Week 2 NFL Daily Fantasy Football Results - My Bayesian Inspired Lineups Took a Hit This Week (-$50.00)

in #sports8 years ago (edited)

Week 2 - Cue the loser horn!

Week 2 was a bit of a bust for me. My r^2 value went down a measly 1%, but I lost more this week, going down about $50. The results are disappointing, but, there were still many insights to gain!

Week 2, 2016

My overall 2016

Why did I stink so bad this week?

First, I tried a different strategy this week by playing a variety of lineups split heavier towards cash games. I attempted a two-thirds split over cash games and a one-third over GPP tournaments. For my cash games, I skewed more towards my my algorithm optimizations. My GPP games were based on extreme scenarios that skewed projections on many players in projected high scoring games to their 99th percentile potential.

Despite my best efforts to craft as many diverse lineups as possible across cash games and GPP tournaments, I can attribute this week's losses to a heavy reliance on older players and underwhelming performances compared to my predictions. Specifically, Antonio Brown, Drew Brees, Danny Woodhead, and Arian Foster were awful choices for me this week.

As always, I exclusively played on Draft Kings for every lineup, with my username of twigsta26. Listed below are my highlights:

My Best Lineup

I had this one in 12% of my lineups. I wish i had it in more! Eli was a very high risk play by my algorithm, but I liked the stack with Sterling Shepard, who I still felt was undervalued. Having Benjamin and Olsen certainly helped as well.

My Worst Lineup

This was my dreaded algorithm optimization lineup. Everyone here I needed to do well underperformed significantly, or got injured in the first quarter of their games. UGH!

Algorithm Results

I had a median algorithm difference of underestimating a player's performance by 0.9 points, with the majority of my quartile data falling between a 0.9 point underestimate and a 2.5 point overestimate. I provided database examples of players that described each range of my projections.

This week, the algorithm tended to overestimate data. This no doubt led to my poor lineup constructions as I picked players that significantly unperformed their projections. In future weeks, I need to try to squish down these outliers some more by taking a much closer look at the relationship between the most recent information on a player and how that relates with their historical projection curves for each week.

I also need to realize that for many players, I have a very small sample set that will only make my algorithm better in future weeks when I acquire more significant information. I need to cool it a bit with my risk overall while the season is young!

The Detailed List - How the Model Fared for Each Player

Note in the difference category, a negative number is in parentheses and represents an underestimate, a positive number represents an overestimate.

