With Mass Multi-Entry (MME) now the hottest thing since sliced bread, everybody is looking for ways to fine tune optimizer processes to bink the Dime Time, Quarter Arcade or most recent flavor of the Millionaire Maker across contests operators. Building lineups for MME can be very profitable or an unmitigated disaster depending on your overall approach, variance, and the settings that you use. Building lineups in MLB is definitely an art, but there are several settings you can take advantage of to help improve your optimizer process and avoid entering lineups that are drawing dead before you start.
- Stack Settings
- Ceiling Projections
- Volatility and Shuffle
- Hitter vs Pitcher Constraints
- Max Public %
It is commonly known that stacking is critical for competing in MLB DFS tournaments. Therefore, your stack settings are a critical component to your MME process in MLB DFS. Our stack settings provide flexibility in a few different ways.
The first is the ability to select “Primary” or “Secondary” stacks. The primary stack is the biggest stack, and should often be a core of your favorite teams on that given night. There are a few strategies MLB DFS pros may use — some players run a very tight pitching core of 4-5 pitchers, and then a broad distribution of stacks — while others may choose to go more diverse at pitcher and narrow down their primary stacks. Your approach should vary based on slate size, but your “primary stacks” should be your favorite GPP teams of that slate, either based on team totals, value or the leverage they offer at their ownership.
If you are going with a tighter core of primary stacks (maybe 4-5 teams at 15-20 percent owned) than it would make more sense to go diverse on your secondary stacks. Your secondary stacks will be smaller, and should almost always include your favorite primary stacks but can also be expanded to teams that are less likely to go off for 10+ runs but still could have mini-stacks perform.
The second feature is the ability to choose your stack size for both primary and secondary options. One flaw I have seen in some other software is that it is rigid and requires you to select your stack size as exactly 4 or exactly 5. Our software lets you dictate the stack size either with an exact number or a “greater than” functionality. For example, one user may decide they want their primary stacks to be exactly four players. This means the optimizer would choose four, and only four, players from a given team. Another user may decide they want their primary stack to be set to 4+ which would allow the optimizer to also mix in a 5th player if it was optimal based on the projection and settings.
The 5-2+ stack is my go-to because it provides maximum correlation on your primary stack, while still offering correlation on your secondary stack. The reason I prefer 5-2+ to 5-3 is that often positional scarcity might dictate what gets chosen as the “3” on that stack format. For example, if the Yankees are playing on a short slate and Gary Sanchez out projects all other catchers by two points, forcing a 5-3 stack may give you a lot of Yankees exposure simply because Sanchez dominates that position. Running 5-2+ would still sometimes capture 3 Yankees stacks but would also mix in Sanchez as a one-off player.
Other stacks to consider: 5/3, 4/4, 4+/3+
One of our features last year included the integration of floor and ceiling projections (range of outcomes) directly into the optimizer. Our models use coefficients for each player to project their requisite floor for cash games or upside for tournaments. Underneath the “Run Settings” you can toggle whether to use Standard, Median (50th), or a range of Floor (10th/25th) and Ceiling (75th/90th) projections.
Theoretically you are seeking a high upside lineup for tournaments, however, the majority of our users simply optimize straight off of our standard projections.
This is part habit (its the default projection) and part lack of trust on behalf of users (I know it says ceiling but does it really produce better lineups)? I backtested this feature using our optimizer for previous slates last MLB season and found two important takeaways:
1. The average lineup generated in a 150 lineup MME run with a 90th percentile projection scored the same amount of points as the average lineup generated with a standard projection.
2. The highest-scoring lineup generated in a 150 lineup MME run with a 90th percentile projection scored 7 more points on average than the highest-scoring lineup with a standard projection.
Don’t be scared of running MME sets based on ceiling projection. Your average lineup is comparable and your best lineup is materially better.
Volatility and Shuffle
These are your go-to diversity tools. If projections were perfect, we wouldn’t need volatility at all, but in MLB projections are far from perfect.
Statistically, projections in sports like the NBA are more than twice as accurate on a daily basis as a sport like MLB. This is driven by the team nature of the game, event-based scoring, and an overall limited number of plate appearances. In MLB, we need to use more volatility and shuffle to account for the randomness in the daily scoring. Volatility runs the optimizer with a random projection +/- that player’s baseline (VOL=10 runs it randomly + or – 10%). Shuffle decreases a player’s projection as soon as they are used in a single lineup. This accounts for real utility – as a player is used each progressive time their value in DFS contests diminishes.
Rather than come up with thresholds for each setting, I prefer to consider an overall threshold for volatility and shuffle and think you can go as high as 50 for MLB batters. This could be done with 45 VOL + 5 Shuffle, 40+10, 35+15 etc. Increasing volatility and shuffle will introduce more diversity to your lineups. There isn’t a magic bullet — lineup diversity reduces variance and makes sense especially for players still trying to figure out their approach to MME.
Hitter vs Pitcher Constraints
One of the final optimizer features that can help you improve your process is the Hitter vs Pitcher constraint. This allows you to tell the optimizer how many batters it can include in a lineup versus the pitcher in that lineup.
The standard industry talking point is that this should be set to zero, as it is unlikely that if your pitcher reaches their ceiling that an opponent will have a good enough game to be on winning GPP lineups. I believe this is mostly true for large field / main slate contests, where a pitcher usually needs a truly ELITE outing to be good enough to ship all the money.
However, I have observed a majority of GPP players apply these same optimizer parameters to smaller slates (Early Only, Late, etc) and I believe that is a leak. As slate size shrinks, the positional value becomes even more scarce, opening up the possibility that an average performance still is optimal for GPPs.
For short slate contests last year, 20% of them were one by players allowing hitters versus their pitcher and 60% of “optimal lineups” included at least one hitter versus your pitcher.
My opinion is GPP players are systemically grouping out the stone nuts on short slates and we can use this to our advantage by allowing 1 or 2 hitters versus our pitcher.
The “right” approach is probably a bit more nuanced than this, as somebody like Gary Sanchez is more viable as a one-off than Brett Gardner hitting 8th at a loaded OF position with plenty of HR power. Dedicated MME players will spend time making groups for players that they only want as part of stacks, creating a rule for specific batters to ensure they are only played as part of stacks.
Once you have set your basic optimizer settings, run your lineups and look at them objectively. If you see illogical roster construction, consider how you can create rules to avoid that.
Max Public %
The last constraint to understand is the Max Public % feature which allows you to control the total maximum threshold for your lineup as it relates to what the field is doing.
Utilizing this field is the simplest way to create contrarian lineups. There is not a cheat code feature I can share and ultimately how you utilize this will need to change based on slate size.
On short slates, the average optimal roster has roughly 20% public % per player, or 180 for the full lineup. On a full slate, a somewhat contrarian player may have lineups with average ownership of 10 percent, or total ownership of 90 percent in large field MME, while on a short slate those percentages could double.
If left uncapped, lineups will be built delivering the highest projected plays, many of which will likely be popular with the field. This type of setting might work for days when our projections are naturally contrarian or for a single entry or 3-max build but it probably isn’t optimal to be entering 150 consensus lineups (other subscribers may also be firing) into a tournament.
Consider how aggressively contrarian you want to be when creating MLB GPP lineups and set a Public % constraint accordingly.
In summary, Optimizers are a critical part of any sound MLB GPP process. If you can master these 5 Key Optimizer Settings and learn to make the optimizer dance you are on the right path towards improve your GPP process.
For more information on utilizing our tools refresh on our Optimizer Tutorials.
MLB Optimizer Tutorials