Last week, I published Lessons from a Mock Draft Decathlon, featuring the results and my observations after organizing 10 mock drafts. Each was a standard ESPN league with 10 teams and 5x5 rotisserie scoring. The original plan was to discover the optimal draft spot to carry out a specific strategy, but like so many scientific endeavors, this one inspired multiple follow-up experiments.
However, planning and executing another series of mocks was impractical -- especially since one of the goals of this project is to develop a process that anyone can do. Certainly, not everyone can fill ample mocks with enough live bodies to render valid results. That's where average draft position (ADP) comes in. Granted, ADP has its flaws. That said, if you can find an ADP that reasonably represents the market for your league, that will suffice for this exercise.
The plan for today is to assemble teams from ADP generated from the Mock Decathlon. This isn't perfect, as the positions of several players are skewed by the frequency with which I drafted them, but it will still provide a suitable foundation.
As a reminder, the mocks from last week all featured the following restrictions:
• If J.T. Realmuto is the highest player on the board, he must be selected.
• One of six ace starting pitchers must anchor the staff.
• One of four elite closers must be drafted.
• The roster should be populated with several multiple-eligibility players.
• Speed specialists will be chosen only when deemed absolutely necessary.
Part of this week's follow-up is to determine if any of these conditions should be part of the initial draft plan for a league of this nature. The best approach is not to be married to any single strategy, though it's always helpful to have an idea of what might make the most efficient roster construction. The experiment will again entail my conducting 10 drafts, each from a different starting position, but with no other roster restrictions. That is, I was free to draft using any strategy I desired, as long as the selected player was still available when I was on the clock. However, I did abide by a few procedural rules, which are recommended if you opt to try this for yourself.
• Draft each team without making any real-time comparisons to the other teams already completed.
• Don't rotate your draft spots sequentially. That is, don't draft from the No. 1 spot in your first draft, in the No. 2 spot in the second draft, etc. Mix up the order of where you start each draft.
• If possible, do the 10 drafts over multiple sittings so you don't fall into the same pattern for each one.
• You're allowed to actually have the ADP list with you so you can use it to judge when players "should be" selected.
• However -- and this is crucial -- draft how you would if you didn't have the ADP list. If a player has a Round 8 ADP and you have him ranked as a Round 4 value, don't wait until the last possible spot to take him according to the defined procedure. Take him where you would in a real draft. No one is grading you. The idea is to mimic a real draft. You learn less by delaying the pick just because you know you are 100% guaranteed to get the player if you wait. In a real draft, you aren't sure when the player will be picked. After all, the A in ADP stands for average. Sometimes the player is chosen earlier, sometimes later.
Here are the 10 teams I drafted.
Last week, in lieu of judging my teams against the field in each individual mock, I pitted them against each other and scored them rotisserie-style. Here's this week's updated standings, with the 10 ADP-generated squads added to the previous mix.
Mock Teams versus ADP Teams
Despite the need to approach the analysis with no preconceived notions, I fully expected the ADP teams to dominate since I was unhindered and trusted my ability to build better teams devoid of preset guidelines. However, there are several reasons that some of the decathlon teams nevertheless continued to occupy the upper portion of the standings.
• Perhaps the mock roster rules are indeed the best way to construct a roster -- or at least a huge part of it. I purposely mixed some pet strategies with those I usually avoid. For example, I rarely draft a stud starter and an elite reliever. Normally, it's one or the other, and sometimes neither. That said, I'll admit 10-team leagues aren't my wheelhouse, so I don't have the experience to know if what works in 12- and 15-team formats transcends to the lusher 10-team player pool.
• While I was free to embark on any path in the ADP drafts, each mock had players of interest who lasted longer than their ADP. As such, I was able to draft some desired options in the mocks, but not the ADP drafts. The best example is Mookie Betts. His ADP was six, but I was able to take him at No. 8 overall in the decathlon, followed by Max Scherzer. In the corresponding ADP draft, Betts was already off the board for Team 8, so I opted for Juan Soto instead, then also took Scherzer.
• As has been demonstrated by my colleague Pierre Becquey's Project GOAT experiment, teams punting a category are favored when drafting against known stats, as categories become more tightly bunched in a standard rotisserie league. The top five teams (three ADP and two mock) all finished with four or fewer points in one of the hitting categories. This was unintentional -- and not necessarily a good thing -- leading to the next thought exercise.