PlayerPosTeamOppMyModelActualDifference
Drew BreesQBNONYG23.414.528.9
Antonio BrownWRPITCIN22.67.914.7
Cam NewtonQBCARSF22.134.82(12.7)
Andrew LuckQBINDDEN22.012.089.9
Russell WilsonQBSEALA20.611.569.0
Ben RoethlisbergerQBPITCIN20.321.76(1.5)
Matthew StaffordQBDETTEN19.916.53.4
Aaron RodgersQBGBMIN19.519.420.1
Julio JonesWRATLOAK19.424.6(5.2)
Kirk CousinsQBWASDAL19.322.56(3.2)
Matt RyanQBATLOAK19.031.84(12.8)
Matt ForteRBNYJBUF18.733.9(15.2)
Joe FlaccoQBBALCLE18.621.18(2.6)
Alex SmithQBKCHOU18.55.6412.9
Derek CarrQBOAKATL18.324.96(6.6)
Ryan TannehillQBMIANE18.228.06(9.9)
Andy DaltonQBCINPIT18.022.34(4.3)
Philip RiversQBSDJAX18.024.8(6.8)
Jameis WinstonQBTBARI18.09.628.4
Sam BradfordQBPHIGB18.019.14(1.2)
Eli ManningQBNYGNO17.516.321.2
Blake BortlesQBJAXSD17.224.56(7.3)
Jordan MatthewsWRPHICHI17.013.13.9
Brandon MarshallWRNYJBUF17.019.1(2.1)
DeAngelo WilliamsRBPITCIN16.923.2(6.3)
Arian FosterRBMIANE16.70.915.8
Jordy NelsonWRGBMIN16.418.3(1.9)
Brandin CooksWRNONYG16.313.82.5
Marcus MariotaQBTENDET16.317.62(1.3)
Carson PalmerQBARITB16.327.32(11.1)
AJ GreenWRCINPIT16.25.810.4
Jay CutlerQBCHIPHI16.04.2811.7
Odell Beckham JrWRNYGNO15.916.6(0.7)
Josh McCownQBCLEBAL15.516.4(0.9)
Ryan FitzpatrickQBNYJBUF15.224.06(8.9)
DeMarco MurrayRBTENDET15.221.5(6.3)
LeSean McCoyRBBUFNYJ15.0132.0
Demaryius ThomasWRDENIND15.016(1.0)
Blaine GabbertQBSFCAR14.922.72(7.8)
Doug BaldwinWRSEALA14.859.8
DeAndre HopkinsWRHOUKC14.827.3(12.5)
Danny WoodheadRBSDJAX14.74.110.6
Emmanuel SandersWRDENIND14.46.97.5
Jordan ReedTEWASDAL14.3122.3
Mike EvansWRTBARI14.219(4.8)
Adrian PetersonRBMINGB14.24.69.6
DeSean JacksonWRWASDAL14.177.1
David JohnsonRBARITB14.117.3(3.2)
Sammy WatkinsWRBUFNYJ14.0410.0
Jeremy MaclinWRKCHOU14.012.81.2
Alshon JefferyWRCHIPHI13.814.6(0.8)
Amari CooperWROAKATL13.612.11.5
Larry FitzgeraldWRARITB13.620.1(6.5)
Randall CobbWRGBMIN13.59.34.2
Case KeenumQBLASEA13.510.063.4
Eddie LacyRBGBMIN13.458.4
Todd GurleyRBLASEA12.984.9
Pierre GarconWRWASDAL12.74.58.2
Dez BryantWRDALWAS12.520.2(7.7)
Eric DeckerWRNYJBUF12.527.6(15.1)
Steve SmithWRBALCLE12.59.43.1
Michael CrabtreeWROAKATL12.313.1(0.8)
Tyrod TaylorQBBUFNYJ12.225.38(13.1)
Brock OsweilerQBHOUKC12.212.42(0.2)
Jason WittenTEDALWAS11.98.13.8
Doug MartinRBTBARI11.92.39.6
Devonta FreemanRBATLOAK11.99.32.6
Lamar MillerRBHOUKC11.611.7(0.1)
Terrelle PryorWRCLEBAL11.36.25.1
Allen RobinsonWRJAXSD11.28.42.8
Latavius MurrayRBOAKATL11.222.1(10.9)
Ryan MathewsRBPHICHI11.016.5(5.5)
Jimmy GrahamTESEALA10.97.23.7
Frank GoreRBINDDEN10.615.3(4.7)
