In early October, our A.E. Schafer solicited article ideas from our readership. There were a lot of great ideas, but one in particular grabbed my eye.
...we can quantify what kind of WAR each draft position has produced over the years on average. Are there systems that have consistently beat those averages over the years.
That idea is ripe for a deep dive. Let’s figure out how Cardinal drafts, specifically, have fared through the years compared to expectation. I’m using 2003 as the beginning of the time frame since that’s when Jeff Luhnow came on board, kicking off the Cardinals modern era.
Step 1. I collected the first 15 rounds of draft picks, for all teams, for every draft since 1990 using the Baseball-Reference draft database.
Step 2. For each draftee, I collected their Fangraphs WAR (fWAR) for each single season after they were drafted. WAR1 is the first year after they were drafted, WAR2 is the second year, and so on up until year 22. For a real world example, Paul DeJong was drafted in 2015 and debuted with a 3.0 fWAR in 2017, followed by a 3.3 fWAR in 2018. Thus, DeJong’s row has a WAR2 of 3.0 and a WAR3 of 3.3.
Step 3. We need a framework for expected career WAR. There are a lot of ways to do this. Former VEB writer Erik Manning once tackled it for Beyond the Box Score, and Matthew Murphy performed some research for The Hardball Times. I’m going to use Sky Andrecheck’s model, developed in 2009. The beauty of the Andrecheck model is that it gives us an expected career WAR, it evolves based on the overall pick number, and his research goes all the way to pick #500. Since I have 15 rounds of draft picks, my sample set fits neatly into his research. There’s one more positive. Andrecheck’s formula adjusts based on whether or not the draftee is a college or high school player, and a position player or a pitcher. An expected career WAR using Andrecheck’s model was produced for each draftee.
Step 4. Finally, I needed to adjust Andrecheck’s career WAR into a year by year expectation. After all, the Andrecheck model gives us a career WAR expectation, but many of the drafts since 2003 have yet to bear full fruit. Jack Flaherty, for instance, was drafted in 2014, and isn’t going to have anywhere near his career WAR. To adjust, I needed to find out what percentage of career WAR the average player would have through X years since being drafted.
Using all draftees from 1990 to 2000, I put together a simple aging curve for the four categories- high school pitcher, college pitcher, high school position player, college position player. For a frame of reference, the average college pitcher will have collected 21.9% of his career WAR through the fourth year after being drafted. A high school pitcher, on the other hand, will have 5.6% of his career WAR through the fourth year after being drafted. With this information in hand, I multiplied Andrecheck’s expected WAR by the percentage of career WAR the average player, by category, would have amassed in the number of years since they’d been drafted.
Using Flaherty as an example, he was the 34th overall pick, and a high school pitcher. Using the Andrecheck model, that player, on average, would produce a 2.16 career WAR. On average, a high school pitcher will have 5.6% of their career WAR through their fourth post-draft year. For Flaherty, then, his expected WAR through 2018 would be 2.16 x 5.6%, or 0.12 WAR. In Flaherty’s case, he posted a 2.3 fWAR last season alone. He has produced 2.18 more WAR than expected. In fact, he has already surpassed the expected level of career production in just his first full season.
For the record, I know there are flaws in this system. For instance, the aging curve is based on players drafted from 1990 through 2000. And Andrecheck’s model from 2009 could probably use an update. That said, the flaws are evenly applied across all draft picks in the sample. There’s no inconsistency in the way my process is applied.
Actual vs. Expected
To gauge how well each team has done in the league, I’ve added the actual career WAR for all draftees for each team, and then divided it by the expected career WAR for the same players. I’ll let the graph get us started. Here is how the Cardinals, in red, have ranked in actual vs. expected value.
The third place finish in 2009 is impressive on its own merits, but it becomes even more impressive when you consider that these results only represent the first 15 rounds. The 2009 draft also netted Trevor Rosenthal and Matt Adams in the 21st and 23rd rounds, respectively. Combined with Shelby Miller, Joe Kelly, and Matt Carpenter at the front of the draft, it’s a massive haul.
The 2011, 2012, and 2014 drafts all land in the top 10 on the backs of Kolten Wong and Seth Maness (2011); Michael Wacha, Stephen Piscotty, Kyle Barraclough, and Tim Cooney (2012); and Jack Flaherty, Luke Weaver, Austin Gomber, and Daniel Poncedeleon (2014).
It’s too early to properly gauge the 2015 and 2016 drafts, but the early returns are very positive. The Correa draft in 2015 yielded Jordan Hicks, Paul DeJong, and Harrison Bader. The latter two have already surpassed their expected career WAR based on draft position and demographics. Hicks is halfway there before his 23rd birthday. The high ranking for 2016 is exclusively caused by Dakota Hudson’s early arrival. Only 10 players from that draft have even reached MLB at this point. That being the case, don’t read anything into that ranking.
