fantasymachinelearning.home.blogMachine Learning for Fantasy Football – Predicting fantasy football performance with machine learnin

g Machine Learning for Fantasy Football Predicting fantasy football performance with machine learning Menu Home Contact 2019 Standard Running Back Rankings Last year, I used machine learning to rank the running backs. The results were mixed: I was right about Dalvin Cook, David Johnson, and Jordan Howard, but wrong about LeSean McCoy, Royce Freeman, and Melvin Gordon. Overall, I was right on 9 predictions, and wrong on 9 predictions, and ranked below average compared with the experts. I’ve made some improvements for this season: my model isn’t as needlessly complex, and my predictions on everyone’s carries and stats are more realistic. My model still made some bold calls, but I only see one outlandish predictions. The Rankings: Sorry for the formatting. Player Carries Rush Yards Rush TDs Rec Rec Yards Rec TDs Points ADP 1. Saquon Barkley 255 1149 9 76 616 4 251 1 2. Alvin Kamara 191 890 9 81 695 4 235 2 3. Nick Chubb 254 1194 9 43 326 2 219 7 4. Christian McCaffrey 184 907 6 82 681 4 218 3 5. Ezekiel Elliott 237 1058 7 54 428 2 204 5 6. David Johnson 230 937 7 52 490 3 201 4 7. Joe Mixon 246 1104 8 45 337 1 198 10 8. Todd Gurley 200 909 8 47 413 3 197 9 9. Le’Veon Bell 247 1035 7 49 404 2 195 6 10. James Conner 210 923 8 46 392 2 192 8 11. Melvin Gordon 185 858 7 44 370 2 181 14 12. Dalvin Cook 212 951 6 49 376 2 179 11 13. Leonard Fournette 242 966 7 36 290 1 177 13 14. Derrick Henry 227 1037 9 16 123 0 171 21 15. Marlon Mack 217 960 8 27 209 1 169 16 16. Damien Williams 168 755 7 44 352 3 168 12 17. Kerryon Johnson 194 907 6 43 318 2 165 15 18. Aaron Jones 183 869 7 34 261 1 161 17 19. David Montgomery 200 856 6 32 261 1 157 20 20. Devonta Freeman 184 792 6 34 271 1 151 18 21. Josh Jacobs 193 838 5 35 284 1 151 19 22. Chris Carson 200 894 7 23 183 1 151 22 23. Philip Lindsay 176 824 6 33 239 1 147 25 24. Kenyan Drake 142 655 4 48 391 2 145 29 25. Mark Ingram 209 856 6 28 209 1 144 23 26. Lamar Miller 194 846 4 27 210 1 138 27 27. Sony Michel 193 840 7 15 110 0 137 24 28. Tevin Coleman 153 708 5 30 251 2 136 26 29. Latavius Murray 182 726 7 24 171 1 135 32 30. James White 69 314 2 66 558 4 123 31 31. Miles Sanders 156 634 4 31 251 1 122 28 32. Austin Ekeler 108 522 3 39 351 2 120 34 33. Derrius Guice 154 655 5 24 206 1 120 30 34. Rashaad Penny 152 671 4 25 198 1 119 33 35. Tarik Cohen 74 345 3 54 482 3 115 35 36. Royce Freeman 161 652 5 21 156 1 113 38 37. Darrell Henderson 136 579 4 28 227 1 111 37 38. Kalen Ballage 143 610 7 23 188 1 111 45 39. Jordan Howard 161 662 5 15 112 0 110 36 40. LeSean McCoy 155 620 3 31 236 1 109 41 The Rest: # Player Points ADP 41 Ronald Jones 105 40 42 Carlos Hyde 101 42 43 Alexander Mattison 98 48 44 Jaylen Samuels 95 46 45 Devin Singletary 92 47 46 Peyton Barber 91 43 47 Justin Jackson 91 52 48 Kareem Hunt 91 39 49 Adrian Peterson 91 44 50 Damien Harris 87 50 51 Duke Johnson 87 51 52 Tony Pollard 86 56 53 Nyheim Hines 84 60 54 Matt Breida 83 49 55 D’Onta Foreman 80 59 56 Darwin Thompson 80 55 57 Justice Hill 79 53 58 Mike Davis 78 58 59 Dion Lewis 76 56 60 Ito Smith 74 59 61 Jerick McKinnon 70 54 62 C.J. Anderson 66 61 63 Malcolm Brown 65 62 64 Jalen Richard 57 — 65 Jamaal Williams 54 — 66 Theo Riddick 53 — 67 Giovani Bernard 51 — 68 Frank Gore 46 — 69 Alfred Blue 45 — 70 Gus Edwards 45 — 71 Dontrell Hilliard 45 — 72 Chris Thompson 45 — 73 Wayne Gallman 44 — 74 Ty Montgomery 43 — 75 Chase Edmonds 42 — 76 Mark Walton 39 — 77 Ryquell Armstead 39 — 78 Rod Smith 39 — 79 Benny Snell 38 — 80 Cameron Artis-Payne 37 — 81 Bilal Powell 36 — 82 Qadree Ollison 35 — 83 Dexter Williams 34 — 84 Jordan Scarlett 32 — 85 Alfred Morris 32 — 86 Andre Ellington 30 — 87 Zach Zenner 28 — 88 Doug Martin 27 — 89 Trayveon Williams 26 — 90 Ameer Abdullah 24 — 91 J.D. McKissic 23 — 92 Zach Line 22 — 93 T.J. Logan 17 — 94 David Fluellen 17 — 95\ Josh Ferguson 5 — 96 Darren Sproles -4 — Bold Picks: Sleepers and Undervalued: Nick Chubb: Rank: 3, ADP: 7 Derrick