Bet the Process - Kirk Goldsberry

Podcast Release Date:
Podcast Link:
hagrinBPR Review Date: 2024-05-07 21:15:42
Rating: 6.00

Notes:
Bias - Aware that "Kurt" Goldsberry is the lowest ranked Lowe pod guest on the ShipTheJustice Lowe guest rankings.

Opens with some discussion on Penix. Jeff takes the position of questioning if you may want a QB who is more polished / NFL ready than say just younger because you can possibly maximize the return (surplus value) while on the rookie deal.

7:55 - Brings up the Cazhi tweets which I mentioned in the Cade review which was two parts - 1) "Cade is overconfident in other people's overconfidence" and 2) general frustration from being "praralyzed by the uncertainty". First, I was glad they brought these tweets up because I did run into them before the Cade pod & I thought they were responses that needed to be addressed / defended.

11:15 - Rufus says "The market for player evaluation is efficient, doesn't mean it's perfect, it's as efficient as it can be as there hasn't been demonstrable predictive skill displayed by participants". Well stated by Rufus.

(I mostly agree, but do disagree slightly. Mostly agree - Rufus is basically stating that we may have, or are, rapidly approaching the "predictive limit" of the draft process. There are many papers that try and determine the predictive limit that we can achieve in predicting sporting event outcomes and I would absolutely consume all of that work for many reasons (including identifying the touts that are touting win%s above the predictive limit, that's always fun). It's possible we have approached some theoretical predictive limit in player evaluations ... but I doubt it (which is where I somewhat disagree). Even Cade mentioned that technology is changing so fast and it may takes several years for that new tech's influence to actually be felt. As someone who has done scouting work now for 8 years, especially surrounding the QB evals the last 3 years, it's hard for me to take those discussions and drafts and describe them as "highly efficient". But hey, maybe I am overconfident that we can still do better at narrowing the probabilities around the range of outcomes for a prospect - I definitely think we can do better even if it's only slightly which I think was Cazhi's second point.)

14:40 - Discount rate discussion. 15:22 - "Ideally the discount rate should be related to a GM being fired." Jeff retorts that this isn't the right discount rate and I agree despite Rufus positing "what are we trying to measure". Almost immediately we venture into this abstract discussion about the discount rate when talking about our existence in this universe and it's just not that difficult.

17:05 - "What it should be if there was no Principal Agent problem, or as I like to call it, the Principal Asian problem, that's the problem with this podcast." Funny line! "I can make that joke, you can't." "Well, I did." Funny! It then gets a little loose from there, but

19:45 - Kirk section starts. The first 45 seconds is so rough.

24:50 - "As data becomes larger, visulaization becomes more important." Agree. 26:20 - Currently working on building a shot chart of the human body, thinks baseball and golf are ahead on this. 31:15 - Brings up that Data Viz at high level technical universities isn't represented well enough in the cirriculum. (Would like to hear from the college students that follow this account, is this true?).

37:15 - State of the NBA, current foul calling situation. 38:00 - "Hell yeah." PP/100, Silver takes over and then for 10 years by season "and then we see what bitcoin people would dream of is a 10 year spike hockey stick shape". Kirk, we don't have to dream a 10 year dramatic increase, it happened buddy. Talks about how this midseason change had ramifications in the betting community and how some people were upset who built models, think a lot of NBA bettors probably saw this as an opportunity so interesting framing. Jeff asks about possible betting angle and KG mentions ShaiGA and Luka, but ultimately might not affect it in that way.

43:00 - Injuries. "I don't have a good analytical answer here." Great! Brings up 65 game change, Haliburton. "82 games is too much."

48:15 - Rufus asks what the theoretical maximum would be for PP/100. Good question. "eFG leap slowing down, 3PT rate". Still improving, but flattening out. Officiating knobs and levers. 50:00 - asks about watchability of the product when we try and maximize for PP/100. Kirk advocates for moving back the 3 point line, "makes the NBA bimodal". 51:30 - Advice for someone entering the visualization space. (Rufus really crushed it with these last 3 questions, extracted a bunch of good responses).