Pick6 vs PrizePicks vs Underdog: A Cross-Book EV Calculator

Three apps own the pickem market: DraftKings Pick6, PrizePicks, and Underdog Fantasy. Every pickem review site ranks them by interface and bonus offers. Almost none compare the math. That gap matters because the same 4-leg slip can return $200, $250, or $410 depending on which app you submit it to — and the cheapest book depends on your edge, your leg count, and the field’s positioning on the slate. This is the cornerstone reference for a pickem expected value calculator that routes across all three. Plug your projection in, pick the highest-EV book, repeat.

The three payout engines, side by side

Pickem looks identical across books — pick two to six player props over or under a line, hit them all, get paid. Under the hood the settlement math is wildly different.

PrizePicks: fixed multipliers

PrizePicks is a peer-to-peer DFS contest, but operationally it functions like a parlay book. Their published Power Play payouts (current as of 2026 across most states) are:

• 2-leg Power: 3x
• 3-leg Power: 5x
• 4-leg Power: 10x
• 5-leg Power: 20x
• 6-leg Power: 25x

Flex Plays trade lower top-end payouts for insurance — go 5 of 6 on a Flex and still cash 2x your entry. Flex multipliers run 2-leg 3x with 1-of-2 protection paying half-back, 3-leg paying 2.25x on 2-of-3, 4-leg paying 5x on perfect / 1.5x on 3-of-4, all the way up to 6-leg Flex paying 25x on perfect / 10x on 5-of-6 / 2x on 4-of-6. We dig into the full Flex math in PrizePicks Flex vs Power Plays.

DraftKings Pick6: parimutuel pools

Pick6 is a different animal — it’s a parimutuel pool, similar mechanically to horse racing. Every entry pays into a single weekly pot. The pot pays out to perfect entries first, then to next-best, and so on, with each tier’s share scaling to how many entries hit it. Two consequences:

1. Pick6 has no fixed payout. A 6-leg perfect on a chalk slate might pay 80x; on a contrarian slate it can hit 140x or higher.
2. Pick6 rewards differentiation. If you fade public favorites, you skip the bottleneck where 60% of perfect entries cluster.

Published Pick6 weekly recaps from late 2025 showed perfect 6-leg entries paying $7,613 on a $100 entry (76x) during a chalk-heavy NFL Week 12, and $14,221 ($142.21 on $1) on a more contrarian Week 14. The structure is closer to a tournament than a parlay. We unpack it in DraftKings Pick6 Payout Curves.

Underdog Pick’em: middle of the road

Underdog uses a hybrid. Higher (their fixed-odds Power equivalent) pays 2-leg 3x, 3-leg 6x, 4-leg 10x, 5-leg 20x. Standard (Flex equivalent) layers in insurance similar to PrizePicks Flex with slightly different protection bands. Underdog’s 3-leg Higher at 6x is structurally generous — PrizePicks pays 5x for the same correlation profile.

The cross-book EV formula

The math is simple enough to fit on a napkin. For a fixed-odds book (PrizePicks Power, Underdog Higher):

EV = (product of per-leg hit probabilities) × payout multiplier × stake − stake

For a parimutuel book (Pick6) you need an extra term — the expected payout, which depends on how many other entries hit your tier:

EV_Pick6 = (product of per-leg hit probabilities) × E[payout multiplier | perfect] × stake − stake

The expected payout multiplier given a perfect ticket is the trickier piece. We approximate it from the public split on each leg — if the four legs you picked are each chalk at 70% public, your perfect entry shares the prize pool with ~24% of the field, so the multiplier compresses. If the four legs are all under 30% public, your perfect entry shares with ~1% of the field and the multiplier explodes.

A worked 4-leg example

You build a 4-leg entry at $25 stake:

• Patrick Mahomes over 268.5 passing yards — your model: 58%
• Travis Kelce over 5.5 receptions — your model: 62%
• Bijan Robinson over 75.5 rushing yards — your model: 60%
• Tyreek Hill over 78.5 receiving yards — your model: 56%

Joint perfect probability: 0.58 × 0.62 × 0.60 × 0.56 = 0.1208 (12.08%).

