beanumber
commented
2 years ago Closing this in favor of #34 #33 #32 #31. All other issues have been addressed.
Closed beanumber closed 2 years ago
beanumber
commented
2 years ago Closing this in favor of #34 #33 #32 #31. All other issues have been addressed.
Pages 3,4, 5: The use of historical tables and data is powerful. However, I worry about the quality of the images particularly Figure 2. Page 6, line 8: “the effect may be overstated” is somewhat ambiguous. Is it the Page 6, line 53: Is the “!” warranted here for conditioning on a 9-dimensional vector? Page 8, line 53: There is also work on expected possession in ice hockey, see Stokes et al. https://hockey-graphs.com/2021/07/06/bayesian-space-time-models-for-expected-possession-added-value-part-1-of-2/
9, 17: Is Player Win Average relevant to win probability? This is more of a player contribution metric than a win probability metric. I would rather see a brief exposition of Tango et al (2007) here. 9, 47: Gridiron football is a lovely term but for this audience, I think it needs to be clarified. 10, 52: “In this paper”, the “this” seems ambiguous. 11, 36: Table 3, why order this table alphabetically, why not by ability level? 11, 50: Given how much fivethirtyeight.com uses Elo ratings for all kinds of sports, it is probably worth mentioning their work as examples. 12, 1: As noted above, Section 4 feels much weaker than Sections 2 and 3 which are strong. Perhaps a look at Pythagorean methods as a historical introduction to this topic. 12, 23: This is not much of a case study. Perhaps some additional information on the data available, ways it could be used to calculate game state expected valued, win probability, player ratings…. 12, 37: Theming teamcolors is a great R package, not sure it is part of the “big ideas” in sports analytics. A single line in the first paragraph of this section would seem to be better. 13, 35: Add a citation for Big Data Cup, https://www.stathletes.com/big-data-cup/, for SABR Diamond Dollars Case Competition, https://sabr.org/analytics/case, Big Data Derby https://www.kaggle.com/competitions/big-data-derby-2022, and maybe more. 13, 43-53: don’t think there is a need for a bulleted list here.