Last week, we gave the background on changes in player value in dynasty leagues based on changes in ADP over time. I encourage you to go back to read it. In this article, we do an analysis of players taken in the middle and late rounds of the dynasty drafts to see how fantasy owners selected players and how those players’ values changed two years later.
Deep Dive – Middle Rounds (ADP of 91 through 450, i.e. 7th thru 30th Rounds)
The Middle Rounds were examined first for a number of reasons. Before beginning the analysis, our expectation was that dynasty owners’ selections in the Early Rounds (picks 1 through 90) are generally uncontroversial. It’s human nature to spend more focus on the sexy picks in the first few rounds (so the picks will generally be ‘good’). Also, the general sentiment is that most of the elite picks in a redraft – even if they were older (such as Chris Sale, Justin Verlander, J.D. Martinez) – generally still produce at a high level the following year. Yes, they may decline, but there wouldn’t be that many falls-off-a-cliff. That’s why they’re early picks after all. In other words, we figured the review of the Early Rounds would be unexciting. Similarly, we expected that the Later Rounds (pick 451 and onward) generally consisted of roster filler and ‘lottery ticket’ prospect selections (the vast majority of whom wouldn’t pan out). Again, nothing interesting was expected.
Therefore, the ‘meat’ of the draft – where it’s believed that leagues are won or lost – was our first focus. We wanted to see how the value of players selected by owners in the Middle Rounds (defined as picks 91 through 450, or rounds 7 through 30), changed when viewed over a two year time horizon.
From the data set described in Part 1, there were six hundred and fifty nine (659) players selected in the Middle Rounds. These split approximately into the same quartiles as in the overall analysis (though with slightly fewer players taken in the youngest cohort):
- Age 17-22 Quartile (n=128, 19% of pool)
- Age 23-25 Quartile (n=170, 26% of pool)
- Age 26-29 Quartile (n=185, 28% of pool)
- Age 30+ Quartile (n=176, 27% of pool)
Again, as would be expected, the older the age cohort, the more likely they would bust (i.e. not be selected two years later):
Survivor Analysis – Middle Rounds
Below is the scatterplot of non-busts from the Middle Round (n=580; excluding the 79 busts) using the same conventions for color and size of the dot as in the overall player pool plot in Part 1.
As was seen in the overall data set (note that the Middle Round plot is literally a vertical slice of the overall ADP plot), most of the dots that are above the line are green; ie younger. Also we can see that several of these selections actually jumped up in value to become “elite” picks two years later (i.e. close to the top of the graph). In other words, there is an opportunity for significant value to be extracted within the Middle Rounds.
Zooming in on the players whose ADP went from the Middle Rounds (7th to 30th) into the first six (6) rounds two years later (ie only examining those with a y-value (ADPYEAR + 2) of between 1 and 90), we see that there were 63 players (out of 580, or 11%) who accomplished that feat:
Of the 63 players, 57 (90%) were age 27 or younger and only six (6, or 10%) were over the age of 27.
Making the case even stronger against the ‘older’ players was that the six (6) older players who reached the Early Rounds (two years later) were clustered near the early section of the Middle Rounds already (ie. had an early ADP to begin with). In other words, they didn’t jump more than approximately 3 rounds in ADP (average improvement of ~56 in ADP). They were:
- Khris Davis (ADP of 175 –> 87)
- James Paxton (ADP of 131 –> 80)
- Khris Davis (ADP of 129 –> 51)
- Brian Dozier (ADP of 103 –> 43)
- Justin Upton (ADP of 117 –> 79)
- Yu Darvish (ADP of 97 –> 78)
Meanwhile, many of the younger players who made the jump into the Early Rounds also had BIGGER jumps; e.g. Juan Soto (from an ADP 365 to 6), Vlad Guerrero Jr. (416 –> 21), Aaron Judge (312 –> 9), Cody Bellinger (336 –> 15), Victor Robles (302 –> 20), Alex Bregman (307 –> 25). In fact, the players under age 27 who jumped into the Early Rounds had an average ADP improvement of ~183 picks.
