The method
How a value is made
Every number on RingChase comes from one question: how many fantasy points will this player score above a replacement starter in your league, and for how long?
The formula
value = Σ over 6 seasons of 0.85t × P(still relevant) × games × E[max(0, PPG − replacement)]
Each season is valued on its own and then added up. Later seasons count less: 15% per year, for time and uncertainty. Taking the expected surplus over the player's full range of outcomes means young, unproven players keep their upside, while aging depth players fall toward zero.
The pieces
A recency-weighted average of real NFL production, shrunk toward what a player drafted at his slot usually scores. The fewer games he has played, the more the draft slot matters.
The chance a player has a real role each future season. It's fitted on NFL history by position from age, recent playing time, production and draft slot. It covers both late-career drop-offs and young players growing into a role.
How production changes with age, learned from what players actually did. Not hand-drawn curves.
Taken from your league's lineup: team count, starters, flex, superflex and scoring. A 14-team superflex league has a very different replacement QB from a 10-team 1QB league.
A QB's value each season is scaled by his chance of starting, from the current depth chart.
Two horizons and your team
Dynasty is usually played in 3-year windows, so every player gets a 3-yr window value beside his 6-season dynasty value. The shape badge shows how front-loaded his value is: Win-now if more than 75% comes in the window, Future if less than 50%.
The Your team setting reweights seasons in the dynasty value. Contend leans on the next three seasons and Rebuild leans on seasons two through six. The window value stays the same.
The market is a benchmark, not an input
We never use other sites' values (KeepTradeCut, FantasyCalc, DynastyProcess, expert consensus) to build ours. The market shows up in only two places: the backtest we grade ourselves on, and the BUY/SELL flag where our rank and the market's are far apart.
How we test it
We freeze the model at past dates (September 2021, 2022 and 2023) using only data available then. We score how well it ranked what players actually went on to produce, and run the market through the same test. Every model parameter is fitted on 2010–2020 only. A change ships only if it improves the test without getting worse on a separate earlier validation period.
Today we roughly match the market one season out and trail it at three. The gap is mostly players whose situation changed in the offseason: new teams, depth charts, signings. That's exactly what comes next.
What's next
- Team context. Depth charts, team changes, vacated targets and carries, team volume, expected touchdowns, a dedicated rookie model.
- Title odds. Simulate your league week by week. A player's value becomes how much he raises a roster's chance of winning a title. Stars beat depth for a reason the math can show, and the same player is worth more to a contender than to a rebuild.
- Draft picks. Valued from what players drafted at each slot have actually produced.