Fantasy Baseball: Draft Strategy, Part 2 – Turning Projections Into Values
In the last installment of my Fantasy Baseball Draft Strategy series, I mentioned some standard draft strategy principles and concepts. I brought up some of the flaws with people relying too heavily on ADP and the effect that various scoring systems and roster configurations can have when evaluating players. This section is more of the meat and potatoes section, where I discuss assigning each player a proper value based on projections. This is where the informed fantasy player who is willing to put in a little elbow grease will really begin to separate himself or herself from the pack.
What?! Your league is not planning on using Fantrax? Inconceivable! Check out everything Fantrax has to offer and I’m sure you’ll come around to our way of thinking.
Fantasy Baseball Draft Strategy for 2020
Before I get further into this I do want to clarify a couple of things. First, I do not have an advanced degree in sabermetrics or analytics. When I started playing fantasy baseball, I played by mail and looked at player stats in Baseball Weekly. For those who are not familiar with Baseball Weekly, it is essentially a newspaper all about baseball. For those who are not familiar with newspapers, well… my point is that I am not going to get too deep into advanced metrics, primarily because these statistics were not a part of my introduction to either real baseball or fantasy baseball. When I started playing, I remember thinking things like, “Jim Thome batted .311 with 38 home runs last year but had 141 strikeouts. I wonder what his numbers would look like if he only struck out 100 times.”
That is not to say that advanced metrics are not useful. They most certainly are when taken in the correct context and utilized properly. But xwOBA is not a category in any league I play in, so I will primarily attempt to stick to the basics. My aim is to make this both an intro for newer players and a refresher with a little bit of game theory for the more seasoned players. If you do want an analytics crash course, our team has a ton of data for you to absorb. They do outstanding work. And I’m not just saying that so they will invite me to the company cookout. People outside of our group like Alex Fast and Alex Chamberlain are also putting out tons of great content, as are countless other men and women throughout the industry who may or may not be named Alex.
Regardless of whether you bring your own rankings and projections into the draft room (you definitely should) or merely rely on the default settings (did you really not read Part One?), you need to determine what those rankings and projections really mean. Corey Seager is ranked 95th among hitters solely based on our ADP. Franmil Reyes is 98th and Mallex Smith is 104th. To simply state that Seager > Reyes > Smith is a trap that too many fantasy players fall into. The three players I mentioned are vastly different and bring their own set of skills to the table. How do we quantify what each of them brings to the table when deciding who is the best to take in a draft? To try to figure this out, you should attempt to convert raw projections into a more usable tool. Here are a couple of methods that will hopefully illustrate how to do so.
The regular season is less than two weeks away! Don’t worry, there’s still time to add that “one more league!” Leagues are still forming at Fantrax.com, so head on over and jump in one today.
Turning Projections Into Values
Hitter Values in Points Formats
First, I will start with a run of the mill Points league. Points leagues are deemed as “simpler” in many circles because they do not incorporate ratios. While that seems logical enough, I do not think that makes Points leagues easy. Each league and format comes with its distinct challenges. Indulge me for a moment while I go down the rabbit hole and see why there is very little black and white when evaluating players, even in Points leagues. Here I will begin have a nice, simple, apples-to-apples comparison of three players who play the same position and whose ADP is relatively close on Fantrax. Projections are courtesy of Steamer.
I mentioned in Part One of this series that our ADP does not differentiate between the smorgasbord of customizable league settings we offer. It appears that those drafting Joey Gallo thus far in Points or Best Ball leagues are getting quite the bargain. His expected production is much higher than his counterparts despite a depressed ADP. Owners in Roto leagues are likely worried about a poor batting average, but Gallo is still an obvious value in Points and Best Ball leagues. You can see how a speedster like Robles benefits from the scoring format in Best Ball leagues. Instead of a hitter netting four points for a home run and two for a steal, big flies and thefts are weighted equally. I would still rate him third among this specific set of players, but his (and virtually everyone else’s) value is affected by the scoring settings of your specific league.
