Survivr

Overview

Portfolio

Preseason

Dead entries

Pools & Entries

Contest

Pick history - all entries in this contest, by week

Current optimized pick path - the single best max-survival line

How hard is each week to survive? - the safest pick still available after your plan

Safe · 78%+ Solid · 70-77% Risky · 64-69% Trap · under 64%

Splash Sports NFL Survivor - rules that drive strategy

One team, straight up, each week; never reuse a team. Some contests require 2+ picks in designated weeks - use the format selector in the Pools & Entries tab.
Tie = not a win = elimination. Our probabilities treat ties as losses.
Pick deadline: Sunday 1:00 PM ET (earlier for TNF/early games). No pick = loss. Set picks early; adjust before kickoff.
Postponed/flexed out of the week = loss. Avoid games with weather/venue uncertainty when alternatives are close.
Co-survivors split prizes at Week 18 (or the latest week with a survivor). In small pools, if everyone else dies with you on the same chalk, your equity survives via the split - mild extra argument for chalk in small pools; in large pools, differentiation matters more.

How to use this dashboard

Before the season / each week: open the Picks tab. The recommended pick is assumed made each week - click any pick (board or portfolio table) to override it; past weeks stay editable if you strayed. Follow the current week’s pick unless news breaks.
Tuesday–Sunday of game week: check the Weekly Board - real lines replace projections automatically as they post (the board pulls the latest odds every time you open it).
Multi-entry: create your entries in Pools & Entries - the portfolio section of the Picks tab plans a diversified path for every one and re-optimizes around the survivors as results come in (an entry whose pick loses is out).
Multi-pick contests: pick the matching preset - team scarcity binds hard (25+ picks from 32 teams) and planning is worth ~50% more survival.
Hunch board (Preseason tab): each conference is shown in the market's strength order - drag teams to where you think they'll finish. Moves are converted onto the market's own point scale (slot values), shrunk by your conviction, and applied only to projected weeks (HUNCH tag) - posted lines always keep the market's word. The optimizer burns your fallers early and can hold your risers for later weeks. At kickoff the board locks and each claim fades on evidence: ~4 games of weight, capped at the original claim, faster when the market moves against you, zero once priced in - so early-season plans still lean on your read while wrong calls die off on their own.

Methodology

Probabilities. Weeks with posted lines: consensus moneylines, de-vigged, then recalibrated on 5,065 games (slope 1.08 - corrects the favorite-longshot bias). Future weeks: market-implied power ratings (fit to 2026 win totals + the 83 posted early-season lines, HFA 1.5) plus an in-season Elo overlay - each final score updates team strength by its margin vs the closing spread (25 Elo pts = 1 pt of spread, 0.10 pts of rating per point of cover margin; FiveThirtyEight-style MOV update as fallback when no line existed; seeded from the win-total prior), so unpriced weeks track current form. Spread → win probability via the historical margin distribution (σ=13.45), plus schedule-known rest adjustments (+1.2 pts off a bye), then horizon-decayed and recalibrated. Backtested 2006–25: future-week log-loss falls from .666 (static preseason ratings) to .643; games with posted lines still use the market unchanged.

Optimizer. Hungarian assignment maximizing Σ log P(win) across all remaining picks (supports multi-pick weeks, used/banned teams, mid-season entry). It reserves teams for weeks where they’re the only good option and spends redundant safety now.

Validation. Every effect was tested with the closing line as an offset (walk-forward out-of-sample, 2012–2025). Claims that failed - including overriding the market’s weekly #1 - are disclosed in the Evidence tab, not shipped.

Data: nflverse (1999–2026 schedules, results, closing lines), DraftKings win totals (May 2026), posted 2026 lines as of July 2, 2026. Projections update as lines post - re-check before every deadline. Not affiliated with Splash Sports. No guarantee of outcomes; play responsibly.

Pick history - all entries in this contest, by week

Current optimized pick path - the single best max-survival line

How hard is each week to survive? - the safest pick still available after your plan

Team usage - entries slotted per team, by week

Bottleneck radar - where the field will herd (projected pick concentration for this contest, ⚑ = leverage target)

Dead entries - eliminated from their contests

Preseason hunch board - drag teams to where you think they'll finish (locks at kickoff, fades on evidence)

QB durability board - fragile starters get spent, not saved (feeds the season plan, the portfolio & the field sim)

Market calibration: favorites 2006–2025 (5,065 games)

The weekly “safest pick” by week - how it actually went

Information decay - why you can’t bank on Week 15 chalk today

What survived 20 years of backtesting (vs the closing line)

Big-favorite losses - the 15-year tail study (2011–2025)

Second tier - every big-favorite (≥78%) loss, 2011–2025 the full audit behind the flags

All losses (newest first)

Tested in 2026 - every challenger rejected

The honesty box - what does NOT work

Splash Sports NFL Survivor - rules that drive strategy

How to use this dashboard

Methodology

Pool settings

Entries