Last March at RotoBaller, we kicked off a cool fantasy baseball research project, exploring what we're calling Expected Draft Values. This offseason, we’ve refined the approach and our research, and taken things a step forward. We're sharing this concept with the fantasy baseball world now, and hope it's as helpful to you as it's been to us.
A huge kudos to Nick Mariano (2018's Most Accurate MLB Draft Ranker) for leading this effort, and having the utmost patience for all my questions, ideas, and waffling throughout this project. You're the man, Nick.
Expected Draft Values is one of the more practically useful fantasy baseball data sets to be produced. It's being used by our writers in their articles and analysis, and now it can be used by you, dear readers, for your draft preparation.
Editor's Note: Our incredible team of writers received five total writing awards and 13 award nominations by the Fantasy Sports Writers Association, tops in the industry! Congrats to all the award winners and nominees including Best MLB Series, NFL Series, NBA Writer, PGA Writer and Player Notes writer of the year. Be sure to follow their analysis, rankings and advice all year long, and win big with RotoBaller! Read More!
What Are Expected Draft Values?
So frequently, we hear fantasy analysts say “Player X is a great value at that ADP.” How do they really know that? And more importantly, just how good of a value is that player? If you're deciding between two "value" picks during a draft, you would want to know which returns more value!
Expected Draft Values (EDV) try to provide concrete answers to these questions, so you're informed as to whether Player X not only provides profits, but how big a profit. Put another way, EDV answers the question:
“What sort of stat production do I need, at a given draft pick, in order to break even, or turn a profit?”
How Do Expected Draft Values Do This?
EDV provides historically-averaged 5x5 Roto stat lines for every draft slot, to help you understand the type of production you should be expecting with any given draft pick. Technically, we averaged out 1,000+ player-seasons and resulting Yahoo overall ranks from the past 5 years, sprinkled in some data smoothing, and came up with a data set that reliably shows the average stat line for the 10th best player, 20th best, 100th best, etc.
Not all players are made equal, though. If you're thinking about drafting a steals-first player, you need to reference that against other steals-first players. Comparing your steals target against power hitter stat lines wouldn't be so helpful, would it?
To make Expected Draft Values easier, we divided up the player pool into seven cohorts:
1) HR+BA+SB 2) HR+BA, 3) SB+HR, 4) SB+BA, 5) HR, 6) SB, 7) BA.
These seven cohorts are not perfect, as many players don't neatly fall into one of them. However, it was important to try and approximate the different types of hitters one targets. Too few cohorts and it would be difficult to find player comps, too many and we wouldn't have enough player-seasons in a given cohort to make the data viable. This is one area we may tweak and improve upon in the future. Here is how we defined the cohorts:
Cohorts | Metric1 | Metric1 Value | Metric2 | Metric2 Value | Metric3 | Metric3 Value | % of Total |
SB | SB | 13 | 19.73% | ||||
HR | HR | 26 | 20.70% | ||||
BA | BA | 0.288 | 20.55% | ||||
HR+SB | HR | 20 | SB | 9 | 10.35% | ||
BA+HR | HR | 20 | BA | 0.284 | 10.87% | ||
BA+SB | SB | 9 | BA | 0.284 | 10.13% | ||
BA+HR+SB | BA | 0.28 | HR | 18 | SB | 9 | 5.36% |
With these cohorts, when you use Expected Draft Values for a player like Joey Gallo, you're able to reference highly-relevant comparisons of other power hitters, rather than a general averaged 5x5 stat line. If you're wondering why we didn't factor RBI and runs into any cohort definitions, it's because RBI and runs are mostly a product of batting order, lineup, and the batting average and power of a player.
Expected Draft Values In Action
Here’s a quick example of how EDVs can be used with one of my draft targets last year and this year: Joey Gallo. Gallo, a power cohort staple, has an NFBC ADP of 80.6 as of Feb. 8 (he's lower on other platforms). Many attribute his ADP to his poor BA, but most analysts are largely guessing as to how much negative value is driven by his batting average. In my opinion, which is supported by EDV, most over-weigh the negative value of Gallo's BA and consequently misjudge how much value Gallo’s pop provides.