It is important to not only consider total points, but points per plate appearance when evaluating players in both Points and Best Ball formats. If you are comparing two players, one of whom has 510 projected points in 550 plate appearances, and another who projects to have 485 points in 510 plate appearances, you may be better served to take the latter because Player B averages more points per plate appearance. If you feel that both players will end up with the same number of plate appearances, the second player makes for the better selection. I know that seems basic, but people far too frequently just look at totals rather than numbers on a per-game or per-plate appearance basis.
Steamer also puts out a separate set of projections known as the Steamer600, which provides projected hitting statistics based on 600 plate appearances. This offers an interesting look into a player’s raw skill set. However, you should realize when to take these numbers at face value and when to take them with a grain of salt. For example, Steamer600 projects Bobby Bradley to hit 33 home runs if given 600 plate appearances. That’s all well and good, but the chances of Bobby Bradley getting 600 plate appearances this season are about the same as the chances of Archie Bradley getting 600 plate appearances. Bobby Bradley is not expected to play much in 2020, which is why he is going well outside the top 600 players drafted. You should only draft him in extremely deep leagues, and even then it is a stretch to expect much from the prospect in 2020.
There are additional factors to consider before blindly drafting the hitter with the most projected points, or even the most projected points per game. If there is a hitter with a distinct home field advantage who is also available at that point in the draft, you may choose to go in that direction, particularly in Best Ball formats. This is a relatively simple and affordable way to take advantage of certain splits. Roster structure from a positional standpoint is also important. If Soler, Robles, and Gallo are there for the picking but you have already selected three outfielders, you may choose to invest in a player at another position instead. There are a bunch of different directions you can go in with each draft pick. The permutations are seemingly endless, but it is easier if you make sound decisions at each step in the process.
Despite plenty of promises from pundits posting clickbait-y articles that definitely aren’t this one, there is no one way to crush your fantasy baseball draft. I said in Part One that drafting a fantasy baseball roster is like putting a puzzle together. But I neglected to mention that the puzzle and the pieces themselves are constantly changing. Drafting is not like one of those puzzles your grandma used to give you when you were a kid. It is much more like a game of Tetris. You try your best to clear out room for that perfect O tetromino (the square piece), but if a bunch of Z tetrominoes start hurling themselves towards you, you have to do your best to turn those bad boys into a makeshift square. Ultimately, that is what drafting a well-balanced roster is all about.
Baselines for Hitters in Roto Leagues
When coming up with ways to project not only a baseball player’s stats, but how those stats equate to fantasy baseball player values, there is more than one way to skin a cat. Some suggest using your league’s historical data to try to create baselines for what the average player would contribute to in each category. I feel that this practice has some holes. This method fails to factor in year-to-year statistical variance, such as the doctored baseball, the launch angle revolution, or which teams are most effective at stealing an opponent’s signs. It also assumes that you are playing in the same league year after year. Certainly, there are those who do, but if someone only played in a 10-team league last season and now wants to test the waters in a 12 or 15-team league, last year’s 10-team league standings won’t amount to much.
Having said that, I do agree with the premise of baselines and find the right ones quite useful. Much has been written about how to go about creating these baselines. I will not pretend to reinvent the wheel, so I would suggest checking out The Process by Jeff Zimmerman and Tanner Bell and Winning Fantasy Baseball by Larry Schechter. These books, along with several others, detail principles such as Percentage Value Method [PVM] formulas and Standings Gain Points [SGP] formulas. For a brief introduction, suppose you are in a 12-team league that starts 14 hitters. You should try and come up with a baseline for the top 168 (12 x 14) hitters to see how they fared. This will give a basic idea of what to look for from the average hitter. These are the numbers that the top 168 hitters contributed last year on average:
For the record, I do not have my own projection system, but there are plenty of excellent systems out there. Guys like Rudy Gamble and Ariel Cohen are among the many industry names who are taking some of the guesswork out of the equation for the rest of us. I highly recommend checking out some of their work. Once you have your preferred set of projections for each player, you can turn these thresholds into values for each player’s contributions in each category. I like to assign a baseline of 100 for each category’s “average” and figure out a player’s value relative to that number. For the purposes of this exercise, I will use Cody Bellinger’s Steamer projections to try to quantify his value.
