Fantasy Football Analytics Interview with Rotoviz27
The creator of Fantasy Football Analytics discussed the website in a recent interview with Rotoviz on the Numbers Game Podcast. We discussed the “wisdom of the crowd”, subscription vs. free projections, reasons to use R over Excel, lessons learned from previous first round fantasy picks, and more.
Fast forward to 25 minutes of the audio to hear the interview:
[soundcloud url=”https://api.soundcloud.com/tracks/271395471″ params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]
But I didnt understand how you can figure out which players to avoid, for example last year CJ Anderson had a high risk, number wise. He had a low risk number with 3.34, this year David Johnson pretty low 3.55. But in the interview you said they are riskier because of very little games played, but it doesnt show up in the risk number.
Thats the part I couldnt understand in the interview.
See my response to Kenneth below for a description of how we calculate risk. Our calculation does not take into account sample size (how many games played i.e., the number of games from which the projections are estimated), so that is an additional factor that is important to consider.
Hope that helps!
Have you considered adding age and lack of data (rookies and sophomores) as additional factors in the risk score? Or at least using lack of data as a factor and adding age as an additional column (Age could supposedly be factored into the projections, but I think many older players are still overvalued)?
We do plan to add columns for age, experience, etc. Probably wouldn’t add them into the risk index because they have non-linear effects (i.e., both young and old players have high risk). Adding them as columns would provide users more info to use in their decision making.
An unexpected error has occurred. Please contact the system administrator.
– Current problem with the projections app
Apps should be up now!
Great web site and I plan on using this in my upcoming draft but the more I think about the snake draft tool…I am in a league where my roster has 3 flex spots (QB-RB-WR-TE-WR/RB-WR/RB/TE-WR/TE).
Should there be some input for my roster since VOR will depend on how I decide to fiil out my roster? Maybe this input is not needed…help, a little confused with draft on Sat.
Hope that helps!
I think risk is based on the range of projections. So if everyone has the same player with the same stats he has a low risk. If there is a big difference in the high/low projections then high risk. I could be wrong….
Yes, risk is based on the variability of projections and rankings across sources, along with their injury risk. For more info, see here:
I have been looking at trying to sort out consistent players versus boom/bust players. Is coefficient of variation the best way to look at that? Is that something you report that I have missed. This is an awesome website. Thank you
Yes the CV would tell you players’ game-to-game consistency or inconsistency. Players with a lower CV are more consistent. Players with a higher CV are more inconsistent (boom/bust as you say).
Hope that helps,
Do you have that information recorded? Is it possible to scrape that from any of the sites? The R program is very new to me, but is very useful.
You can download historical weekly performance from our Projections tool or from ProFootballReference. You can then calculate game-to-game CV from that.
I am trying to come up with a little more rigor for drafting, particularly with VOR as well as dropoff. Any ideas how to come up with a rule? Or is this where we have to shift more into drafting as an art than a science?
Oh, and I already have a rule for risk-adjusted VOR, so that’s incorporated, although it could be better. I am just subtracting the risk value less five, and then I am subtracting additional amounts for players that have very low experience or are nearing the dropoff ages (e.g., AP, Forte). Could be better I’m sure, but at least it is something.
I also have a rule to adjust VOR for starters to weight their floor higher, and to weight ceiling higher for bench players.
If you were to build an algorithm/rule, it would probably take into account other factors, as well (e.g., risk, positions you need, positions your opponents need, etc.). If you just want to focus on VOR and dropoff, I’d suggest focusing first on VOR and use dropoff to distinguish between players who have similar VORs.
Just wondering why seattle and carolina defenses are ranked so low in projections…i mean if vikings out perform the two previous as u have forecasted..then u guys have truly mastered the art…but not a single other site has taken such a bold prediction.
We aggregate multiple sources of projections using an average. The reason for the lower projections is that FantasySharks have projected the Seahawks DST at 0 safeties and 0 TDs.
Isaac, apologies if you have a post elsewhere on the site about this topic, but have you ever looked at (and is there enough data to reliably evaluate) position-level projection quality? For example, maybe CBS is great at projecting receptions and FFToday is consistently the best at XP? Possibly making the case that each position should be weighted separately?
You can examine the accuracy of projections by position in our tools. For more info, see here:
Will the weekly lineup optimizer be available for weekly use with 2016 projections? So far I’ve only been able to run the optimizer for the full 2016 season. Will the week 1 optimizer come out when the major websites put out their weekly projections?
I have seemed to have found a problem with QB/RB/WR/TE flex set to 1, AAV set to ESPN and PPR scoring for the Auction Projections. It sets all the auction values very low like to $1.
Also, is there any way to download the auction projections?
You can download from the Projections tool, not currently from the Lineup Optimizer (but it’s on our to-do list).
Should the dropoff value column in the snake draft optimizer be dynamic as players come off the board (or if they begin the draft off the board in the case of keeper leagues)?
Yes, that’s on our to-do list, thanks!