Luckily, we can reference our Expected Draft Values research to clear this up. If we go to the HR cohort and follow it down to row 81, we reach the average stat line for players who finished the season ranked 80th overall. The important assumption here is that if you're drafting Gallo at 80th overall, you're expecting a player who'll finish the year ranked at least 80th overall, otherwise you're taking a loss on that draft pick.
What we see in the power cohort of the EDV is that, on average, power hitters who finished the year ranked 80th have produced a stat line of .265-31-88-85-4. So, if we draft Gallo at his ADP of 80 and he outdoes that stat line, we profit. That's the power of EDV, we have a clear and simple break-even point that's rooted in real results around which to make judgments on value.
But the real power of EDV comes when we combine the EDV with either a ranking or a projection. RotoBaller’s esteemed rankers have Gallo at 56 overall, yielding a profit opportunity at an ADP of 80. If you prefer a comp with real projections instead of our rankings, RotoBaller's Nick Mariano projects a .232-44-101-89-8 line in 607 PAs for Gallo. If we compare Gallo's projected stat line to his break-even point (265-31-88-85-4), it shows less BA than we need (.232 vs .265, a 13% deficit), but significantly more HR (41% gain), a bunch more RBI (14% gain), a few more R (5% gain), and double the SB (100% gain).
Granted, we can't simply average these percentage gains / losses together. Because of volume, Gallo's 100% 4-SB gain is way less impactful than his 13% batting average loss. Nevertheless, it's clear that what Gallo loses in batting average, compared to the EDV break-even point, he more than makes up for in the other categories.
Additionally, Nick's projection assumes 145 games for Gallo, so a fully healthy season for Gallo would see his stat line top Nick's projection for him, and widen the gap even further from the EDV break-even point. Bottom line, if you believe in Nick's projection for Gallo being close, then he's a great value at his current ADP.
But just how good of a value is Gallo? This is where it gets interesting, as finding an exact comp for Gallo's projection in our EDV stat lines is not simple. As the draft cost column gets more expensive, BAs tend to rise. That makes sense - better ranked players usually have better BAs.
Gallo is a unique player, and in a perfect world we would have a "+Power -BA" cohort for him. Without that, we have to approximate and find our way to the ~48-50 overall range, where we find where the projected BA loss seems to even out with his HR / RBI / R / SB profit. And voila, we've used Expected Draft Values to find that Gallo's projected stat line sits near a 48-50th overall ranked player, one taken in the early part of the 5th round in a 12-team draft.
Quick Note on Draft Cost / Rankings
The Rankings (Draft Cost) we used came from Baseball Monster. We find them quite similar to Yahoo!, and superior to ESPN and CBS. Since we spent a lot of time averaging and smoothing the EDV data, we like to think that BBM's rankings would mostly converge with other platforms, and thus using BBM is as good as using any other platform's rankings. This is one area of this project which can and will be improved next year. Real fantasy rankings need to be league dependent, because the number of started / owned players in a given league influences the relative scarcity of stat categories, which impacts player rankings. Next year, we'll roll out EDVs which can be adjusted for league depth.
EDV Summarized
EDV shows us that Joey Gallo is being drafted in a spot (80) that has historically returned a stat line which he is projected to beat. Ergo, Joey Gallo is a solid draft target based on his current draft price. Use this to inform potential targets as you go.
A slightly longer explanation: Gallo's ADP at 80 seems a bit low right now. Gallo's EDV (the expected production for a power hitter taken 80th overall) is 265-31-88-85-4. If healthy, RotoBaller's projection for Gallo show he should have an easy time beating this break-even stat line, making him a a fine target after the top 50 picks are off the board, and a great target after the top 60. Additionally, RotoBaller has Gallo ranked at 56, further cementing the notion that Gallo is a nice value at his current ADP.
What about Pitchers?