Tavon AustinWRLASEA10.610.6(0.0)
Julian EdelmanWRNEMIA10.414.6(4.2)
Mark IngramRBNONYG10.48.71.7
Thomas RawlsRBSEALA10.43.86.6
Jarvis LandryWRMIANE10.425.7(15.3)
Stefon DiggsWRMINGB10.436.2(25.8)
Willie SneadWRNONYG10.416.4(6.0)
Gary BarnidgeTECLEBAL10.17.72.4
TY HiltonWRINDDEN10.08.11.9
Jermaine KearseWRSEALA9.93.16.8
Golden TateWRDETTEN9.93.36.6
Greg OlsenTECARSF9.826.2(16.4)
Rashad JenningsRBNYGNO9.863.8
Kenny BrittWRLASEA9.815.4(5.6)
Isaiah CrowellRBCLEBAL9.724.8(15.1)
Jamison CrowderWRWASDAL9.615.9(6.3)
Darren SprolesRBPHICHI9.66.82.8
Anquan BoldinWRDETTEN9.514.8(5.3)
Cole BeasleyWRDALWAS9.512.5(3.0)
Mike WallaceWRBALCLE9.220.1(10.9)
Jeremy HillRBCINPIT9.28.90.3
CJ AndersonRBDENIND9.118.3(9.2)
TJ YeldonRBJAXSD9.011.8(2.8)
Davante AdamsWRGBMIN9.04.64.4
Kyle RudolphTEMINGB9.012.1(3.1)
Carlos HydeRBSFCAR8.97.21.7
Vincent JacksonWRTBARI8.88.40.4
Kelvin BenjaminWRCARSF8.832.8(24.0)
Tyler LockettWRSEALA8.813.9(5.1)
Alfred MorrisRBDALWAS8.76.72.0
Torrey SmithWRSFCAR8.714.5(5.8)
Delanie WalkerTETENDET8.620.3(11.7)
Allen HurnsWRJAXSD8.611.4(2.8)
Michael FloydWRARITB8.68.8(0.2)
Shane VereenRBNYGNO8.68.60.0
Seth RobertsWROAKATL8.65.53.1
Justin ForsettRBBALCLE8.67.61.0
Duke JohnsonRBCLEBAL8.49.6(1.2)
Andrew HawkinsWRCLEBAL8.45.82.6
Brandon LaFellWRCINPIT8.36.91.4
Bilal PowellRBNYJBUF8.21.36.9
Antonio GatesTESDJAX8.110.5(2.4)
Spencer WareRBKCHOU8.111.5(3.4)
Charles SimsRBTBARI8.05.82.2
Tim HightowerRBNONYG7.90.97.0
Mohamed SanuWRATLOAK7.94.93.0
Giovani BernardRBCINPIT7.828.7(20.9)
Travis KelceTEKCHOU7.78.4(0.7)
Victor CruzWRNYGNO7.612.1(4.5)
Robert WoodsWRBUFNYJ7.625.6
Donte MoncriefWRINDDEN7.61.95.7
Jonathan StewartRBCARSF7.63.64.0
Theo RiddickRBDETTEN7.510.5(3.0)
Cecil ShortsWRHOUARI7.52.25.3
Melvin GordonRBSDJAX7.424(16.6)
Julius ThomasTEJAXSD7.411.1(3.7)
Nelson AgholorWRPHICHI7.48.2(0.8)
Marvin JonesWRDETTEN7.322.8(15.5)
Charles ClayTEBUFNYJ7.37.7(0.4)
Jerick McKinnonRBMINGB7.21.16.1
Andre JohnsonWRTENDET7.27.9(0.7)
Dorial Green-BeckhamWRPHICHI7.13.83.3
Travis BenjaminWRSDJAX7.132.4(25.3)
Terrance WestRBBALCLE7.16.70.4
Chris ThompsonRBWASDAL7.08.8(1.8)
Christine MichaelRBSEALA6.910.6(3.7)
Matt JonesRBWASDAL6.813.5(6.7)
Coby FleenerTENONYG6.74.91.8
Jared CookTEGBMIN6.67.1(0.5)
Ameer AbdullahRBDETTEN6.53.82.7
Lance DunbarRBDALWAS6.55.21.3
Ted GinnWRCARSF6.56.6(0.1)
Lance KendricksTELASEA6.410.1(3.7)
Charcandrick WestRBKCHOU6.47.3(0.9)
Brandon ColemanWRNONYG6.43.52.9
Dennis PittaTEBALCLE6.422.2(15.8)
James StarksRBGBMIN6.32.93.4
Eddie RoyalWRCHIPHI6.315.2(8.9)
Richard RodgersTEGBMIN6.14.51.6