Winning the league versus winning the draft
Ultimately, the goal is to build a foundation to win the league, not the draft. Is dedicating a lower percentage of assets to a chosen category optimal? We talked about this a bit last week, suggesting that sitting low in batting average and/or steals isn't a dealbreaker and, in fact, is a viable approach.
Team ADP 4 finished fourth, in large part by dumping runs while accruing 39 of the possible 40 points in both homers and RBIs. However, this is a terrible foundation. It's difficult to envision any kind of in-season management where runs are improved with a net gain. Having such a disparity may be an argument for tracking stats during a draft, as a balanced team usually garners a few more runs than RBIs. That said, I'm going to put this on me. I should have been more aware of the "team build," and I'll never argue with someone who is more comfortable with tracking.
Something else to keep in mind is that the categories are close by nature of the process. This is why punting something is effective in a draft against known stats. If you look at your teams in this manner, don't focus on absolute finishes. Use the standings to identify teams that are worthy of further study. Look for patterns among the top teams (perhaps a general strategy to employ) or trends with the lesser squads (approaches to avoid).
Along those lines, the top six clubs all scored higher in pitching. Is this the best way to leave the draft? To be honest, there isn't a universal answer. The ability to perform in-season moves is a huge factor. How often are pickups? How many reserve slots are allowed? How effective are you at streaming pitching?
To this last point, if you feel your streaming skills are deficient, team construction akin to that of the top teams in this experiment is more beneficial. That is, pay for solid ratios so you aren't as reliant on identifying strong spot starters and, when you do go that route, there's a natural buffer already on your roster. Personally, I count managing pitching as a strength. Therefore, even though the pitching-heavy rosters "won," they don't provide me with the most beneficial infrastructure on which to build. Ideally, I should bully hitting more in the draft and rely on my in-season managing skills to bolster pitching.
There's some game theory supporting this approach, though to be fair it hasn't been vetted in 10-team leagues. In a study conducted on 15-team leagues, the top couple of teams almost always improve their pitching ratios in-season. Meanwhile, the other clubs hurt their ERA and WHIP, presenting a Catch-22. In order to win, one's ratios need to be improved, but it's hard to do so.
Please realize this isn't drafting a staff that outperforms expectation. The research involves determining the season-ending ERA and WHIP based on the opening day staff's actual numbers and comparing that to the team's final stats. In-season moves obviously affect the ratios, with the best teams doing so to the greatest extent. Ergo, my mindset is that I feel I'm good at managing pitching -- and studies show it's needed to win, so my optimal draft footing should favor that approach.
Fear of missing out
I think the kids call it FOMO. I come into my drafts with a list of arms I know that I favor more than the market. My biggest drafting flaw is feeling like I have to put as many of these hurlers on my team as possible, hence FOMO. It pains me for a competitor to enjoy the breakout of one of my guys, and at a bargain price, no less. In leagues with frequent trading, this is less of an issue. However, most of my leagues are no-trade formats.
Examples of midround pitchers showing up on a multitude of these experimental rosters are Chris Paddack, Mike Soroka, Brandon Woodruff and Mike Minor. Some may contain three or even all four. Candidly, this is a mistake I need to rectify. Some of these picks should be dedicated to bats, helping me to boost counting stats. I can replace these early picks in my rotation with guys I like late such as Joe Musgrove, Ryan Yarbrough, Adrian Houser, Yonny Chirinos, Caleb Smith and Anthony DeSclafani. Sure, the ratios would suffer, but that's fine.
Part of the rationale for this approach is a flaw in rankings. Conventional valuation assumes a player will be active all season. For a pitcher, this encompasses favorable and risky starts. I won't have guys like Musgrove or Yarbrough active each week. Yet their raw rankings bakes in the outings to be avoided. Yes, I could figure out a means to quantify it, but I should l also be adept enough at adjusting intuitively.
Other flaws in valuation
When outlining this experiment, one of the things I wanted to do was look at the correlation of standings points and projected earnings. Even though these were drafts, valuation principles can be borrowed from auction formats and expected earnings can be determined for each player based on the format specifications. After doing all the math, earnings and standings points yielded a correlation constant of just 0.49. This suggests some level of a direct relationship between earnings and standings, but I thought it would be stronger.
Two explanations immediately came to mind. Since all that's needed to finish above the adjacent team is one homer, one RBI, one save, some of the discrepancy is distribution-based. The "value" attributed to everything more than just the stats needed to be one ahead of the next team is wasted. Alternatively, my valuation process could be flawed. To be frank, all valuation processes are fraught with error, but maybe my algorithms don't properly measure the impact of ratios, or whatever.