Pitchers were way, way, way easier to approach for EDVs. This is because there really aren't different "types" of pitchers fantasy managers target, thus no need to break them out into cohorts. You might be thinking of Miles Mikolas and Kyle Hendricks, low-ERA and low-K guys. They do exist of course, but most high-K pitchers are also going to be low-ERA / low-WHIP pitchers, and low-ERA / low-WHIP are also going to be high-K pitchers. Here are some interesting observations with the Pitching data set:
- Elite pitching is worth paying for (surprise!)
- The "load up on cheap elite RPs" strategy is legit. Many more relievers than are actually drafted end up returning value equal to starters drafted in the mid-to-late rounds.
- Once you get past the Top ~110 players, inning-eater pitchers tend to be over-drafted. It would be smarter to target higher-upside low-IP guys (Rich Hill and Hyun-Jin Ryu are classic examples).
What Are the Different Ways Expected Draft Values Might Be Used? Q&A Time...
- If you're deciding between three players in a draft slot, could EDV be utilized to indicate which player is going to provide you with better value at this pick?
- If you are utilizing projections in your draft prep, then you can match a player's projection up with the best EDV comp, which informs the overall value they are projected to return. If all three players are available at the same point in the draft, then whichever one of them projects for the highest overall value is likely your best bet, all things equal. In that sense, we're using EDV to simply turn a player's projection into an overall projected rank as an alternative to ADP.
- Can EDVs be used for determining whether a player's ADP is justified? Can it also be used for general draft strategy? If one decided to use EDV as a draft strategy, they would just become a "best value available" drafter, which isn't ideal, right?
- Correct, EDV, when combined with rankings or projections, can be used for determining whether a player is a good value at their ADP, and how good of a value they are.
- General strategy should come first in any draft before a simple "best available player" approach. Roster construction is one of the most critical factors when drafting, and may influence you to go "off board" at times.
- The Speed cohort doesn't start until "draft cost 73", does that mean I shouldn't draft a speedster until pick 73?
- No. A better way to interpret the data is that a "speed-only" guy shouldn't need to be taken before pick 70 because historically, players who finished ranked 70 or higher were in the other cohorts (SB+BA, SB+HR, SB+HR+BA), meaning they returned more value than a speed-only guy would. In other words, never draft a Mallex Smith mold in the top 70. Even if you think you're locking up a category, you're setting yourself up for a net loss. This may feel intuitive, but it's nice to see objective confirmation.
- How do we find the best EDV comps for the guys that are average (maybe above-average) across the board? I.e. what cohort would we place Andrew Benintendi in, and is his ADP (107) outlandish right now?
- Benintendi's projection (272-18-77-96-13) makes it particularly hard to comp him to any EDV. He's a five-category guy by virtue of not being really good or really bad at anything, but not good enough in BA to meet the BA+HR+SB tier. My guess is, if he returns his projection, he'd be fair value around pick ~130.
- Is EDV just as reliable to spot small values as it is to spot big values?
- Trying to find value with a 10-pick profit is a tough exercise, for two reasons. One, the profit can be wiped out by a small AB increase or decrease. Also, you'll notice that sometimes in the EDV data, the stat lines in 10 consecutive rows are quite similar, so finding exactly where a player belongs is a tough task to begin with. EDV is much more reliable in finding clear value gaps, as Gallo illustrates.
Some Concluding Thoughts
Many fantasy managers are swayed by brand names, for better or worse. This may help with some popular sleepers, but often hurts because the herd mentality causes many managers to target the same players, drive up their prices, and over-target well-known players, even if they’re declining. Our cognitive biases lead us to be overly reactive in fantasy drafts, reacting to last season(s), the last day or week of news, or the number of times we heard a name in draft season, at the expense of the larger picture.
EDV aims to help with these common draft problems, by establishing break-even baselines for every type of player for every draft slot. With these baselines in hand, we can make better informed decisions at every point with both draft prep and in the draft room itself.
More Fantasy Baseball Analysis