How Much Is a Draft Pick Worth in 2014? by Matthew Murphy May 21, 2014 Justin Verlander was the only league-average player drafted in 2004’s top 10 picks (via Keith Allison). Much was made during the past offseason about the value of draft picks. Under the new Collective Bargaining Agreement, a player who turns down a qualifying offer will cost the team that signs him a draft pick – its highest pick outside the top 10, which are protected. While a number of free agents seemed unaffected by this additional cost, three players ended up signing deals of equal or lesser value than the initial qualifying offer (and one of the three had to wait two months into the season to do so), and a fourth player remains unsigned. This raises an important question – are major league teams and fans properly valuing draft picks? A number of factors make this a difficult question to answer. What is the team’s financial situation? A small-market team that relies on homegrown talent might place a premium on draft picks, since it won’t be able to afford the high-priced free agents. Is the team a contender or a rebuilder? A team that looks to be competitive in the present and immediate future may not worry as much about a value that won’t be felt for three or more years. Meanwhile, a rebuilding team might not mind having to wait a few years to get a return on their its investment, but may actually prefer it, as the cost-controlled player’s contributions will be more meaningful. With these variables, it’s easy to see why such a wide range of values gets thrown around for draft picks. This past March on THT, Matt Swartz estimated that the 30th pick of the draft was worth about $6.7 million of present value. On the other hand, Dan Szymborski of ESPN wrote that the 26th pick that the Braves gave up to sign Ervin Santana was worth $19.6 million. To better understand the true value of a draft pick and explain the differences between these projections, we need to do three things: Estimate the total value that each draft pick will produce, and how that value will be distributed Calculate the net value of each draft pick (which means estimating how players are compensated relative to their performance while they are under team control) Determine how value of the drafted players changes over time Today, we’ll tackle the first question. We’ll tackle the second and third tomorrow, where we’ll fine-tune the model to project net value and tackle the issue of time, in addition to looking at how draft pick compensation affected free agents this past offseason. And on Friday, we’ll bring everything together. How Much WAR Does Each Draft Pick Produce? Luckily for us, this is a question that has been investigated by a number of baseball analysts. In 2009, Sky Andrecheck developed a model to estimate the average WAR that each draft pick produces, both in their career and their first six years of team control. If you’ve read anything about modeling player production by draft pick, you’ve probably seen this graph before: (The graph above is for Career WAR; the model for WAR under team control can be found here.) The first thing you might notice is how quickly production drops off. On average, only the top few picks will average at least one WAR per year while they are under team control. In the compensation round and second round, the average draftee isn’t even cracking two total wins before hitting free agency. When accounting for the recent adjustment of the WAR baseline by FanGraphs and Baseball-Reference, this model gives us an average of 7.65 WAR for a top-five draft pick during his team-controlled years, while a late first-rounder (picks 21-30) produces an average of just 2.24 WAR. Over the years, Andrecheck’s work has become the gold standard for modeling production of draft picks, but many others have performed similar work. Erik Manning at Beyond the Box Score performed similar research in 2009, using the 1990-1999 draft classes. In that span, draft pick value doesn’t appear to decline quite as rapidly after the first 10 picks as Sky’s model would predict. I compiled my own data set (using Baseball-Reference’s draft pick search tool), looking at 15 years of drafts starting with the most recent year in which the vast majority of early picks have reached free agency (2005). Cumulative WAR, pre-Free Agency Draft Pick Sky Model Manning, 1990-1999 1991-2005 1-5 7.7 8.3 8.5 6-10 4.1 5.9 6.1 11-15 3.2 4.6 5.6 16-20 2.7 5.0 4.3 21-30 2.2 1.7 2.6 31-40 1.9 – 1.8 41-50 1.6 – 1.9 It is important to remember that each year in the draft is extremely variable, even when looking at groups of picks. In 2002, the 21-30 draft picks matched their top-10 counterparts in production under team control. Meanwhile, only two players picked between 21st and 30th in 1999 produced any positive WAR for their team, and nobody surpassed four wins. The same variation is seen in the upper end of the draft. The top 10 picks in the 2004 draft produced only one league-average player (Justin Verlander). By contrast, the 2005 draft was extremely deep, with seven of the first 12 picks producing at least 15 WAR while under team control. This may not be news, but it’s something to keep in mind while evaluating draft picks and determining what the best model is to use moving forward. Sky’s model is based on decades of data, and given the extreme variability in the draft, it’s difficult to trust numbers from a five-, 10-, or even 15-year sample. At the same time, baseball has changed dramatically in the past couple decades. Improved scouting, analytics, and player development may help teams get the most out of their draft picks. The fact that the data from the most recent drafts look very similar to the numbers from the 1990s hints at the possibility that this is in fact a trend, rather than being a blip on the radar. Supporting this idea is the fact that evaluation of prospects is improving, as Chris Cwik investigated last week for Sports on Earth. While the article focuses primarily on the increased accuracy at ranking minor leaguers over the past 25 years, the same improvements in scouting and technology would likely apply to amateur players as well. If we perform a nonlinear regression on the 1991-2005 sample, we see the same power relationship between WAR and draft pick (R2 = 0.57). Here’s how the more recent drafts compare to Sky’s model: (For reference, the adjusted equation for Sky’s model is y = 12.03x-.52) As we saw with the raw numbers, it looks like teams may be getting better at finding quality players in the middle of the first round. Nearly every pick in the top 25 has exceeded expectations based on Sky’s model. However, right around the 25th pick, there appears to be a pretty steep drop-off, and the majority of the picks fall below both models. This actually makes some sense. Since the number of quality players in every draft is limited, an increase in average value at one point in the draft needs to be offset by a decrease elsewhere. This decrease may be seen in the later rounds, or there’s a chance that players who used to fall through the cracks to the end of the first round or the second round are now getting scooped in the top 25. While both equations have the same R2 value, the updated model is much closer to the actual average WAR of 3.71 for the first 50 picks over the past 15 years (3.76 for the updated model, 2.90 for Sky’s model). However, this improvement is based almost entirely on changes in the first 25 picks, and the old model actually does a better job of estimating WAR for players picked in the late first and early second rounds. Distribution of WAR, by Model Type Draft Pick Actual WAR Sky’s Model 1991-2005 Model 1-25 5.6 4.0 5.0 26-50 1.8 1.8 2.5 Average 3.7 2.9 3.8 You could make an argument for using the new model for the first 25 picks, then switch to Sky’s for the rest of the draft. While 15 years sounds like a lot of data, with the variability in the draft it’s always difficult to tell whether we’re seeing a true trend. For simplicity, I’ll use the new model (1991-2005) moving forward, but I will revisit this idea later. References & Resources WAR figures for draft picks from 1991-2005 are from Baseball-Reference’s draft tool. WAR figures for recent arbitration-eligible players are from FanGraphs. Arbitration salaries are from Baseball-Reference and Cot’s Baseball Contracts.