our end. It's going to take a lot of planning to build a tool that is as fool-proof as possible. So the development may take a while. So please be understanding and patient, knowing our history, the first release will have a lot of bugs, and you guys will essentially be bug testing for a while.

I've heard a bunch of cool things, (like csv import), but if you'd like to give us some sort of mock up that we can work off of. It may accelerate the process ;), just saying.
How do you guys get each sport done every day? do you pull the data from somewhere or just manually enter it every day? I'm trying to get creative here to make things as easy as possible, and kind of thinking out loud. If the "Player Name" and "Salary" columns were editable like the "My Proj" & "Exposure" columns were, I could start spitting out lineups for other sports immediately, but then again I think I might think about lineup construction a little differently than most. I would set all the projection numbers equal to the salaries to kind of fool the algorithm into treating everyone equally (I already do this when using the tool now anyway, and it definitely works), then decide which players I want by checking the box, locking, or reducing exposure to get the types of lineups I want. If the idea is to get away from using projections, and building a tool that's flexible enough to use for every sport, then projections really aren't necessary, and different types of limiting factors could be used to build the desired lineups. In fact, NOT using projections gives the power to get exact (or really close to) exposure to the desired players.

I would imagine it's not overly complicated to take each result and have it spit it out into a .csv file, if you haven't checked out RG's tool on exporting to .csv then uploading L/U's to DK, you might want to check it out to give you an idea of how this could be done. They use a chrome plugin as well that adds a button on the DK "My Lineups" page where you select a contest and upload the csv"

Unfortunately I'm not a programmer, so I'm naive when it comes to the complexity of tasks like this.... but if it were me, and I were designing something like this I would probably first attempt to do things like this...

1) Make every column editable
2) Add an "Add New Player" button
3) Under "advanced options" make the team names in the "Players Per Team" editable
4) Make the created lineups spit out into a .csv (I have example ones if you need for any reason, but I doubt it)

The importing lineups via DK's .csv's that they give you is just a time-saver, I don't think that would be the highest priority.

I know this wouldn't be easy, but for an early "alpha" stage, maybe you can take a few shortcuts and get something working and play around with it and use your own creativity to make it better. Then again, maybe this isn't the direction you guys are interested in going in. I'm very happy with what you guys already have, and since I can adjust the projections and trick the algorithm I've really been able to build the kind of lineups I want very easily for the sports you guys already offer. I was just proposing this idea to both save you guys time in the long run and to create a more powerful lineup creation tool.
Obviously there would be a lot of issues if the users incorrectly enter a field, and a lot of new issues would arise, but I really think you could get something working fairly easily by copying the NFL page for instance (I know it's not that easy to just simply copy it, but you know what I mean) and then make the "Player Name" and "Salary" columns editable.

If you wanted to create CFB lineups for this week for example, you'd essentially have to assign players on each college team to an NFL team so that the "Players Per Team" function doesn't mess things up, but logically I think just editing the two columns and setting all projections equal to salaraies are the only two things necessary to start spitting out lineups for other sports.
I'm not sure how interested you guys are in this at the moment, but if you'd like to sit down and talk for a bit on skype or teamspeak I'd be willing to do that.
We understand that you'd essentially import (player, position, salary, team) at the minimum. But you'd also need to import when each team plays, the positions for each different sport (remember people are asking to use this on every sport i.e: mma, esports) what the flex positions are. How do we do team stacking and stuff like that. I'm not even sure if that's all the essentials to be honest. But a REALLY important factor is: how do we present this in an organized way so that it's usable for all the users that want to use it?

You can't simply just make every column editable even with something like NFL to NCAAF, because there are WAAAAAAY more teams in NCAAF. So you can't assign 1 to 1.
I really don't think you'd need to import when each team plays. The outgoing csv file with your teams only needs player names. The contest/sport/time selection can be done when uploading the lineups to DK. I don't know if you've looked at other lineup uploading methods, but other places do theirs like this...



The position part was a total oversight by me, I was thinking about my excel solver and how I don't have a problem with that, but then of course all of the positions are defined. Yeah, that would be a big issue. The only thing I can think of is for the user to have to define which positions are necessitated by DK for the contest of choice, where QB/RB/RB/WR/WR for example would be a/b/b/c/c, but for baseball it might be SP/C/1B/2B/3B or a/b/c/d/e. The cruncher of course would need to know that it needs to fit x amount of a's/b's/c's and the order they need to be in. I guess this could be done, but at this point it might just be easier to create an optimizer for each sport since you guys already have a good thing going.

Team stacking in my opinion could be handled more easily with a minimum exposure function for most sports, although if you're trying to stack a bunch of combinations of 9 batters in baseball, then it'd be more difficult. Football/NHL/PGA/eSports/NBA/NCAAB I don't think you'd have a problem as you're probably locking in the QB, and might want 70% minimum exposure to WR1 and 30% to WR2, or whatever other combination you want. I don't think any issues should arise with accidentally putting the wrong team defense against your stack, or putting a pitcher that's playing against your batter stack, because the user should be excluding those players anyway.

Presenting it in an organized manner could definitely be done, and I think a single optimizer for everything could still be done, but as I said earlier, it might be a bit too much work, and you guys already have a good thing going.

I don't see why the amount of teams matters, because in any given NCAAF slate there's about the same number of games that week as there are in the NFL. I think things could be made to where the team someone plays for is irrelevant. I personally approach my L/U construction by excluding all players from the start, selecting the ones I want, then assigning their exposure. So I'm never going to have an opposing D or something accidentally on my lineups. But I can see how teams are relevant and could pose an issue if you're taking the "I like the patriots, now gimme your best players with them" approach.

I haven't spent a ton of time thinking about any of this, so some of it's probably just flat out wrong, so take it for what you will.
For the "position" thing, make a popup box open when you open the optimizer that asks you to select the website, then select the sport. I mean, we all know the positioning format for all the fantasy sites, so have those parameters set already once the player gets to the lineup creation page.
There were requests for a skeleton tool months ago......we asked for this back around the same time as the 1st milli PGA contest. This would be a great tool to have so that users can import their own data.....but I do understand the issues and bugs. I would vote to have it built as simple as possible....and then let the users test it out so that we can enhance it in the future.

Simple dropdowns to choose the Sport....that would flex in the available positions would be the best start
I will be honest guys, and I hate to burst your creative thought processes here, but I am 99% sure this will never happen. There are a lot more moving parts than I think you guys realize, and it would actually be less work to keep adding more sports. The other reason this will not happen is you guys are all very advanced users, but that does not truly represent our user base as a whole. If I were to spend the time developing this, I might as well just add a new sport and then all of our users can benefit from it.

We will be adding more sports, and I am actually in the process of hiring more development help, so hopefully we can get it done sooner than later. I appreciate all the good ideas, and honestly I mulled this over for a bit in my head, but I really cant see it ever happening.
Fair enough, thanks for taking the time Dave.
so a add player feature would be impossible pretty much as well?