Then I had an epiphany. Projected earnings are based on assigning an empirical budget split to hitting and pitching, usually dedicating 65-70% to bats. My standard is 69%. While that's most favorable for adjusting to the market in drafts or auctions, in studies of this nature, earnings should be computed on a 50/50 basis since there's an equal number of hitting and pitching points. This may be difficult to grasp since the standard roster has more hitting roster spots, but that's moot. When looking backward, the number of points directs the split.
To put this in perspective, after switching to a 50/50 split, Ronald Acuna Jr. went from $46 to $34 while Jacob deGrom climbed from $38 to $62. After running the expected earnings with the appropriate split, the correlation coefficient ballooned to 0.89, displaying a strong relationship between team earnings and standings finish. So even though my preference is to bully hitting and manage pitching, if properly ranked, the top pitchers blow away the top hitters. What's a drafter to do? How about taking a pitcher early, then waiting to add more arms? I know, this is far from revolutionary advice, but it's supported by game theory and valuation.
Dude, what's the deal with Realmuto?
Part of the initial outline was forcing some non-Realmuto teams to test the efficacy of drafting him or fading for a later catcher, a common ploy in leagues of this format. That said, I wanted the first iteration to be freeform, with no defined strategy. There is an inherent issue with the particular ADP I'm using, since Realmuto made it onto eight of my mock teams: seven in the fourth round and one with the first pick of the fifth. His ADP was 38. After polling the participants, Realmuto's "true ADP" would have been around 75. Should I adjust this accordingly or keep it at 38?
I decided that if I knew his new ADP, I would be too tempted to wait until the sixth or seventh round in this exercise, so I left it as is, forcing the decision by the end of the fourth round, which is in sync with a real draft since I received a sizable number of "I would have taken Realmuto if you didn't" responses. As it turned out, I still opted for the Phillies backstop eight more times, only missing out when using the top two picks, as that would have required taking him in the early third, voiding most of the advantage enjoyed by ranking him properly.
To reiterate, this isn't always the case. It just happens that, by my numbers, Realmuto is so far ahead of the catching pool he warrants special treatment.
Closing time
So far, two of the strategy points in question have been addressed. I desire an ace early, then wait on more arms and I favor Realmuto. Please realize this isn't a global approach. Your mileage may vary, based on your own managerial strengths and weaknesses. In fact, this may not be my preferred tactic in different league formats. It's my design in 10-team, mixed rotisserie leagues.
All that's left to discuss is closers. Was forcing an elite stopper the right call? I'll be honest, I still don't know. In deeper leagues, I'll rarely pair an elite starter and an elite closer, fearful of missing out on offense. However, my gut is leaning to grabbing Josh Hader or Kirby Yates, if possible. There are game theory reasons for doing so, especially with the likelihood of a shortened season. Maybe it's best to lock down saves, unsure of how closers will be used in an abbreviated and perhaps condensed campaign. However, I drafted with a full season in mind.
What's next?
One of the side benefits of undergoing a study of this nature is unearthing a plethora of follow-up studies. For example, in order to gauge the efficacy of drafting closers early, the benefit beyond saves should be quantified. Most ignore the influence closers have on ratios, and to a lesser extent strikeouts. That's an undertaking for another day.
Here are a few more topics to be explored. Obviously, redoing the ADP drafts with the mindset of backing off the middle pitching tier and hitting the late tiers harder makes sense. The end products may not finish atop the mythical standings, but they may well provide a more solid base to finesse in season.
Something I've already done on a couple of teams (and plan to do on all of them) is finding mistakes. Even though I'm aware picks aren't always directed by the raw rankings, they're a strong influence. Overall roster construction trumps "value." This is highly recommended after doing your own ADP mocks.
For example, Decathlon 4 is the club with 39 points in homers and RBIs, but just one in runs. Obviously, the team is overloaded with meat-of-the-order batters while lacking leadoff types. Even though I have Marcell Ozuna ranked ahead of Kris Bryant, Chicago's combination 3B/OF is actually a better pick for this squad. After going through all your rosters and identifying similar scenarios, you'll be better accustomed to focusing on roster construction and not simply trying to beat the market. Remember, you want to win the league, not the draft.
The experience of the Mock Decathlon and ADP drafting was quite satisfying, not to mention eye-opening. As such, repeating the mock process with points-league scoring is warranted. If you're interested in taking part in this next phase, please look for details on my Twitter account @ToddZola.