reviewer 1 specific comments

Below I have several more specific suggestions/comments, which I lay out in order of where they appear in the paper:

In the second sentence, baseball is introduced as a “naturally discrete” sport, but there are certainly many things about baseball which are not discrete, as talked about later in the article. It would be helpful to clarify what is meant by discrete in this setting. Also, later on in the first paragraph, it would be useful to have a citation or two about the changes being brought about in baseball due to analytics.
In Section 2.1, Figure 1 (George Lindsey’s calculations) is introduced before the reader even knows who George Lindsey is. Furthermore, the expected run matrix is displayed “in its more familiar form”, but I would venture for most readers of this article, no form would be familiar. I think the caption does explain the matrix nicely, although it might be useful to point out from which column in Lindsey’s table the matrix is derived.
In Figure 2, I found it quite surprising that the expected point value was so negative at the 95 yard line. Were safeties very common in this era of American football? This might be worth mentioning/discussing.
The title of Section 2.3 seems to indicate that the aforementioned sports have only discrete states, but this is not always the case. Some clarification would be useful here.
In general, the organization of Section 2 is confusing. It first seems to be going sport by sport, but then 2.3 is about several different sports and then 2.4 is on a general topic altogether. As is, it reads somewhat disjointed and is hard to follow. Section 3 flows much better in this sense.
Table 2 seems unnecessary, as the content is entirely summarized in the text.
After the comprehensive views in Sections 2 and 3, Section 4 seems sparse on details. Much detail is given on the Bradley-Terry model, but only one example is given of a possible usage. Has this been used in other sports? Has it changed how we think about those sports or team strategies? An explanation of Elo rating should be given; I for one have no knowledge of how to compute it.
There is a large body of literature on Bayesian state-space models in sports outside of the Glickman and Stern paper, mainly in soccer/association football. See for example Koopman and Lit (2015) and the various references in its introduction.
As mentioned earlier, Section 5 seems sparse on details, and furthermore it seems odd to choose chess as the case study when it has barely been mentioned previously in the article. I think this section could use a table of some sort, perhaps something that summarizes the CRAN task view at a high level. Another potential would be a figure which shows a typical workflow of sports analytics as explained in the text.
Figure 6 doesn’t add much to the text; I don’t think the reader better understands sports analytics by visualizing many team colors. Instead, it would be better to show how the theming can lead to a more impactful graphic for sports, i.e. show how that package might be used in a real-world setting.
For Section 6, there is also now a “Big Data Derby” in horse racing. All in all, I think the article explains the ideas well and covers a lot of ground, but more care should be taken towards the organization and what to include as figures. The last half of the paper has a very different pace than the first half. I think a common theme in the paper is the rise in tracking data and how that has impacted sports–perhaps that idea could be better unified and stated in a specific section of the article, or at the very least noted in the conclusion.

In addition to the above general comments, I also want to note some minor typos.

Page 4, line 56: Needs to be some sort of separation between the “absorbing states” and “touchdown”
Page 5, line 56: Typo for “causal”
Page 5, line 57: Should be “too conservative”
Page 7, line 37: Should be “led” not “lead”

It was a pleasure reading your article! I hope the comments prove useful.

beanumber / wire21

reviewer 1 specific comments #28