• terrymandangle

    I have been hearing alot of people referring to “their projections”. How do you come up with your own projections?

  • tianyihe

    @dstars5 said...

    Seth – You mention in there that a more elaborate projection system would include a variety of other factors. How exactly would you include these? Would you follow a similar process for each factor (DvP, Carries, Vegas Totals, etc) and average them?

    Do not average them. That’s the cop out way of running data analysis and is always wrong. You need to have a correct loading factor for each variable and sometimes, your model might not be linear. In fact, when I first ran a linear model to predict FTPS, it became pretty clear that the linear model is inappropriate because I always underprojected top end guys and overprojected the fringe guys.

    Creating a system where you predict fantasy points is too complex. If you can predict changes in minutes played for upcoming games and the associated change in usage, you pretty much have everything you need to identify the players you want to play. Then you plug in stuff like DvP, vegas over/under, standard deviation of a player’s fantasy output, etc.

  • Tele21

    After reading a lot of Jonathan bales in my quest for knowledge that allows me to eventually become a dfs pro…its my understanding that most pros started making their own models….even in the incentive pkg I have the only ways I’m able to modify are pro filters….a few of which are pros who I don’t respect their opinion as much as others…i curious do you use your own model and projections as a pro or do you use rotogrinders as your sole provider for projection. If you have the time as a player would you say there are more/less benefits to making your own projections. And since you are behind the scenes is there anyway to see the sources that RG uses for their projections or the make up of their models? I ask bc in review I find that some (not most ) of the projection highly differ from other sites when it comes to good plays that w9uld normally be underowned and turn out to be contrary to what the RG projection may be coincidence but if it weren’t often enough to bring it up I owuldnt. I love RG and their premium stuff I use it along with other sources but I’m feeling like not only is it cheaper to create my own model but I’m able to trust myself to not tilt the opinion of the public… (not saying RG or any site does)

  • billyelflaco

    Exactly. Thank you. Either get fish from someone, or learn to fish. I’m not one to be fed by others. Great job with this reply.

  • brianasian

    @huitcinq said...

    This stuff is like alchemy. There will never be a perfect projection system. Sometimes a simple vegas based projection system will be more accurate than a full PFF one (using individual player ratings based on watching every player on single play of the season) or Numberfire (who I believe lean heavily on body types and speed and how they correlate to historical players with comparable skills and build in addition to a players matchup and performance).

    You can literally use anything you can think of as long as you can get your hands on some data. For PGA I looked into building a player rating system based on how they perform in rainy weather and using it to project tournaments where the weather is poor. But that became overkill real quick.

    hey would you be able to walk me through how to make projections through email?

  • mtdurham

    @sethayates said...

    Serious question here. Why would you want to create your own projections? It’s quite a bit of work. Why wouldn’t you just subscribe to RG incentives or one of the other sites that already offers projections? If you buy RG incetnives you get the projections as well as the lineup builder. I highly doubt anyone who doesn’t have an advanced knowledge of statistics (the math discipline, not looking at box scores) would be able to build anything better than what you can already get here for relatively cheap. I think incentives is $100 for the whole season correct? The NBA runs from Nov-April. So you are getting 7 months worth of projections for $100. Divide that by the month and you are paying about $14 per month. Divide that by the day and you are paying just under $0.50 per day. Creating your own projection system is going to involve at least two hours per day just updating stats and expected minutes. I can’t see how that is worth it unless you simply enjoy doing the work yourself and also have the free time.

    A SOLID case can be made for creating your own projections OR using others projections…..

    Think of it like investing in the stock market…. one can take a Macro or a Micro approach…. Macro you analyze the Economic landscape as a whole…you look at interest rate forecasts, GDP Growth, political climate, etc and select industries you think will do well… or you buy stocks/sell bonds…. or you buy calls/sell puts and try to make money off the underlying volatiltiy of markets….

    With a Micro approach you look under the hood at individual companies… you see who has the best line of product development, look for core growth opportunities, good/bad management, etc.

    Me? I’m a Macro DFS’er. I’d rather take your projections at face value and assume they are mostly correct. Then I use scenario analysis and game theory to try to gain an advantage. Oh Player A is projected to score 0.3 points more than Player B but be owned twice as heavily? Give me Player B. Oh this team is projected to have a huge game? Who are the ancillary guys who may benefit whose picture isn’t on the frontpage of RotoGrinders right now?

    Others might be Micro DFS’ers….. They think they can tweak a vanilla projection system that is backtesting and overfitting regression lines…. and project forwards instead of backwards….

    I can tell you this…. the most accurate projections in the world quickly become the least profitable projections in the world if EVERYONE is using them. I don’t want to split a GPP with 186,000 other people using the same lineup.

    I’d rather my projections be relatively inaccurate than frequently replicated.

    Besides all that…. most people do not have enough understanding of basic statistics to recognize the relationship between explanatory variables and their predictive capabilities. They are so busy trying to explain the past they forget we make money by trying to predict the future.

    Some people will read this and understand it and apply it. And others will not. And that cognitive dissonance will keep this industry intact and give one of us out there a chance to become a Millionaire every week for the rest of their life…

  • leokev21

    @colinwdrew said...

    I create my own projections for NHL and would be happy to talk you through the process I use over email. It is more suited for a blog than forum though. I think it is the single thing that has improved my DFS game the most this year and would highly recommend it. Especially for a sport like hockey where fewer resources are out there.

    It isn’t just about the numbers, going through the process teaches you to think critically about the game and more instinctively adapt. John Tavares is projected for 4.84 pts on NumberFire, but even if I thought that number was good (I don’t), would you understand why? How would that change if a different goalie was starting? What are the odds that he puts up the 6+ points I want for a GPP? Building your own projections allows you to have a data set to look at stuff like that and forces you to think through things over time.

    My opinion on this might be different if I was playing NBA which gets more love, but if you don’t know what a site incorporates into their projections (DvP or player specific matchups?) it is hard to just trust them.

    hi i just saw this post and i would like some help in developing my own projections for DFS please this is my email please i have been trying to figure it out i need a push thanks

  • X Unread Thread
  • X Thread with New Replies*
  • *Jumps to your first unread reply

Subforum Index is the home of the daily fantasy sports community. Our content, rankings, member blogs, promotions and forum discussion all cater to the players that like to create a new fantasy team every day of the week.

If you or someone you know has a gambling problem, crisis counseling and referral services can be accessed by calling 1-800-GAMBLER (1-800-426-2537) (IL). Gambling problem? Call 1-800-GAMBLER (NJ/WV/PA/MI), 1-800-9-WITH-IT (IN), 1-800-522-4700 (CO), 1-800-BETS OFF (IA), 1-888-532-3500 (VA) or call/text TN REDLINE 1-800-889-9789 (TN).