Draft Buddy has its own brand 2019 fantasy baseball projections. This is how we made it happen.
For the first time we have our own Draft Buddy brand fantasy baseball projections.
These projections are available in Draft Buddy draft software and Last Player Picked. They form the basis for our 2019 fantasy baseball rankings.
The projections are based on Steamer with our own playing time adjustments. Steamer projections from Fangraphs include playing time that is inconsistent with other trusted sources. This creates significant outliers in our ranking results.
If we improve the fantasy baseball projections then it ultimately improves the rankings output. That is the motivation for the Draft Buddy projections.
Steamer Base + Playing Time
Steamer is a proven solid resource for projecting hitter and pitcher metrics with its model dating back to 2008. There is little reason for me to pull Steamer apart.
However, the projected playing time is another story. Projected plate appearances for hitters and projected innings for pitchers is something to analyze for reasonableness, risk, etc. And, we can make changes to better reflect our best-educated guesses for the upcoming season.
Instead of “simply” going team-by-team through the depth charts and pegging a playing time projection for every player, I thought a better first exercise is to gather existing playing time data to review. This way, we can see the range of playing times for each player.
Fantasy baseball analysis naturally lends itself to a lot of number crunching. If you note a player you are particularly high or low on is at the opposite end of your expectations when you calculate his value against his projections, it may not be because he is a better or worse ballplayer than you thought. His underlying metrics may be right on par with your expectations. However, his projected playing time may be far different than you envisioned, suppressing or inflating his value.
If we review the range of projected playing times, then we can assess if a particular projection set is unusually high or low in expected playing time for a player. Once we determine our own playing time estimate for a player, we apply those to the Steamer projections to form our own projections set.
Collecting Playing Time Data
I began pulling projected playing time data from various sources. These were from projection sets already included in Draft Buddy: ATC, THE BAT, Steamer and Zeile from FantasyPros.
In addition, I collected playing time data from Razzball, by Rudy Gamble, who does a similar process for his projections, using Steamer and applying his own playing time adjustments. I also added Roster Resource, an excellent go-to for projected MLB depth charts and therefore a natural to estimate playing time.
After compiling this data in my baseball database, I produced playing time comparison reports by team for each of hitters and pitchers. Here is the Minnesota Twins hitters as an example:
+--------------------+------+-----+--------+--------+------+------+------+------+-------+-------+-------+------+-----+ | Name | Team | Pos | Roster | batOrd | ATC | BAT | Razz | RR | Steam | Zeile | Count | High | Low | +--------------------+------+-----+--------+--------+------+------+------+------+-------+-------+-------+------+-----+ | Jorge Polanco | MIN | SS | 25 | 1 | 617 | 614 | 635 | 656 | 641 | 606 | 6 | 656 | 606 | | C.J. Cron | MIN | 1B | 25 | 2 | 460 | 497 | 507 | 313 | 529 | 499 | 6 | 529 | 313 | | Eddie Rosario | MIN | LF | 25 | 3 | 623 | 635 | 654 | 604 | 622 | 599 | 6 | 654 | 599 | | Nelson Cruz | MIN | DH | 25 | 4 | 557 | 621 | 603 | 610 | 606 | 593 | 6 | 621 | 557 | | Max