Under each book:

• PrizePicks 4-leg Power (10x): EV = 0.1208 × $250 − $25 = +$5.20. Edge = +20.8%.
• Underdog 4-leg Higher (10x): Same formula, same EV = +$5.20.
• DK Pick6 (parimutuel, E[mult] = 11.5x in our example): EV = 0.1208 × $287.50 − $25 = +$9.73. Edge = +38.9%.

Pick6 wins this slip. Now redo the same legs with all four players being heavy public chalk (think Mahomes, Kelce, McCaffrey, Lamb in primetime). E[mult] on Pick6 compresses to 7.5x because too many entries hit your tier. Now PrizePicks’ flat 10x is better. The cross-book calculator at /pickem does this comparison automatically — you input projections, it spits out the highest-EV book.

When parimutuel beats fixed-odds

Pick6 has a structural edge whenever the slate is heavily contrarian — i.e., your perfect entries don’t cluster with the field. Three diagnostic signals:

1. Average public split on your legs is below 50%. If you’re fading chalk on most legs, Pick6’s pool dilution drops and your effective multiplier rises.
2. At least one leg has a public split below 25%. A single low-owned leg gates a large share of would-be-perfect entries.
3. Slate is small. Pick6’s pool is fixed by entries, not by event count. A Thursday Night Football mini-slate can deliver 200x multipliers on perfect 4-legs.

PrizePicks wins whenever your edge sits on chalk. If every leg is 65%+ to hit and the public agrees, the fixed multiplier protects you from Pick6 pool dilution.

The break-even win rates you need

Inverting the EV formula, the per-leg hit rate you need to break even (assuming independence) is p_break = (1 / multiplier)^(1/legs):

• PrizePicks 2-leg Power (3x): 57.7%
• PrizePicks 3-leg Power (5x): 58.5%
• PrizePicks 4-leg Power (10x): 56.2%
• PrizePicks 5-leg Power (20x): 54.9%
• PrizePicks 6-leg Power (25x): 56.0%
• Underdog 3-leg Higher (6x): 55.0%
• DK Pick6 6-leg (E[mult] ≈ 100x): 46.4%

The 5-leg PrizePicks Power is the easiest fixed-odds break-even at 54.9%. Pick6’s headline 46.4% is misleading — it requires the parimutuel pool to actually deliver 100x, which only happens on contrarian slates. The realized break-even on chalk Pick6 slates is closer to 53-55%, similar to PrizePicks.

A standard sportsbook spread at -110 needs 52.4%. Every pickem app demands real edge on top of that.

Correlation: where simple EV math breaks

The formula above assumes leg independence. In practice, pickem legs are correlated. A Mahomes over passing yards leg correlates positively with a Kelce over receptions leg (same offense, same game script). A Bijan rushing over leg correlates negatively with a Falcons team-total under (if Bijan is hitting, the team is moving). PrizePicks blocks the most obvious same-game correlations on Power Plays (no QB + WR in same lineup), but cross-game correlations from game-script clustering are unrestricted.

The implication: when your legs correlate positively, your true joint probability is higher than the product, and your real EV is better than the calculator says. When they correlate negatively, EV is worse. The cross-book calculator on /pickem exposes a correlation slider — drag it to 0.1 for typical same-game stacks and the displayed EV rises 10-25%. Our same-game parlays math piece covers the correlation theory in depth.

Bankroll: which app for which budget

Each app has a different bankroll personality.

• PrizePicks: low variance per slip (fixed payout, no field surprises). Ideal for steady players who want to log thousands of entries and let edge compound. Pair with our bettor desk to track CLV across slips.
• Underdog: similar variance profile to PrizePicks. Better on 3-leg Higher (6x vs 5x); slightly worse on 6-leg.
• Pick6: high variance per slip (parimutuel pool surprises). One contrarian 6-leg perfect can fund 100 entries. Use in moderation; cap weekly spend.