The average Cardinals rank since 2003 is 13.6, fifth in baseball. In order, they trail Arizona, the Cubs, Colorado, and Boston. They’ve been one of the very best in the game. Anecdotally, I suspect they’d climb higher if my data set had expanded past the 15th round. In addition to the aforementioned Rosenthal and Adams, the Cardinals have drafted several future big leaguers after the 15th round. That list would include Luke Voit (2013, 22nd round), Ryan Sherriff (2011, 28th), Kevin Siegrist (2008, 41st), Sam Freeman (2008, 32nd), Tommy Pham (2006, 16th), Luke Gregerson (2006, 28th), Jaime Garcia (2005, 22nd), and Jason Motte (2003, 19th). It’s hard to imagine any other team has consistently collected more talent out of the back of the draft.
Of Cardinal draft picks in the first 15 rounds, which players have most exceeded their expected value? Here they are ranked by WAR divided by expected WAR (expWAR).
Best Cardinal Draft Values Since 2003
This is strictly since 2003. Obviously, the best value pick the Cardinals have ever made is Albert Pujols, who has exceeded his his expected career WAR by... well, by an inner circle Hall of Fame career. Since 2003, Paul DeJong’s WAR/expWAR represents the third best value in all of baseball behind only Mookie Betts and Cody Bellinger. Matt Carpenter ranks 10th, Bader ranks 13th, Hicks is 28th, and Flaherty is 38th. That’s out of 1,584 draft picks since 2003 who have reached MLB.
Those numbers will shift as Bader, Hicks, Flaherty, and DeJong age. The older a player gets, the higher a percentage of their expected WAR they have to collect. Keeping with the DeJong example, through his 3rd year as a college hitter draftee, he is expected so far to only have accrued 4.67% of his eventual career WAR. By the end of 2019, that percentage rockets up to 10.8%. To maintain his current value ranking (WAR/expWAR), he would have to have an MVP season approaching 9 fWAR.
Most likely, that’s the lesson in here. Where the Cardinals have excelled in recent years is in aggressively pushing their youngsters to the MLB level. When they’ve pushed their youth to the MLB squad, they have succeeded at slightly younger stages in their development than happens for prospects with other teams. They excel at collecting MLB value earlier in the careers of their prospects.
Let’s wrap up with an easy metric. How about simply converting draft picks into Major League players? Here is how the Cardinals (again in red) rank in all of baseball in percentage of draft picks who reach MLB, by year, for the first 15 rounds.
You’ll see I’ve excluded 2017 and 2018, as it’s obviously much too early to gauge those drafts. It’s probably too early for 2015 and 2016 as well, but there are at least some draftees from those years arriving already. They’ve been very good at converting draft picks into Major League players other than a dud in 2013. Most impressive is how steady it’s been. In 9 of 11 seasons from 2006 to 2016, they’ve been a top 10 team, with four of them falling in the top 5. Their average rank since 2003 is 10.8, which trails only the Red Sox at 10.33. Since 2006, the Cardinals have the highest average rank in the league.
With all of this data, we can start to draw soft conclusions. First, there’s a connotation that the Cardinals do more with less in the draft. There’s some truth to that, but it’s not as obvious an edge as you might expect. In fact, several other teams have outperformed them from the standpoint of an average ranking. It shouldn’t be lost on you that of the four teams ranking ahead of the Cardinals, three of them have been led at various times since 2003 by either Theo Epstein or members of the Epstein family tree.
Second, where the Cardinals truly excel- particularly in recent years- is in shepherding their players to St. Louis, and placing them in roles where they excel earlier than the rest of the league. Third, there’s a volume game at play. It’s no surprise they have as much depth, consistently, as they’ve had for the last decade or so. They’re consistently placing a high number of draftees on the Major League squad, a demonstrably higher amount than all other teams except the Red Sox. Their limited draft slot prevents them from reaching higher quality players with any frequency, but they do their best to make up for it with raw depth.
Finally, I think there’s something to be said for what isn’t shown here. After seeing all of the data, I wish I’d expanded my data set to the full draft. I imagine that data would paint a different picture. Also not shown is the “brain drain,” which could have sunk their success in draft and development. The loss of Dan Kantrovitz, Jeff Luhnow, Sig Mejdal, Mike Elias, and Chris Correa, amongst others, was surely hard to weather. You can’t really capture something like that with this data. Even if you could, it would take several years to properly gauge. On the face, they seem to have survived and even found new ways to thrive.