Younger Players (25 or younger) vs Older Players (26 or older)
To explore further the significance of player age on future value, we bisected the Middle Round player pool into a ‘younger half’ (age 25 or younger, n=280) and the ‘older half’ (age 26 or older, n=300).
This can be seen very clearly in the two plots below; the one on the left is comprised of players age 25 or younger whereas the plot on the right is comprised only of players age 26 or older:
Not only can we see that the plot (on the left) of younger players has a much higher concentration of dots above the ADP line but it also has far more dots situated near the top of the graph (i.e.with an Early Round ADPYEAR + 2); conversely, the plot (on the right) of older players has far fewer dots above the line (and even fewer near the top of the graph).
Of the players age 25 or younger, about half of them (136, or 49%) improved their ADP two years later; whereas of the players age 26 or older, only 20% of them (59 out of 300) improved their ADP over the same time span. In other words, it is very unlikely (only 1 in 5 chance) that a player age 26 or older who is taken in the middle rounds improves (or even retains) their value over two years.
For an even deeper dive, the two (‘younger’ and ‘older’) halves were bisected again (into the same age quartiles used earlier). Within each of these age cohorts (Age 17-22, Age 23-25, Age 26-29, Age30+), we ranked the players in order of ADP improvement and then created quintiles of equivalent player sizes (including busts).
The average ADP increase/decrease within the quintile (including busts) is tabulated below (positive ADP means ‘improvement’):
The above table can be shown most strikingly on the following bar graphs (with each quintile’s average two year ADP change plotted along the same x-axis (where a bust was considered to be an ADPYEAR+2 of 1500)
Note: The names are an example of the player with ‘median’ ADP performance within the quintile and the player whose name is in (parentheses) had the ‘best’ performance within the quintile:
Above, we see that the top quintile of the age 17 to 22 cohort had an average ADP improvement of 257 (with Gleyber Torres being the player within the quintile showing the median ADP improvement and Vlad Guerrero Jr being the player within the quintile that had the largest two year ADP improvement). Similarly, of the players in the middle quintile, Alex Reyes had an average ADP improvement of 19 (and Amed Rosario had the best performance).
I think these graphs show quite effectively that essentially any player taken in the Middle Rounds who is over 30 years old (the last graph) will lose value over two years. Only the top 20% performing players retain their value (ie stay at the same ADP). The other 80% worsen in value by an average ADP of approximately 700(!). And the age 26-29 cohort (third graph) doesn’t do that much better: 80% of the players (aka the bottom 4 quintiles) also lose value in two years.
The younger players perform remarkably better overall (the top 20% in each age cohort average an ADP jump of over 200 slots!)
Younger Players (age 25 or younger): Hitters vs Pitchers
Clearly then, to maximize value over two years, the data shows that younger players should be targeted in the Middle Rounds of a draft. Does it matter if these young players are hitters or pitchers? Let’s separate the younger players (age 25 or younger) into hitters and pitchers:
Again, this strikingly shows that shows that the majority (58%) of younger hitters (from the graph on the left) selected in the Middle Rounds (7th to 30th) improve their value, and they often vastly improve two years later (as can be observed by the clusters near the top of the graph on the left). In fact, 40 out of 165 young hitters (or nearly 25%!) became an elite dynasty asset (ie an Early Round selection) in two years.
Meanwhile, pitchers (from the graph on the right) do not tend to show an ability to be considered an elite player (and are hence rarely selected two years later by owners in the Early Rounds) and, in fact, they only sparsely populate the plot above the line at all (with only 33% of the young pitchers even increasing in value two years later).
Conclusion (Middle Rounds)
In order to maximize value in the Middle Rounds, avoid selecting players age 26 or older. This surprised us (as we expected that the age threshold for players – when taking a two year window into consideration – would have been around age 28 or age 29).
To further maximize retained or improved value, focus on hitters over pitchers. Young bats. Young bats. Young bats.