Bellinger is projected to score 99 runs in 2020. If I compare that to the baseline of 80.3, Bellinger will get credit for 123 projected run points ((99/80.3)*100). If you prefer to skip multiplying the quotient by 100 and the number ends up being 1.23, that is fine. Whatever floats your boat. I can conduct the same practice to determine Bellinger’s projected totals in home runs (167), RBI (149), and stolen bases (146). This would give Bellinger a projected weighted total of 585 points. This does not include batting average, which I will tackle in the next section.
I like this methodology in theory, but this system has flaws which will necessitate tweaking some of the formulas. First and foremost, the value assigned to stolen bases is completely out of whack. Bellinger only accounts for 12 steals, yet his projected weighted total is 146. That seems rather harmless on the surface. But Steamer projects Mallex Smith to steal 39 bases this year, which would give him a projected weighted total of 476 points in stolen bases alone. Under no circumstances should Mallex Smith ever come close to Cody Bellinger in any projection system, much less surpass him in overall value. You would have to adjust the value given to stolen bases in order to compensate for this glitch. Otherwise, every player who steals a fair number of bases is going to be massively overvalued.
This method also does not account for some of the factors that I touched on in the first paragraph of this section. For example, home runs were hit at a record breaking pace in 2019. Will that trend continue in 2020? We cannot say for sure. If home runs drop this season, the value of each home run will increase. Whereas 25.1 home runs is considered the baseline for 2019 (and thus given 100 “home run points”), that same home run total would have been worth 121 points using the 2018 baseline of 20.8 home runs.
Because of this, it is hard to use past statistics to definitively assign future value. Still, having an idea of what stats to target is a helpful endeavor. Instead of averaging out the statistics of all “starter-worthy” players, I can narrow the field by weeding out some of the bottom feeders among this large set of players. After all, the goal is not to be above average – it’s to win. To do this, I can limit the pool to the top 112 hitters as opposed to the top 168. If I od this, the baseline targets expectedly increase quite a bit:
By reducing the control group from 168 to 112 players, the projected thresholds increase by about 10 percent. This raises the bar for what the average hitter should look like. Perhaps a more thorough and accurate approach would be to take the numbers of the top 168 players (or 112, or whichever number) in each category and create baselines that way. So instead of ranking the 168 hitters overall and creating baselines off their averages, I would rank the top 168 hitters in runs, then rank the top 168 in home runs, etc. These smaller subsets would help zero in on projected production in each category. Interestingly enough, when I tested this theory using 2019 baselines, the results came very close to the initial baselines:
The top 168 run scorers averaged 80.8 runs.
The top 168 home run hitters averaged 26.3 home runs.
The top 168 run producers averaged 78.7 runs batted in.
The top 168 base stealers averaged 9.9 stolen bases.
Before I pat myself on the back, there is another caveat that must be considered – positional scarcity. Of the top 168 hitters outlined in my original baseline exercise, only nine of them were catchers. Ideally that number would be 12 since the pool of hitters is, in theory, the hitters that are considered starter-worthy in a 12-team league. By contrast, there are 25 first basemen and 23 third baseman. Even when accounting for CI and UT spots, there is an overabundance of quality hitters at these positions. A more thorough experiment would be to create baselines within each position group. This would be the most efficient way to give each player, position group, and statistical category their proper value. Remember when I said this would be like going down a rabbit hole? I wasn’t lying. And I haven’t even mentioned pitchers!
As I mentioned earlier, there have been some great books written on different valuation methods and draft and game theory. I highly recommend checking them out.