The pickem desk at /pickem tracks per-book EV, hit rate, and ROI separately so you can see which book is actually paying you.

Live updating: ignore the public, follow your model

Pickem lines move with public action just like sportsbook lines. PrizePicks shifts a passing-yards prop from 268.5 to 266.5 if 65% of entries take the over. Underdog moves similarly. Pick6 doesn’t shift lines but does shift implicit payouts as the pool weights move. Whichever book you use, lock your projection first, then check whichever book’s line is most divergent from your number. Locking the line before submitting is a habit. Tracking is the audit.

Pulling it together

Cross-book pickem math is not optional for anyone betting more than a few entries a month. The same 4-leg slip can give you +$5 or +$10 of EV depending on the book. Build your projections once in /tinker (NFL spread/total models, NBA prop models, MLB hit/HR models — they all output per-leg probabilities). Push the per-leg numbers through the EV calculator at /pickem. Take the book with the highest EV. Log every slip. Recompute after every Sunday.

That loop — projection → cross-book EV → entry → log → recompute — is what separates the recreational pickem player from someone with real positive EV across hundreds of weekly slips. The math is small. The discipline is everything.

A deeper look at the per-leg math

The per-leg hit rate is the single most important input. Two ways to estimate it from your model:

1. Direct projection. Your model outputs a mean and standard deviation for each player prop (e.g., Mahomes passing yards ~ Normal(283, 41)). For a line of 268.5, the over probability is 1 − Φ((268.5 − 283) / 41) = 1 − Φ(−0.354) = 0.638. Round to 64%, that becomes your per-leg input.
2. Calibrated bucket. Look at historical hit rates for similar lines at similar offered numbers. If sub-key Mahomes passing-yard overs at -1.5 vs prop line have hit at 61% across 187 historical games, use 61% directly.

Both approaches need calibration. Track the realized hit rate of your projected 60-65% buckets across at least 200 logged legs. If your projected-60% legs hit 56% in reality, your projections are over-confident by 4 points. Apply a flat shrinkage and recompute every slip. Calibration is the difference between a model that prints and a model that confidently bleeds bankroll. Our closing line value piece is the rigorous validation tool — beat-the-close pace tracks real edge faster than realized hit rate.

Common per-leg projection mistakes

• Ignoring usage variance. A WR projecting 6.5 receptions with high target variability is genuinely a worse over than a WR projecting 6.0 receptions with low variability. The standard deviation matters more than the mean.
• Treating recent form as a step change. Tee Higgins coming off 9 and 11 reception games doesn’t shift the model to 80% on his next over. Bayesian shrinkage toward his season baseline is the correct adjustment.
• Double-counting injury news. If a starter is out, the public bumps the backup’s line. By the time the line lands on the board, the news is priced in. Your edge comes from forecasting the news, not reacting to it.

Bonus offers and the EV of a one-time boost

Each book runs constant promotional traffic — deposit matches, first-entry free, "get $50 if your first slip wins" type stuff. These create one-time bumps to EV that should NOT change your structural routing. Two clean rules:

1. If a promo says "get $X in site credit if your first slip wins" — value the site credit at 60-70% of face (because conversion to real cash is friction-heavy).
2. Don’t enter a slip just because a promo runs. The promo bumps EV by a fixed amount; if the underlying slip is negative-EV by more than the promo bump, you’re still losing in expectation.

Track promotion EV separately in your log. Most bettors mentally over-weight the promotional value and bleed elsewhere to capture it.