Vernon DavisTEWASDAL6.110.1(4.0)
Travaris CadetRBNONYG6.13.72.4
Tevin ColemanRBATLOAK6.115.1(9.0)
Jeremy LangfordRBCHIPHI6.19.4(3.3)
Eric EbronTEDETTEN6.09.3(3.3)
Devante ParkerWRMIANE6.021.6(15.6)
Dontrelle InmanWRSDJAX5.91.74.2
Shaun DraughnRBSFCAR5.82.13.7
Paul RichardsonWRSEALA5.76.5(0.8)
John BrownWRARITB5.72.43.3
Luke WillsonTESEALA5.61.64.0
Quincy EnunwaWRNYJBUF5.615.2(9.6)
Brian QuickWRLASEA5.51.83.7
Chris JohnsonRBARITB5.411.4(6.0)
Clive WalfordTEOAKATL5.417(11.6)
Jordan CameronTEMIANE5.415.9(10.5)
Benjamin CunninghamRBLASEA5.41.93.5
Austin Seferian-JenkinsTETBARI5.33.41.9
Jeremy KerleyWRSFCAR5.15.9(0.8)
Crockett GillmoreTEBALCLE5.03.21.8
Martellus BennettTENEMIA5.025.4(20.4)
Adam ThielenWRMINGB5.08.1(3.1)
Zach MillerTECHIPHI5.07.3(2.3)
Kenny StillsWRMIANE4.911.9(7.0)
Charles JohnsonWRMINGB4.93.51.4
Jimmy GaroppoloQBNEMIA4.821.36(16.5)
James WhiteRBNEMIA4.84.9(0.1)
Alfred BlueRBHOUKC4.81.13.7
Devin FunchessWRCARSF4.79.9(5.2)
Darrius Heyward-BeyWRPITCIN4.71.73.0
Quinton PattonWRSFCAR4.65.5(0.9)
Dwayne AllenTEINDDEN4.64.50.1
Jacob TammeTEATLOAK4.618.5(13.9)
Phillip DorsettWRINDDEN4.540.5
Adam HumphriesWRTBARI4.312.7(8.4)
LeGarrette BlountRBNEMIA4.321.3(17.0)
Mike GillisleeRBBUFNYJ4.18.8(4.7)
Will TyeTENYGNO4.031.0
Albert WilsonWRKCHOU4.03.10.9
Vance McDonaldTESFCAR4.014.5(10.5)
Andre HolmesWROAKATL3.97.6(3.7)
Denard RobinsonRBJAXSD3.82.41.4
Andre EllingtonRBARITB3.82.11.7
Rishard MatthewsWRTENDET3.88(4.2)
Marqise LeeWRJAXSD3.812.5(8.7)
Justin HardyWRATLOAK3.77.8(4.1)
Corey BrownWRCARSF3.64.5(0.9)
Danny AmendolaWRNEMIA3.520(16.5)
Jesse JamesTEPITCIN3.411.9(8.5)
Zach MillerTESEAPHI3.37.3(4.0)
Garrett CelekTESFCAR3.32.21.1
Chris ConleyWRKCHOU3.23.5(0.3)
Larry DonnellTENYGNO3.26.4(3.2)
Ryan GriffinTEHOUKC3.01.51.5
Josh HuffWRPHICHI2.90.92.0
Darren FellsTEARITB2.97.1(4.2)
Jack DoyleTEINDDEN2.77.7(5.0)
Greg SalasWRBUFNYJ2.718.9(16.2)
Chris HoganWRNEMIA2.79.9(7.2)
Cameron BrateTETBARI2.64.6(2.0)
Tyler KroftTECINPIT2.56.5(4.0)
Anthony FasanoTETENDET2.11.11.0
Fozzy WhittakerRBCARSF2.118.1(16.0)
Fitzgerald ToussaintRBPITCIN2.12.3(0.2)
Rashad GreeneWRJAXSD1.61.7(0.1)
James WrightWRCINPIT1.52(0.5)
Jordan NorwoodWRDENIND1.55.4(3.9)
Marcedes LewisTEJAXSD1.49.7(8.3)
Jay AjayiRBMIANE1.47.5(6.1)
Robert TurbinRBINDDEN1.37(5.7)
David JohnsonTESDTB0.917.3(16.4)
Ka'Deem CareyRBCHIPHI0.81.6(0.8)
Jaron BrownWRARITB0.815.8(15.0)
Niles PaulTEWASDAL0.82.6(1.8)
Dion SimsTEMIANE0.63.1(2.5)
Cody LatimerWRDENIND0.53.2(2.7)
Virgil GreenTEDENIND0.47.6(7.2)
Alex SmithTECINHOU0.25.64(5.4)

