Methodology
How the Value is arrived at, and how far off it is against real trades.
Every dynasty site publishes a number. This page is about how ours is arrived at, and — the part nobody publishes — how far off it is when it meets a trade two real managers actually agreed to.
What the Value is
It is a model of scarcity in your league, not a table of prices. The same player is genuinely worth different amounts in a 10-team and a 14-team league, and the number moves accordingly.
- The replacement line. Every asset is scored against the last player your league can actually roster. That player is freely available, so he is worth nothing — and everyone above him is worth what he is worth more than nothing.
- A premium for starting. Players your lineup has to field every week are scarcer than the ones on your bench, and the premium tapers as the requirement is filled rather than falling off a cliff at the last starter.
- An age curve per position. Careers are not the same shape. Backs fall off first and fastest, receivers peak later and hold it, tight ends arrive slowly and decline gently, quarterbacks are productive into their mid-thirties. Deliberately shallow — the market already prices age, and a steep curve counts it twice.
- A weighting between now and later. A win-now roster and a rebuilding one do not want the same players, and the same asset is priced differently for each.
The inputs are public market data and this season's projected scoring. What the model does with them — the scarcity, the curve, the league in front of it — is the product.
How wrong it is
Every week we pull completed trades out of real Sleeper leagues and price both sides on the chart as it stood. If the model were perfect, two managers who agreed to a deal would be trading equal value and every gap would be zero. They are not, and it is not.
Read that middle number as the honest one: on 688 usable trades from 464 leagues over 90 days, the median gap we show between two sides who shook hands is 14.9%. A quarter of trades come in under 6.3%; a quarter are over 27.2%.
The slope of 0.91 is the other half. It says that when one side gives up more by our reckoning, they get back very nearly proportionally more — which is what you would want from a chart that tracks a real market, and it is measured across 1,376 sides with an r² of 0.83.
Where we are off, by position
The same run solves for what our values would have to be multiplied by, per position, for those trades to balance. This is our own measurement of our own error:
| Position | Scale | Reading |
|---|---|---|
| Quarterbacks | 0.99× | About right |
| Running backs | 1.15× | We price these about 15% low |
| Wide receivers | 1.08× | We price these about 8% low |
| Tight ends | 0.92× | We price these about 8% high |
| Rookie picks | 0.85× | We price these about 15% high |
Not corrected away silently. A model tuned until its own regression reports zero error is a model fitted to its test, and the next quarter's trades would find it out. These are published so the correction is visible when it is made.
What the measurement cannot see
- Not every trade can be priced. 3,953 of 5,218 items in the window had a value; the rest were startup picks, FAAB and far-future selections that no chart covers. Trades containing them are dropped rather than guessed at — 492 of 1,180 in this run.
- An agreed trade is not a fair trade. Managers accept bad deals, and rebuild, and do favours. A gap is evidence about the chart only in aggregate.
- It is a snapshot. The run is dated 2026-08-14 and covers the 90 days before it.
What the values are not
They are not a prediction of what a player will do on the field, and they are not advice. They are an estimate of what an asset is worth to your roster in your league, against a market that changes daily.
Why the history matters
The other sites in this category publish today's number and only today's number. Ask any of them what a player was worth last March and there is no answer — the figure on the page is always the current one.
charted.gg keeps a dated archive, taken every morning and never rewritten. That is why every player page can draw where his value has been, and why this page can quote an error rate at all: measuring a chart against trades requires knowing what the chart said on the day the trade happened, and a site that only knows today cannot do it.
Values on the site are from the snapshot taken 2026-08-15; the settings above default to a 12-team superflex league. The regression is re-run by hand — see the board, who has moved and the calculator.