How To Build NHL DFS Lineups: Stacks, Uniques & Randomness

Learning how to build NHL DFS lineups is really learning how to build a portfolio. Hockey scoring travels in packs: goals are rare, almost every one carries assists, and the skaters collecting those points share an even-strength line or a power-play unit. Which is why daily fantasy hockey on DraftKings and FanDuel is a stacking game first and a player-picking game second. And once you move past hand-building two or three entries a night, the edge stops living in who you like and starts living in the settings that assemble your builds: stacking rules, uniques, randomness, exposure. This guide walks through that workflow end to end, from loading projections to uploading a finished CSV. One number matters more than any other in it, and it is four. We will get there.

If terms like line stacking or confirming the starting goalie are new to you, start with our NHL DFS beginner guide. This article assumes those fundamentals and covers the construction layer on top of them.

In Summary (TL;DR)

• Stack Full Even-Strength Lines. A standard NHL tournament build pairs two three-forward line stacks, then fills defense and goalie around them.
• Uniques Are Your Diversification Dial. In NHL, four uniques is the threshold that stops the same two-line pairing from repeating across your entries; it also roughly triples your player pool.
• Randomness Widens The Builder's Vision. At 10%, a 5.0-point projection is treated as anything from 4.5 to 5.5 on each new lineup, which spreads your exposure.
• Correlate Your Goalie with at least three forwards from his own team so your netminder wins when your stack does.
• Upload Via CSV, Then Verify. Export against the correct slate, paste into the site's entry template, and check your live entries page before lock.

Watch The Video

The full walkthrough this guide is built on, including the on-screen build of a 20-lineup two-stack portfolio, is here:

Watch on YouTube

Why Building NHL Lineups Is A Portfolio Problem

Hand-building works fine for a couple of entries. Around 20 or 30 lineups it collapses, not because clicking players takes too long but because you can no longer see your own portfolio: how often one line appears, how many entries differ by only a single defenseman, whether your goalie ever opposes your stack. A lineup builder exists to enforce those rules for you at scale. But a builder is only as smart as its settings, and the default settings are built for sports where correlation is optional. In hockey it is the whole game, which is why the settings below are NHL-specific.

The raw material comes first, though. Every build starts with projections.

Step 1: Load Projections And Sort By Value

Whatever builder you use, it will optimize toward two columns: projected fantasy points and value, meaning projected points per $1,000 of salary. Value is the column doing the real work. The optimizer's job is to cram as much projected scoring under the salary cap as possible, so a mispriced third-line winger with a huge value number will show up in nearly every build the tool spits out. Know this before you crunch anything: an outlier value score is the single biggest lever on your exposures, so decide whether you believe it before the builder decides for you.

Do not feel obligated to project players yourself. Building your own projections from scratch is a spreadsheet project measured in weeks, and unless you are unusually comfortable in Excel it is a headache with no edge attached. Start from a trusted source instead, like Stokastic's NHL projections in the DataHub. Whatever source you use, make sure it tells you each skater's line and power-play unit, because that context is the whole next step.

Step 2: Set Stacking Rules Around Full Lines

Because assists travel with goals, the standard NHL tournament build is two stacks of three forwards, each stack a complete even-strength line, with defensemen and a goalie filled around them. In your builder's team-stacking settings that means: forwards only, line-level grouping, three players per stack.

Say you want 20 lineups anchored to one team's second line. Set stack one to that specific line, then clone the rule for stack two and leave it open to first and second lines from any other team. The builder now has a fixed anchor and a rotating partner. Which brings us to the first trap in the whole workflow: run that exact build with default settings and you will not get 20 different lineups. You will get one pairing, 20 times.

Step 3: A Worked Example In Uniques (Why Four Is The NHL Number)

With no diversification settings, the optimizer has no reason to deviate from the highest-projected combination, so it will marry your anchor line to the same partner stack in every single entry. Uniques fix that. A uniques setting of N forces every lineup to differ from every other lineup by at least N players.

Here is the worked example, and the payoff of the number four. An NHL line stack is three forwards. If your uniques setting is three or lower, the builder can satisfy it while still repeating the same two-line pairing, differing only in goalie and defense. Set uniques to four and the math changes: two lineups can no longer share both full three-man stacks, so no line pairing ever repeats across your portfolio. The video's 20-lineup crunch shows the difference on a real 7-game slate:

Uniques Setting	Players in pool	Can a two-line pairing repeat?	What actually varies
1 (Default)	27	Yes, freely	Often just a goalie or one defenseman
4	83	Never	The partner stack itself rotates

The pool number is the row worth staring at: one setting change took the same slate, with the same projections, from 27 rostered players to 83. One setting is the entire gap between lineups that look different and lineups that are different.

More uniques means more players in your pool, which cuts both ways. Aggressive pros run fewer uniques on purpose: if they love a line, they are comfortable running the same pairing at 40% or 50% of a 150-entry block, alternating only the goalie and a defenseman. Done knowingly, it is a conviction play, not a mistake. But until you know exactly what your settings are doing, four uniques is the honest default for NHL, because it forces real roster diversity instead of look-alike lineups.

