Diamond Plots and Diamonds In The Rough, Part 5

Let's put a bow on our holiday season examination of prospect values (see what I did there?) by retreating ever so slightly to the comforts of a simple scatter plot. 

In the third part of our series, we built out the list of z score values for the hitting and pitching prospect groups of all 30 MLB teams from 2019 - 2024 (data courtesy of Fangraphs). That full list of 180 z scores (30 teams, 6 seasons) can be found HERE in my GitHub. 

We then pivoted last week in Part 4 to roll up those z scores by team and look at cumulative success across the timeframe. Let's undo that quickly and go back to looking at all those 180 team-season combos in graph form. This way, we can really see how hitter and pitcher z scores interact with one another, further informing our understanding of how teams build their farm systems.

To start, let's just simply put all 180 team-seasons on a graph, like so: 

I've added in lines that bisect 0,0 so we can see the distribution of groups more clearly. This is very similar to the diamond plot analysis that we did to start the series off, but with more observations to bulk up our analysis a little. Indeed, while bulked up, not much actually stands out in this chart - most observations are somewhere in the -1,-1 to 1,1 square of "perfectly cromulent organization" when it comes to prospects. To be sure, this is to be expected! 

Let's try adding in a trend line. Now, to be clear, I do not even for one second see this line as predictive or overly explanatory. I'm including this visual solely to lend a little bit more visual flair to the graph itself. 


Again, not much of a trend at all here. In the middle, there's a total lack of relationship, as you can see by the straight line going from -2 to 1 on the hitter z score axis. Once we get above a hitter z score of 1, maybe you can squint and start to say that strong hitting development or pitching proxies for overall organizational excellence, namely in the complementary position group. To these eyes, however, it's much more a story of outlier organizations and individual prospect classes than of anything determinative. 

We can also organize the graph by team. 


Above, we've plotted out every team's z scores by season within the time frame and slapped a crude "Line of Average-ness" at 0 just for quick glances. It helps us see, for instance, the relative struggles of Washington and Philadelphia vs. the consistent success of the Rays and even the Twins. These are dynamics we've previously discussed, just packaged and presented in a slightly new way. 

Finally, let's again add a quick little heuristic to help us. This time, we'll shade a box between -1 and 1 total z score. Any team-season falling in this box is within one standard deviation of the average value that year, making it more or less ____. Anything outside of the box is where we want to draw our attention towards. 


This helps more to drive quick, dirty narratives around the organizations themselves. The Cardinals, for instance - pretty average! The Angels and Orioles - distinctly not so average (for distinct reasons)!

Again, as these are high-level visuals, I'd caution any sort of prescriptive analysis coming from this. Instead, think of these as conversation starters, especially in the context of any priors you might have about each organization's success or struggles to develop their prospects. 

But you tell me - drop a note in the comments! 

Next week, we'll layer in some brand new data to start to put some rigor behind the old adage of "Grow the Bats, Buy the Arms". Does old baseball wisdom hold up after all? 

Thanks for reading! 

--------

Code is HERE on my GitHub. 

--------

Notes: 

Two-way (batter and pitcher) prospects are split (0.5/0.5) between the batter and pitcher tiers for valuation purposes.

From: 

https://thef5.substack.com/p/how-to-diamond-plots-in-r

https://blogs.fangraphs.com/an-update-to-prospect-valuation/

https://blogs.fangraphs.com/putting-a-dollar-value-on-prospects-outside-the-top-100/

https://blogs.fangraphs.com/how-the-draft-and-the-trade-deadline-affected-our-farm-system-rankings/

https://www.fangraphs.com/prospects/farm-system-rankings/2024-prospect-list 





Comments

Post a Comment

Popular posts from this blog

Diamond Plots and Diamonds in the Rough

Image
Well, blog, it's been awhile! Here's a quick one to get back on the board. My hope is to make posts like these much more of a regular occurrence, though, so stay tuned!  Today we'll be working with diamond plots in R. This post is much more proof of concept than anything earth-shattering, but the reps are needed, so away we go.  First, thanks to Owen Phillips of the F5 for his "How To: Diamond Plots in R" post  which I followed to a significant degree to build this.  Traditional X-Y plots are not necessarily the most intuitive to compare data points intra-graph. By flipping the plot 45 degrees, we better align the graph's visuals with our own human nature - namely, making quality judgements top to bottom in a descending order.  Here, let's use diamond plots to get a sense of the varying ways MLB farm systems are built, namely, the number of (ranked) hitters vs. (ranked) pitchers in a given team's system. This will be a quantity exercise to start - the ...
Read more

Diamond Plots and Diamonds In The Rough, Part 2

Image
Last week , we introduced diamond plots and very quickly looked at applying the visual to MLB farm system composition. From that post: "Let's use diamond plots to get a sense of the varying ways MLB farm systems are built, namely, the number of (ranked) hitters vs. (ranked) pitchers in a given team's system. This will be a quantity exercise to start - the value of those players will be the focus of next week's post. We'll use the Farm System Rankings from Fangraphs 2024 Preseason Prospect Report  to get the breakdown of each system. Remember, these are essentially the prospects of note in a system (ranked with a 35+ FV or higher), not a comprehensive look at how many players each organization has rostered." Here's what we got:  Hard to quibble with it from a visual perspective - nice even spread, some clear winners and losers, and even some high-level trends to parse out. Be sure to check out my post from last week for some more thoughts there.  Which brin...
Read more

Diamond Plots and Diamonds In The Rough, Part 3

Building off of last week's post - where we moved the lens towards MLB farm system value, not just prospect count - let's take a look at z scores by organization and by year. Here, we'll run the same exercise as we did last week, creating a z score value (number of means above the standard deviation) but for each year, getting us 30 teams x 6 years = 180 observations of z scores.  Critically, each z score is anchored to the season in question. That is to say, the value of farm system's hitting, pitching and total prospect count is benchmarked to the league-wide values from that year alone. I think measuring z scores within years like this would more accurately help us track population-level changes in prospecting. Normalizing within each year ultimately helps us make comparisons across years that reflect newfound understanding of valuation at a macro level.  For instance, comparing the nominal values of SFG's 2020 hitting prospects ($191.5M) and Milwaukee's 202...
Read more