The Flaw of Averages
Another factor to consider in Roto leagues is that they incorporate averages much more than other fantasy sports. In fact, .300 of 5×5 categories (see what I did there?) in standard Roto leagues are ratio-based. However, you cannot just use the baseline value method discussed in the last section and give each player a “score” based on his contributions in batting average. Simply dividing the team batting average plus or minus a player by the baseline average does not value these batting contributions properly. You would get a number too close to 100 in nearly every instance. Even DJ LeMahieu and his .327 batting average from 2019 would be worth barely more than 100 points. This obviously fails to quantify the edge that his ability to hit for a high average truly provides. It is also important to remember that not all averages are created equally.
For this trick, I will use the Steamer projections Daniel Murphy and Jean Segura. Both hitters are projected to bat .288. Your initial inclination may be to consider their impact on batting average equal. That inclination would be incorrect. Murphy is projected to have 501 at-bats, while Segura is slated for 618. Those extra 117 at-bats have value. To prove this, imagine that either Murphy or Segura fills by 14th active hitter slot. The other 13 hitters in my lineup have the baselines that are based on the top 168 overall hitters from a year ago. My total team would have a total of 1798 hits in 6,565 at-bats, good for a .2739 average. If I add Murphy’s projected totals, my team batting average would rise to .2748. However, if I add Segura’s projected totals instead, my team average would be .2751.
It may seem like a negligible difference, but it illustrates that average is affected not just by the average itself, but by the number of at-bats a player has. And, yes, this goes both ways. In the previous example, the team batting average rose more when adding the player with more at-bats. The same would hold true for players whose expected batting average is below the threshold. Adding Mike Moustakas’ .256 average over 563 at-bats would drop the team batting average to .2724, while Joc Pederson and his .256 average in just 433 at-bats would also lower the team batting average, but to a lesser degree to .2728. My favorite method for quantifying batting average is one that was created by Professor Tom Zownir and highlighted in Gene McCaffrey’s Wise Guy Baseball series. Zownir suggests the following formula:
1 + 3(AB/505)(BA/273.5 – 1)
This method is in line with the baseline methodology above where we are assigning 100 points to the “average” contribution to a given category. In this formula for batting average, the “AB” would be the projected number of at-bats for the player and the “BA” would be that player’s projected batting average. The 2019 baselines of 505 at-bats and a .2735 batting average are included as well and can obviously be tweaked to fit whatever baselines that you deem sufficient. When using this formula, Segura would earn 119 projected batting points and Murphy 116. Pederson would have 83 projected batting average points as compared to Moustakas and his 79 projected batting average points. This is a much more logical way to view a player’s contributions in batting average, and one that is in line with the other “front four” categories.
The baseline methods I have described for Roto leagues are not perfect, but they do provide a quantifiable value for every player. I have found that using these baselines makes drafting decisions much easier. Of course, you can adjust the baselines as needed for your own purposes. Creating values for each player based on their projected statistical output in each category will do wonders for you during the draft. Instead of quickly trying to decide whether to take Kyle Schwarber or Kyle Tucker in the 12th round without any real knowledge of their projected impact on the rest of your offense, you can calmly glance at your sheet of values and see something like this:
Again, I will repeat – this method wildly skews stolen base production, so keep that in mind. Selecting Tucker because he has a higher total is overly simplistic. However, if at that point in the draft you are flush with power but looking for a nice power/speed combo and think Tucker can exceed his current playing time and productivity projections, then drafting Tucker might turn out to be the prudent decision. Either way, you will be able to make a more informed choice when you are able to see exactly what you are getting with either player when compared with the rest of your roster.
For all the rankings, strategy, and analysis you could ever want, check out the 2020 FantraxHQ Fantasy Baseball Draft Kit. We’ll be adding more content from now right up until Opening Day!
Fantrax was one of the fastest-growing fantasy sites of 2019 and we’re not letting our foot off the pedal now! With multi-team trades, designated commissioner/league managers, and drag/drop easy click methods, Fantrax is sure to excite the serious fantasy sports fan – sign up now for a free year at Fantrax.com.