Uniques diversify by rule. There is a second dial that diversifies by noise.

Step 4: Randomness Widens The Projection, Not The Rules

Randomness tells the builder to treat every projection as a range instead of a point. At 10% randomness, a skater projected for 5.0 fantasy points is priced anywhere between 4.5 and 5.5 each time a new lineup is generated; at 20%, that window stretches from 4.0 to 6.0. The practical effect is more players and more line combinations drifting into your builds, because the "best" lineup is slightly different on every pass.

Uniques and randomness solve the same problem from different directions, and stacking both before you understand each one is how portfolios go sideways. A sensible progression: uniques only, until your exposures make sense to you; then layer in 10% randomness if you want wider coverage of a slate. Skip the temptation to force exposure with three other common features first:

• Boosts And Docks (thumbs up or down on a player) typically move that projection by a fixed amount per click, about 8% in the builder shown in the video. Two clicks is a 16% swing, which will quietly warp an entire crunch.
• Locks force a player into 100% of entries, which is almost never what you actually believe.
• Groups are the scalpel: a rule like "at least one of these three defensemen in every lineup" shapes exposure without forcing any single name.

If you want a player in roughly 70% of your entries, the clean path is adjusting his projection or running multiple smaller crunches, not locking him and hoping.

Step 5: Correlate Your Goalie With Your Stack

Goalies score on wins and saves, and your stacked team scoring three goals is exactly the scenario where its goalie banks a win. So add one position rule: your goalie must come with at least three same-team forwards. In a two-stack build, that snaps the goalie to one of your stacks automatically, and every lineup becomes one story: this team controls the game, its line scores, its goalie wins. Position rules are powerful enough to break a build if you overdo them, so start with the goalie rule alone before experimenting further.

One warning from the beginner guide bears repeating here because it voids everything above: confirm the starting goalie before lock. A backup start turns a correlated portfolio into 20 copies of the same zero.

Step 6: Export, Upload, Verify

The mechanical last mile, same routine every slate:

1. Export against the correct slate. Exporting against the wrong slate scrambles every player ID, and the site will reject or misread the file.
2. Download the entry template from the DraftKings (or FanDuel) edit-entries page for the contests you have entered.
3. Paste your exported lineups into the template's lineup rows. The export already carries each player's site ID, so it is a straight copy-paste.
4. Upload, then verify. Open your live entries page and confirm the lineups actually changed. Ten seconds of checking beats discovering a slate of placeholder lineups after lock.

And with that, the classic workflow is complete: projections in, stacking rules, four uniques, measured randomness, a correlated goalie, CSV out. Which leaves the honest question of whether the classic workflow is still the best one.

Where Sims Beat The Classic Optimizer Build

Look back at what randomness is doing: faking uncertainty by jittering a single projection. That is a crude stand-in for what you actually want, which is a real distribution of outcomes for every line, skater and goalie on the slate. That is the difference with Stokastic's NHL Sims, which simulate the slate thousands of times and rank builds by how often they take down a tournament rather than by one projected score with noise sprinkled on top. The stacking logic you just learned still applies; the Sims simply replace the jitter with modeled variance and fold in ownership, which is the piece a classic optimizer never sees. Our Sims-versus-optimizers breakdown covers that gap in depth. For the line-pairing decision this whole guide revolves around, there is the NHL Top Stacks tool: the tool surfaces the estimated chance a given three-player line stack lands in the winning lineup, alongside that stack's projected ownership, which is the exact tradeoff you were approximating with uniques and randomness. And if your volume lives on pick'em boards rather than salary-cap contests, Stokastic Prop Tools (PrizePicks and Underdog projections) covers that format directly.

You can try the Sims free at the link above. When you are ready to run full NHL slates with projections, ownership and simulated win rates in one place, code NHLLINEUPS10 takes 10% off your first payment.

Get Stokastic NHL Sims

FAQ: Building NHL DFS Lineups

How many uniques should I use for NHL DFS? Four is the standard for even-strength line stacking on a full slate, because an NHL stack is three forwards and four uniques prevents any two-line pairing from repeating. Slate size changes the answer, though: on a 2-3 game slate there are only a handful of viable lines to rotate through, so a high uniques setting can leave the builder unable to fill your lineup count. Scale uniques down as the slate shrinks, and drop to two or three on big slates only as a deliberate conviction play on a line you love.

Do I need my own projections? No, and for most players building them is a mistake; it is weeks of spreadsheet work to reach parity with numbers you can already get. Use a trusted source like Stokastic's NHL DataHub and spend your time on stack selection instead, or go a level deeper with our full NHL tools walkthrough.

Hockey rewards the builder who respects its structure. Points cluster in lines, so you stack; stacks duplicate, so you set four uniques; projections are guesses, so you widen them; the goalie rides the same wave, so you correlate him. Get those four decisions right and the last mile is just a clean CSV.