Kepler | MIN | RF | 25 | 5 | 600 | 607 | 579 | 592 | 594 | 569 | 6 | 607 | 569 | | Jonathan Schoop | MIN | 2B | 25 | 6 | 561 | 552 | 594 | 583 | 525 | 547 | 6 | 594 | 525 | | Miguel Sano | MIN | 3B | 25 | 7 | 506 | 586 | 573 | 565 | 545 | 544 | 6 | 586 | 506 | | Jason Castro | MIN | C | 25 | 8 | 307 | 410 | 311 | 367 | 339 | 296 | 6 | 410 | 296 | | Byron Buxton | MIN | CF | 25 | 9 | 479 | 517 | 488 | 512 | 484 | 504 | 6 | 517 | 479 | | Ehire Adrianza | MIN | 2B | 25 | | 229 | 145 | 109 | 64 | 177 | 225 | 6 | 229 | 64 | | Marwin Gonzalez | MIN | OF | 25 | | 511 | 595 | 516 | 515 | 542 | 523 | 6 | 595 | 511 | | Tyler Austin | MIN | 1B | 25 | | 219 | 172 | 181 | 156 | 175 | 245 | 6 | 245 | 156 | | Mitch Garver | MIN | C | 25 | | 313 | 171 | 267 | 176 | 208 | 279 | 6 | 313 | 171 | | LaMonte Wade | MIN | LF | 40 | | NULL | 7 | 50 | NULL | 7 | NULL | 3 | 50 | 7 | | Luis Arraez | MIN | 2B | 40 | | NULL | 7 | 50 | NULL | 7 | NULL | 3 | 50 | 7 | | Lucas Duda | MIN | 1B | 40 | | 316 | NULL | 111 | 0 | NULL | 312 | 4 | 316 | 0 | | Michael Reed | MIN | RF | 40 | | 50 | 90 | 48 | NULL | 60 | 173 | 5 | 173 | 48 | | Willians Astudillo | MIN | C | 40 | | 293 | 125 | 276 | 97 | 186 | 272 | 6 | 293 | 97 | | Ronald Torreyes | MIN | 2B | 40 | | 98 | 138 | 110 | 0 | 125 | 178 | 6 | 178 | 0 | | Jake Cave | MIN | CF | 40 | | 289 | 193 | 179 | 193 | 198 | 272 | 6 | 289 | 179 | | Zack Granite | MIN | CF | 40 | | NULL | 28 | 50 | NULL | 27 | 36 | 4 | 50 | 27 | | Nick Gordon | MIN | SS | 40 | | 22 | 34 | 50 | NULL | 33 | 63 | 5 | 63 | 22 | | Tomas Telis | MIN | C | Minors | | NULL | NULL | NULL | NULL | NULL | 60 | 1 | 60 | 60 | | Adam Rosales | MIN | 2B | Minors | | 20 | NULL | 82 | 0 | NULL | 179 | 4 | 179 | 0 | +--------------------+------+-----+--------+--------+------+------+------+------+-------+-------+-------+------+-----+ 24 rows in set (0.01 sec)
There are some pretty big differences across our six data sets. Roster Resource is not on board with regular at-bats for C.J. Cron at only 313 as of late last week. This is updated to 354, but still a far cry from the 460 and up from other sources.
Nelson Cruz has a 64 PA difference, Jonathan Schoop almost 70 and Marwin Gonzalez over 80. Prospects are even tougher to peg because of the uncertainty when they will get the call up to the Majors.
Here are the Philadelphia Phillies pitchers:
+----------------------+------+-----+--------+---------+----------+------+------+------+------+-------+-------+-------+------+-----+ | Name | Team | Pos | Roster | Role | Rotation | ATC | BAT | Razz | RR | Steam | Zeile | Count | High | Low | +----------------------+------+-----+--------+---------+----------+------+------+------+------+-------+-------+-------+------+-----+ | Aaron Nola | PHI | SP | 25 | SP | 1 | 197 | 194 | 197 | 213 | 194 | 198 | 6 | 213 | 194 | | Jake Arrieta | PHI | SP | 25 | SP | 2 | 173 | 181 | 173 | 172 | 181 | 170 | 6 | 181 | 170 | | Nick Pivetta | PHI | SP | 25 | SP | 3 | 162 | 169 | 149 | 143 | 160 | 155 | 6 | 169 | 143 | | Vince Velasquez | PHI | SP | 25 | SP | 4 | 134 | 150 | 130 | 140 | 137 | 124 | 6 | 150 | 124 | | Zach Eflin | PHI | SP | 25 | SP | 5 | 144 | 145 | 137 | 153 | 145 | 139 | 6 | 153 | 137 | | David Robertson | PHI | RP | 25 | CL | | 68 | 65 | 65 | 67 | 65 | 66 | 6 | 68 | 65 | | Seranthony Dominguez | PHI | RP | 25 | SU | 1 | 67 | 65 | 71 | 70 | 65 | 68 | 6 | 71 | 65 | | Tommy Hunter | PHI | RP | 25 | SU | 2 | 61 | 55 | 55 | 55 | 55 | 61 | 6 | 61 | 55 | | Juan Nicasio | PHI | RP | 25 | Bullpen | | 53 | 40 | 22 | 52 | 25 | 48 | 6 | 53 | 22 | | Adam Morgan | PHI | RP | 25 | Bullpen | | 46 | 45 | 32 | 52 | 45 | 47 | 6 | 52 | 32 | | Jose Alvarez | PHI | RP | 25 | Bullpen | | 49 | 40 | 25 | 45 | 30 | 45 | 6 | 49 | 25 | | Hector Neris | PHI | RP | 25 | Bullpen | | 57 | 40 | 44 | 30 | 40 | 56 | 6 | 57 | 30 | | Pat Neshek | PHI | RP | 25 | Bullpen | | 54 | 55 | 44 | 51 | 55 | 49 | 6 | 55 | 44 | | Adonis Medina | PHI | SP | 40 | SP | | NULL | NULL | 8 | NULL | 9 | NULL | 2 | 9 | 8 | | Jerad Eickhoff | PHI | SP | 40 | SP | | 54 | 9 | 8 | 88 | 9 | 57 | 6 | 88 | 8 | | Ranger Suarez | PHI | SP | 40 | SP | | NULL | NULL | 7 | NULL | 9 | 16 | 3 | 16 | 7 | | Drew Anderson | PHI | SP | 40 | SP | | NULL | 19 | 26 | NULL | 19 | 16 | 4 | 26 | 16 | | Enyel De Los Santos | PHI | SP | 40 | SP | | 51 | 74 | 69 | 60 | 74 | 38 | 6 | 74 | 38 | | Edubray Ramos | PHI | RP | 40 | Bullpen | | 31 | 20 | 19 | NULL | 10 | 34 | 5 | 34 | 10 | | James Pazos | PHI | RP | 40 | Bullpen | | 37 | 40 | 29 | NULL | 35 | 46 | 5 | 46 | 29 | | Victor Arano | PHI | RP | 40 | Bullpen | | 39 | NULL | 24 | NULL | 20 | 39 | 4 | 39 | 20 | | Yacksel Rios | PHI | RP | 40 | Bullpen | | 7 | 20 | 24 | NULL | 15 | 21 | 5 | 24 | 7 | | Austin Davis | PHI | RP | 40 | Bullpen | | 10 | NULL | 27 | NULL | 10 | 34 | 4 | 34 | 10 | | Edgar Garcia | PHI | RP | 40 | Bullpen | | NULL | NULL | 24 | NULL | 10 | NULL | 2 | 24 | 10 | +----------------------+------+-----+--------+---------+----------+------+------+------+------+-------+-------+-------+------+-----+ 24 rows in set (0.02 sec)
Pitcher changes are going to be very injury driven, so I expect to see less variation in projected innings than hitters’ plate appearances. I input the Roster Resource projections manually using a cut off of 75 PA and 30 IP, so there are additional players projected by RR but those numbers are not captured.
Draft Buddy Playing Time
Initial playing time values for the Draft Buddy projections are not a straight average of the available data. To me, that doesn’t make a lot of sense, in part because ATC and Zeile are averages themselves. Why take an average of an average?
Buddy numbers are unlikely to fall outside the range from the six sources. To start, they are a weighted average of some of the sets that I consider the most trusted. Going forward my plan is to manually review and adjust playing time by team. To help with this, individual team reports allow for news, research and comments about changes to playing time.
To produce the projections, playing time is applied to Steamer. The ratios such as batting average, ERA and WHIP should be the same (or close, due to rounding) between Steamer and Draft Buddy.
Jonathan Schoop has 525 PA per Steamer, but Draft Buddy playing time bumped him to 579. The image below from Schoop’s player page indicate AVG, OBP and SLG ratios are all the same.
The Curious Case of Chris Sale
Steamer loves Chris Sale. I mean, loves him. This is a good example of a projection system unable to adequately account for different risk levels between players. Running Steamer through Last Player Picked with 15-team NFBC settings, Sale values at $49.
Maybe the pitchers as a group are ranked too high relative to hitters by LPP, but regardless of that consideration, Sale is way ahead of Max Scherzer, Jacob deGrom and every other hurler. At issue is his projected ERA of 2.76 (2.94 for deGrom) and projected WHIP of 0.97 (1.03 for Scherzer). The differences really increase the value of Sale. Yet, few people would draft Sale ahead of Scherzer because of concerns over Sale’s relative probability of staying healthy.
Even reducing Sale’s projected IP, he still ranks very high thanks to the considerably lower ERA and WHIP projections. Instead of SP1, a more appropriate ranking for Sale is Top 4-5 among starting pitchers.