Sort:  

Don't forget the rookies! Tajae Sharpe and Will Fuller were just too cheap this week, and a few guys with situations that are improved via injury should have been a bit higher like Travis Benjamin. It's a tough thing to model since the samples are so small for bench players, even smaller for rookies, and people are always getting injured. Next week make sure to tweak a couple guys that are benefiting from injuries.

But tough break with Antonio, crushes every week...except one a season. Last year's dud game made sense: he went up against Jason Verrett who is one of the best corners in the league and got shut down. Sunday he just didn't seem to get targeted much, while Sammie Coates (who I think is awful) was targeted a bunch. Weird day for him. Same for Brees, how does he not murder vs the Giants? I projected him for 330 yards haha, one of the highest I can ever remember.

Have you ever tried DFS baseball? It's my worst sport by far, but I think these models would be really interesting for it cause the samples are so huge.

Also really interested if a future version of your model can start beating human projections. Models crush in NBA and MLB, do ok in golf, but seem to be giving up edge in football and mixed martial arts. Lack of sample sizes make eyeball numbers do a bit better, but there has to be a way to model it better than human brains are capable of.

Thanks for the feedback! The model here takes a huge hit for not having a projection on a player until they hit at least two sample points in a season. Even then, their gamma curves are really weird looking and inconsistent until they've hit their 8-9 game mark. In backtesting this is around where the model started winning consistently, hoping that trend continues.

In news of the positive, this upcoming week is my first with rookies present. Although as I will write soon, their estimates tend to be very conservative to start because of the lack of a body of work.

One thing I am tweaking a bit this week is the statistical significance of defenses and the impact that has on point projections. Modeling has it count for a little more of the variance in points projections than what I have already accounted for.

I hope no one hates me for saying this, but I haven't tried applying the model to baseball yet because I'm not the biggest fan of baseball. Casual watcher at best. I am though a huge NBA fan and am currently looking into the applicability of a Bayesean algorithm there!

I completely get the lack of love for baseball. I played for 13 years and even I have trouble watching at times. But the statistics part of it is fascinating. Events seem to occur randomly at certain frequencies, the sample sizes are enormous, and the game can be modeled with simple markov chains. You would fall in love with baseball stats. But basketball is still more fun so really interested to see what you can put together there

Shouldn't your title be Week 2 ... not Week 1?

Yep, bleary eyed team-leibniz was bleary eyed last night ;)

Hey @team-leibniz and @rdaut44, good stuff. I wanted to bring your attention to this site, which is interesting because they already have a platform which has statistical R and gamma curves as well as outside analysis factored into their model; http://fantasyfootballanalytics.net/category/r
FWIW, its useful to glance over it as you compare to your own research.
So team-leibniz from your Week 2 projections, you had some gold-nuggets in there, specifically with Arizona D, which had a monster game you had them ranked numero uno, that was a solid pick. Im looking foward to your week 3 projections. I looked at your optimizer spreadsheet and have a question, how do you easily/quickly update the projections per player ? I can do that with vlookup to update the salaries, but the projections are a manual effort? thx!

Thanks for the great feedback! I will be sure to look into that site in the weeks to come!

I have actually been update the projections for player once each week on Thursday when I release my projections. On my side, everything is automated, I use a program called R to populate this information for me after I make global changes to injuries, changes in weather, etc. But, Vlookup in excel would be just about the fastest way to do so otherwise if you had information of your own you would like to substitute in!

Coin Marketplace

STEEM 0.18
TRX 0.14
JST 0.029
BTC 56948.01
ETH 3056.88
USDT 1.00
SBD 2.40