Tax considerations on cross-book pickem

Pickem winnings are gambling income at the federal level (with state variation). Three operational notes worth flagging because they show up at year-end and surprise people:

• Per-book reporting. Each operator issues separate 1099 forms when annual net winnings clear $600. If you split entries across PrizePicks, Pick6, and Underdog, you may receive three forms.
• Losses are itemized only. Federal law allows gambling losses to offset winnings only if you itemize deductions, capped at total winnings. The standard deduction usually wins for most filers — meaning realized winnings are taxed without offset.
• State variance. A handful of states (notably New Jersey) treat DFS-style pickem differently from sportsbook winnings. Verify your state’s treatment before tax season.

None of this changes the EV math directly, but a 30% effective tax rate on net winnings means your operational hurdle rate is closer to 30% above break-even, not 0%.

The contrarian-vs-chalk decision tree

Before you submit any slip, run through this:

1. What is my joint perfect probability across all legs?
2. What is the joint public-clustering probability (product of public selection rates)?
3. If clustering < 1%, prefer DK Pick6 for the parimutuel multiplier.
4. If clustering > 3%, prefer PrizePicks Flex (6-leg) or Power (3-5 leg) for the fixed multiplier.
5. If a same-game stack is involved, force Flex (Power blocks most stacks).
6. If 6-leg, default to Flex (strictly better than Power across all realistic hit rates).

This tree captures 90% of the routing decisions. The remaining 10% involve specific promotion overlays, withdrawal limits, or per-state availability and require manual judgment.

Slate-construction patterns that show up across hundreds of slips

After logging several hundred entries across all three books, three persistent patterns emerge that the math alone won’t tell you:

Game-script over-stacking

Bettors love stacking the QB + WR1 over on the same passing-script narrative. The math on independence says it’s correlated +EV; the reality on PrizePicks is that those lines tighten 2-3% precisely because the book knows the correlation. The +EV you’d expect from correlation gets clawed back at the line. The same stack on DK Pick6 is unchanged because Pick6 doesn’t adjust lines for correlation — they let the parimutuel pool sort it out.

Late-week injury-news entries

Wednesday-morning entries vs Saturday-evening entries on identical slates show dramatically different per-leg quality. Late-week entries benefit from injury reports, depth-chart confirmations, and weather forecasts that all narrow the per-leg distribution. Sharps almost always wait until late Friday or Saturday morning before locking. The five-day-old entry from Tuesday is structurally worse no matter how good the model is.

Cross-sport diversification

A pure NFL Sunday slip variance is large because all six legs settle on the same afternoon. A slip that mixes NFL legs with NBA or NHL legs (where books allow it) reduces variance per slip because settlements stagger across days. The portfolio-level Sharpe ratio improves even when per-slip EV is unchanged. Underdog supports cross-sport slips most generously; PrizePicks restricts some combinations; Pick6 varies by contest.

Putting the calculator to work end-to-end

The day-of-slate workflow:

1. Pull projections. Open /tinker. Generate per-player prop projections for the slate.
2. Identify edges. Filter for projected hit rate above 56% (the rough floor for all three pickem books on 4-5 leg slips).
3. Build candidate slips. Stack 4-6 legs. Check for same-game pairs (force Flex if used). Compute joint perfect probability.
4. Route via the cross-book desk. Plug the slip into /pickem. The calculator compares PrizePicks, Underdog, and Pick6 EV after fees. Take the winner.
5. Submit and log. Enter on the routed book. Log the slip in /desk with per-leg projections and final book chosen.
6. Settle and review. After slate close, mark winners/losers. Compare realized payouts to projected EV. Recalibrate per-leg projections if a pattern emerges.

Research note board

Use this board before clicking a prop, DFS build, or same-game entry. The table is intentionally about thresholds, not fake certainty.

Bet responsibly — pickem variance is real, set deposit limits, never chase losses.

Breakeven win % at common American odds

The win rate you need to break even at each price. Pick odds shorter than -150 and you must win >60% just to stay flat — a hurdle most casual handicappers never sustain.

Prop OVER hit rate vs line distance from median

Empirical hit rate of OVER bets as the prop line moves away from the player projection median, measured in standard deviations. A line set 1sd below the median hits ~84% of the time — but books price the juice to match.