How Splash Play's DFS Projections and Simulations Work

By Splash Play Staff · A guide to reading the tools

Splash Play's DFS tools work in four layers: inputs, projections, simulations and lineup building. A projection is one forecast of a player's fantasy points. A simulation plays the slate out many times to show a range of outcomes. The lineup tools then pick rosters under the salary cap, roster rules and your own exposure limits. Each layer answers a different question, and a good number in one does not prove the next.

The four layers

1. Inputs. The platform, the slate, the player pool, salaries and the latest player and game information.
2. Projection. A forecast in that platform's scoring system. SPAGS is the projection board.
3. Simulation. Modeled outcomes for players and lineups across many versions of the slate.
4. Lineup building. Candidate lineups or a full portfolio, chosen under eligibility, salary and the limits you set.

Keep them separate when you read a result. A lineup that fits the cap can still rest on a bad forecast, and a large simulation can still assume a field that looks nothing like your contest.

What a projection is

A projection is a single expected score. For NFL, the model starts from each team's expected plays and implied points, splits targets, carries and dropbacks by role, and adjusts for the opponent. Injuries and depth-chart changes run through the same team totals, so when a starter sits, his work shows up in his teammates' projections instead of vanishing.

The useful question about any projection is what would make it wrong. Start with the role: check whether a player's expected workload has changed before you credit or blame the matchup.

What ownership is

Ownership is an estimate of how much of the field will roster a player. It is not the chance he hits his projection, and it is not your own exposure to him. Say a receiver is projected for 14 points at 18% ownership (example numbers, not a current slate). Those are two separate facts: a forecast of his score and a guess about the crowd. A popular player can still be the right play. Ownership tells you how much company you will have if you are right.

What a simulation adds

A simulation replaces one expected score with a range. Floor and ceiling describe that range for a player, and lineup results show modeled cash rate, wins and top-10 finishes. Three lineup views sort the pool by what you want from it:

• Cash frequency emphasizes modeled cash rate.
• Pure upside emphasizes modeled wins and top-10 finishes.
• Balanced combines the two.

Sim% describes simulated lineup usage. It is a model result, not a promise about your final exposure. Your own exposure is simpler: a player in 6 of your 20 lineups is 30% exposure.

When two results differ, name what changed

Most confusion comes from comparing two results that differ in more than one way. Four things can move a number:

• A projection update changes the forecast.
• An ownership update changes the estimate of the field.
• A different payout structure changes what a given finish is worth.
• A new exposure limit changes which lineups are eligible.

If you change projections, run the simulation again so the results include them. Then compare like with like: same platform, same slate, same settings. "The sim got better" tells you nothing until you know which of the four moved.

A worked example

Here is the kind of note worth keeping before lock. The slate and numbers are made up.

Example note: DraftKings main slate, checked 6:10 p.m. One starting running back is questionable. His backup's projection assumes the starter plays. If the starter is ruled out, the backup's workload changes, so re-check his projection and rerun before trusting any lineup that includes either of them.

That note takes a minute to write, and it tells you exactly what to re-check when news breaks instead of rebuilding everything from scratch.

What the numbers cannot tell you

Projections and simulations are model outputs. They describe what the model expects under its assumptions; they do not guarantee a score, a finish, a payout or a profit. A simulation is only as good as its assumptions about the players and the field, and real contests differ from modeled ones. This page explains how to read the tools. It is not an audit of every input or setting behind them.

Where each piece lives

Squirt Squad covers matchup screens and projection matrices. Squirt Squad VIP adds SPAGS projections, exports and simulation results. Apex adds lineup browsing and the Portfolio Builder for NFL and MLB. Available tools vary by sport, slate and contest format.

Open the NFL tool to see it on the current slate, read the field guide for the step-by-step entries and export workflow, or compare memberships.