Deep Dive – Later Rounds (ADP of 451+, i.e. 31st Round onward)
The same analysis that was performed above in the Middle Rounds was repeated for the players selected in the Later Rounds. Again, the conventional wisdom is that owners use these rounds to take their lottery ticket picks and fill out their roster. Busts are expected and therefore risks can be taken – in hopes that one of these Later Round picks (31st round or later) jump up in value. The breakdown of players selected in the Later Rounds (after pick 450 is tabulated below (n=419):
The above is depicted below in a histogram of players based on their age at the time of their selection:
Interestingly, it is more heavily skewed (with younger players (n=258) taken more often than older players (n=161)). This supports the idea that many of the Later Round picks are populated with prospects.
Although the previously used age cohorts are not equivalently-sized in this player data set, the same buckets will be used (for an apples-to-apples comparison with the other analyses). As a reminder, they are age 17-22, age 23-25, age 26-29, and age 30+
The bust rate of each age cohort is depicted graphically here:
As expected, these Later Round selections have a significantly higher likelihood of busting (compared to the Middle Round selections). Also, again, the older the player, the more likely that they bust two years later (with the players who are age 30 or older busting 56% of the time).
Survivor Analysis – Later Rounds (ADP of 451+)
The subset of players selected in the Later Rounds (ADP of 451 or later) who didn’t bust is graphically depicted below (n=296 survivors, excluding the 123 busts).
Although finding it initially surprising that there are a reasonably high proportion of Later Round selections (54%) that improved in value (ie are above the y=x [ADP, ADPYEAR + 2] line), upon further investigation, it shouldn’t be all that surprising: because there is a marginal perceived difference between, for example, a 48th round pick and a 43rd round pick, any observed ‘improvement’ may not actually be all that real-world significant.
Let’s dig deeper into the Later Rounds and perform the same age cohort quintile distribution analysis as we did with the Middle Rounds.
Later Rounds – Quintile Distributions of the Age ‘Quartiles’
The average ADP increase/decrease within the performance quintiles of each age cohort (including busts) is tabulated below (again, positive ADP change means “improvement”):
Again, the player names depicted below on the graphs represent the ‘median’ ADP improvement within the quintile and the player name in (parentheses) is the player who had the ‘best’ ADP improvement (within the quintile). For example, in the age 17 to 22 cohort, the players in the top 20% performing quintile (aka the 80th percentile) averaged a 450 ADP improvement (with Edwin Diaz having the best performance within the quintile and Luke Weaver having ‘median’ ADP improvement within the quintile).
Here we can see that taking players age 22 or younger in the Later Rounds generally pays off 50% of the time The top 3 quintiles (comprised of 60% of players) increased in value in two years (by an average of 259 picks of ~17 rounds.
As expected, the older the player, the more likely they will both reduce in value and reduce in value by a greater amount). But surprisingly, the top 20% performers in each age cohort (with the exception of the oldest age cohort) each had significantly high average ADP jumps (of an average of approximately 400 picks (or 26 rounds)). This seems to support the idea that lottery tickets can be found with reasonably high frequency (around 20%) in all (but the oldest) age groups.
Also, as usual, the players who are 30+ years old continue to be terrible bets. Even if you happen to select one of the top 20% players from the age 30+ cohort, they only increase in value by an average ADP of 57 (less than 4 rounds of improvement) from an ADP of 533 to pick 477. This is a marginal increase in value with little real-world significance. Therefore, one should have no illusions that taking an older player in the Later Rounds is anything more than filling out a roster spot (and only for a year or so).
Although not shown here, as opposed to the Middle Round analysis, there was no observed difference between hitters or pitchers. This was mostly attributed to the fact that such a high percentage of players selected (29%, nearly 1 in 3) of both hitters and pitchers ‘bust’ in two years.
The “Big Jumpers” (change in ADP of > 300)
There were sixty four (64) players that jumped more than 20 rounds (an improvement in ADP > 300) two years later. This can be illustrated by adding a line (dashed in red) corresponding to translating the original y=x trend-line upward by an ADP of 200 and observing the dots above the new line.