To accomplish that, I manually (subjectively) adjusted his dollar value and ranking. This is something that will happen with more players going forward, but since he is such a significant player I felt an early adjustment is warranted for him.
Missing In Action Players
One problem using the Steamer projections as a base is, what if they don’t project a certain player who we otherwise have projected playing time? This is not very common but it is worth noting the players this applied to. They are shown in the following two tables.
+-------+-----------------+------+-----+--------+-----------+-------+ | id | Name | Team | Pos | Roster | Positions | _PA | +-------+-----------------+------+-----+--------+-----------+-------+ | 6687 | Matt Joyce | CLE | RF | Minors | OF | 133.0 | | 17584 | Mark Reynolds | COL | 1B | 25 | 1B | 116.0 | | 18038 | Hunter Pence | TEX | CF | Minors | OF | 112.0 | | 18599 | Drew Butera | PHI | C | Minors | C | 68.0 | | 18961 | Adam Rosales | MIN | 2B | Minors | 2B | 51.0 | | 19658 | Lucas Duda | MIN | 1B | 25 | 1B | 221.0 | | 19658 | Lucas Duda | MIN | 1B | 25 | DH | 221.0 | | 20351 | Brandon Guyer | CWS | RF | 25 | OF | 125.0 | | 20573 | Isaac Galloway | MIA | CF | Minors | OF | 142.0 | | 20911 | Ryan Flaherty | CLE | 3B | Minors | 3B | 69.0 | | 21286 | Bryan Holaday | MIA | C | Minors | C | 71.0 | | 22993 | John Andreoli | SF | CF | Minors | OF | 67.0 | | 23731 | Jack Reinheimer | BAL | SS | 40 | SS | 99.0 | | 25060 | Charlie Tilson | CWS | LF | FA | OF | 67.0 | | 26247 | Daz Cameron | DET | CF | Minors | OF | 73.0 | | 26249 | Mike Gerber | SF | CF | Minors | OF | 98.0 | | 26284 | Bo Bichette | TOR | SS | Minors | SS | 101.0 | | 26294 | Brendan Rodgers | COL | SS | Minors | SS | 154.0 | +-------+-----------------+------+-----+--------+-----------+-------+
+-------+----------------+------+-----+--------+-----------+--------+ | id | Name | Team | Pos | Roster | Positions | _IP | +-------+----------------+------+-----+--------+-----------+--------+ | 10758 | Jason Hammel | TEX | RP | Minors | RP | 63.00 | | 10758 | Jason Hammel | TEX | RP | Minors | SP | 63.00 | | 16352 | Homer Bailey | KC | SP | Minors | SP | 20.00 | | 20478 | Tim Collins | MIN | RP | Minors | RP | 15.00 | | 21787 | Oliver Drake | TB | RP | Minors | RP | 27.00 | | 22380 | Danny Barnes | TOR | RP | Minors | RP | 33.00 | | 22565 | Brandon Maurer | PIT | RP | Minors | RP | 23.00 | | 23058 | A.J. Cole | CLE | RP | Minors | RP | 28.00 | | 23176 | Merrill Kelly | ARI | SP | 25 | SP | 138.00 | | 23555 | Derek Law | SF | RP | Minors | RP | 36.00 | | 26019 | Jared Miller | ARI | RP | Minors | RP | 16.00 | +-------+----------------+------+-----+--------+-----------+--------+
The only player I truly feel should be included in our projections but is not at this time is Merrill Kelly. He will either show up in Steamer next time, or I will manually add him to our projections.
Final Thoughts
As a first attempt at creating our own projections, I am pretty happy with the results. Not that we want to ultimately create a rankings list that is the same as Average Draft Position, but there are fewer outliers as a result of odd playing time estimates. We have a good base to build on.
This process increased my already healthy respect for the work prognosticators put in to help us value players for fantasy sports. It is a lot of numbers and a lot of time, and a potentially never-ending saga making adjustments, trying different approaches, fine-tuning things, etc. Props to Ariel Cohen, Derek Carty, Jared Cross (and crew), Rudy Gamble, Jason Martinez and FantasyPros for their contributions to fantasy baseball that help this project.
It also drives home for me how pure number-generated rankings are not perfect (i.e. Chris Sale). The numbers help immensely, and then we also need to look beyond the numbers as a sanity check. A combination of quantitative analysis and qualitative risk-reward decisions is the best way to consider all angles and to ultimately draft the best fantasy baseball team possible.