Henry: Rank: 14, ADP: 21 Joe Mixon: Rank: 7, ADP: 10 Melvin Gordon: Rank: 11, ADP: 14 Kenyan Drake: Rank: 24, ADP: 29 Kalen Ballage: Rank: 38, ADP: 45 Nyheim Hines: Rank: 53, ADP: 60 Busts and Overvalued: Le’Veon Bell: Rank: 9, ADP: 6 David Johnson: Rank: 6, ADP: 4 Damien Williams: Rank: 16, ADP: 12 James Conner: Rank: 10, ADP: 8 Kareem Hunt: Rank: 48, ADP: 39 Jerick MicKinnon: Rank: 61, ADP: 54 fantasymachinelearning Uncategorized Leave a comment August 18, 2019 August 20, 2019 2 Minutes Evaluating my 2018 Running Back Rankings Last year, I trained a machine learning model to rank the running backs based on how it expected them to do during the 2018 season. The resulting rankings were different from the consensus, to put it mildly. When I posted my rankings , the results were mixed. Some were interested and wanted to learn more, some thought that my rankings were unrealistic, and others accurately pointed out flaws with my methodology. Now that the 2018 season is in the books, it’s time to evaluate how my model did! First, a look at the boldest predictions: Dalvin Cook: Model: RB27, ADP: RB10, Actual Rank: 31st What’s up with Dalvin Cook? F” for efficiency? I get the injury concern but…that makes no sense” I’m not a huge Dalvin Cook truther or anything, but come on man, you know that his ranking is absurdly low here unless the model is predicting him to be hurt again” When I posted these predictions, this was the one I got the most criticism on. Cook was an exciting running back who had a very good start to the 2017 season before getting injured. He was supposed to be healthy in 2018 and everyone expected him to bounce back. However, I thought he would finish in 27th and finish one spot below his backup, Latavius Murray. Actually, Cook finished in 31st and finished one spot above Murray. Result: CORRECT David Johnson: Model: RB7, ADP: RB3, Actual Rank: 10th Did DJs injury influence the results? He seems crazy low at 7.” Last year, David Johnson was one of four running backs considered to be in the top tier. Me ranking him all the way down at 7th was pretty controversial. However, David Johnson was weighed down by an awful Cardinals offense and he ended up finishing 10th, meaning that my prediction was actually too optimistic. Result: CORRECT Latavius Murray: Model: RB26, ADP: RB44, Actual Rank: 32nd Latavius Murray over Kerryon Johnson huh” Murray being ranked so high was a residual effect of my model expecting Dalvin Cook to bust. Dalvin Cook did bust and Murray did get more of an opportunity, but he didn’t really take advantage of it. He did do much better than most people thought, though. Result: Slightly Correct LeSean McCoy: Model: RB10. ADP: RB17, Actual Rank: 40th If McCoy finishes above Gordon I’ll shit a brick” Yeah, this pick was pretty bad. McCoy was getting older and stuck on the Bills offense, but he looked like he was going to get a lot of carries. Instead, he only got 161 carries and came in 40th in points. Result: INCORRECT Royce Freeman: Model: RB11, ADP: RB18, Actual Rank: 47th there is 0 way royce freeman sees 50 more touches than melvin gordon.” This was by far the worst pick my model made. Royce Freeman looked like an exciting rookie running back. People were already optimistic on him, but my model took that to the extreme, expecting him to close to being a top 10 RB in his first season. A Broncos rookie running back did almost make the top 10, but it was Philip Lindsay, not Royce Freeman. To be fair, nobody saw Lindsay coming, and I didn’t even make a prediction for him. Maybe if I did, Royce would have been ranked lower. But still, Royce didn’t even look impressive in the playing time that he did get, and ranking him 11th is bad. Result: INCORRECT Jay Ajayi: Model: RB14, ADP: RB22, Actual Rank: 78th In what universe is Jay AJayi getting more carries than Melvin Gordon?” After a solid 2016, Ajayi had a down year in 2017. My model expected him to bounce back, but that didn’t happen. He had a very good first game, but was then put on injured reserve after only three...