Only fifteen (or 23%) of the “big jumpers” (above the red line) were age 26 or older; the other 49 (or 77%) were age 25 or younger. The older players, for interest’s sake, were:
- SP Mike Clevinger (ADP 641 –> 76)
- C/2B Austin Barnes (ADP 728 –> 174)
- RP Kelvin Herrera (ADP 654 –> 261)
- SP Andrew Heaney (ADP 565 –> 225)
- SP Matthew Boyd (ADP 694 –> 358)
- RP Kyle Barraclough (ADP 733 –> 415)
- RP Brad Hand (ADP 724 –> 140)
- SP Patrick Corbin (ADP 583 –> 50)
- RP Alex Colome (ADP 640 –> 228)
- SP Drew Pomeranz (ADP 661 –> 268)
- SS Andrelton Simmons (ADP 661 –> 285)
- SP James Paxton (ADP 499 –> 129)
- 3B/2B/1B Travis Shaw (ADP 458 –> 119)
- RP Blake Treinen (ADP 556 –> 132)
- SP Hyun-Jin Ryu (ADP 635 –> 302)
Interestingly, most of these “older” breakouts are pitchers that fall within one of the following categories:
- Relief Pitchers who got the Closer role [Barraclough, Colome, Herrera, Treinen]
- Starting Pitchers who got a permanent role in the rotation [Clevinger, Heaney, Boyd, Corbin, Pomeranz, Paxton, Ryu]).
This may suggest that older pitchers may be worth a flier in the Later Rounds as they seem to be the type of older player that has the potential for a large jump.
Jumpers into the Early Rounds (ADPYEAR + 2 of < 91)
Only fifteen (15) players (5%) of the Later Round selections became a ‘top 100 pick’ two years later. Again, for interest’s sake, these are the fifteen (15) players selected with an ADP later than 450 and who became Early Round picks two years later:
- Walker Buehler (pick 488 at age 22 –> pick 27, two years later)
- Jose Ramirez (pick 488 at age 23 –> pick 31)
- Rhys Hoskins (pick 471 at age 23 –> pick 38)
- Matt Chapman (pick 494 at age 24 –> pick 43)
- Eloy Jimenez (pick 566 at age 19 –> pick 44)
- Patrick Corbin (pick 583 at the ripe old age of 27 –> pick 50)
- Robbie Ray (pick 464 at age 24 –> pick 53)
- Jack Flaherty (pick 552 at age 21 –> pick 61)
- Luis Castillo (pick 481 at age 24 –> pick 66)
- Fernando Tatis Jr (pick 634 at age 18 –> pick 68)
- German Marquez (pick 466 at age 22 –> pick 71
- Mike Clevinger (pick 641 at age 26 –> pick 76)
- Sean Newcomb (pick 513 at age 24 –> pick 90)
- Jake Lamb (pick 503 at age 25 –> pick 95)
- Bo Bichette (pick 535 at age 19 –> pick 100)
Note that the oldest player who made the leap from Later Round selection to be an Early Round selection two years later was age 27 (Patrick Corbin).
Conclusion (Later Rounds)
In the Later Rounds, players over the age of 30 rarely retain value over two years. In fact, they have a greater than 56% chance of not even being rostered two years later.
There is a very low chance (5%) that a player taken in the Later Rounds becomes an elite dynasty selection in two years. Still, within the age 17-22 cohort, the age 23-25 cohort, and the age 26-29 cohort, the top performing 20% players within the cohort averaged an ADP improvement of ~400. The best performing age cohort was the age 17-22 group where the top performing 20% of players within the group averaged an ADP improvement of over 30 rounds.
If an older player selected in the Later Rounds were to jump significantly in value, it is more likely that they are a pitcher. Next week in Part 3, we will complete the analysis by diving into the Early Rounds (rounds 1 through 6) and identify overall conclusions of how we can apply these findings.
Leave a Reply
Friends don't let friends talk to themselves.
You must be logged in to post a comment.