NBA DFS Tools: Lineup Optimizer, Daily Picks and Injury Data

Daily fantasy contests use a roster assembled under the contest’s salary and eligibility rules. A useful tool can check those constraints and compare supplied point estimates, but it cannot remove uncertainty from player performance or guarantee a payout. This site does not run contests or accept funds.

The optimizer accepts a local salary file and runs an integer solver in a browser worker. You can enter projections, lock or exclude players and limit overlap across up to twenty lineups. The solver reports infeasible settings instead of filling missing slots with invalid players.

The role-change watchlist is separate. It requires a verified availability change and a supported lineup or minutes comparison. No current picks are invented from an old injury sample.

Begin with a specific contest file rather than a generic list of NBA players. Salaries and eligible positions belong to that slate. A player’s source roster position may be broader than the operator’s eligibility, and the same athlete can have a different salary in another contest. Those differences matter before any projection is considered.

Next, review the objective you are supplying. A historical average describes past results under a scoring system; a projection estimates a future outcome. The tool does not silently convert the former into the latter. If a file has no projection column, enter your own estimates or leave the player excluded until the input is ready.

Then check participation information against the game date. This build has no live injury enrichment, so a CSV status is only as current as the file or edit that supplied it. An Out label is enforced by the solver, but a missing label is not an all-clear. Rechecking the source is a separate task from optimizing the roster.

Finally, inspect the constraints and output together. The cap, slot eligibility, team limits, locked players and allowed overlap define which rosters are feasible. A higher objective total can reflect a changed assumption rather than a better forecast. Use the result as a transparent calculation, not as a substitute for understanding the contest or as a promise of profit.

This build does not attach a live injury or minutes feed to the uploaded file. It uses the status and projection fields supplied by the user or the file. Out players are excluded when that status is present. Historical average points are not silently treated as current projections.

A future feed integration would need to match player identifiers and slate dates before it can add a status. Matching only by a familiar name is not enough to establish that two records describe the same contest player.

A calculation reflects the input at the time it was run. Recheck a contest’s slate, salaries, eligibility, locks and current availability before using any output. The site does not submit entries or perform automatic late swaps.

A high projected total describes the objective being optimized; it is not a confidence score. Small input changes can produce different rosters without proving that either roster is a better real-world choice.

21+ only. Availability depends on your state and the operator’s terms. No guaranteed results. Gambling problem? Call 1-800-MY-RESET (1-800-697-3738).

This site does not publish a fixed list of eligible states. Operators may use different rules, contest formats and requirements, so the tool’s editable presets are not a substitute for the actual contest terms.

A constrained solver can tell you that one roster has the largest sum of supplied estimates among the feasible assignments it considered. That mathematical statement is narrower than saying the roster will score the most in a real game. The estimates, game context and contest rules can all be incomplete or wrong.

Diversifying several generated rosters changes overlap; it does not establish a particular level of financial safety. The tool’s overlap setting is a player-count constraint, not a model of correlations or a guarantee about a set of contest outcomes. Keeping those meanings separate helps prevent a technical control from sounding like a promise it cannot support.

A team roster can contain players who are not in the salary pool for a particular contest. The salary file is the tool’s eligibility starting point; it is not rebuilt from the team directory. A player absent from the contest file should not be added merely because that player appears on a team page.

The same separation applies to dates. A future NBA matchup, a historical report and a current contest can all mention the same team while describing different events. Combining those records requires a verified match, not just a shared abbreviation. This build does not perform that enrichment, so its limitations remain visible beside the upload and result controls.

The review CSV is intentionally a readable output rather than an operator-entry format. Inspect the slot, player identifier, salary and supplied projection for each row. A successful local export means only that the browser created a file; it does not mean an operator accepted the roster or that any entry fee was paid.

The optimizer clears previous results whenever a projection, status or constraint changes. That keeps an earlier calculation from looking like the answer to a new set of inputs. If you change the salary cap after generating a roster, solve again rather than copying the old output into another context.

The same principle applies to an imported file. A different slate can change eligibility, salaries and the available player pool even when many names look familiar. Review the import message and the table before solving. Invalid rows or duplicate identifiers produce a visible error; they are not quietly removed to make the pool appear usable.

A useful result is transparent enough to inspect: which slots were filled, which players were selected, how much salary was used and which supplied estimates were added together. Keep that review separate from any decision to enter a paid contest. The calculation does not make that decision for you.