SignalTuned. Beta Guide

Rankings tuned to the signal, not the noise.

Welcome. This is a private beta of SignalTuned, a math-grounded fantasy football engine I've been building. It does NOT use consensus opinions. Every ranking is computed from on-field NFL production data, calibrated to your league's exact scoring + roster + format, and the underlying model recalibrates against realized outcomes every season.

(The product was previously titled "Fantasy Football Engine." Same engine, official brand as of May 2026.)

You're getting an early look. Expect rough edges. The trade-off is that you can shape what gets built next.

Getting started (5 minutes)

Beta URL: https://signaltuned.com

Sign in. Click the user icon (top-right) and enter your email. You'll get a magic link, click it, you're in. No password, no oauth, no app install.
Connect your Sleeper league. Click "+ Add League" on the home screen. Two options:
Paste your Sleeper league URL or ID, OR
Type your Sleeper username. The app will find your leagues for you.
Pick yourself. When prompted, click your team name from the league roster. This is what unlocks personalized rankings.
Wait ~10 seconds. Roster syncs in the background. The header switches from "no roster data" to "Roster loaded." Refresh if it doesn't auto-update.
You're ready. All 7 tabs (Rankings, Recommendations, Trade Central, League Overview, Draft, Redraft Board, Keeper) are live.

What's in the engine

Every ranking comes from the same math layered together:

Weighted production. Recency-weighted PPG over your player's last 1-5 NFL seasons, scored under YOUR league's exact scoring settings.
Age curves. Position-specific decline schedules (RB peaks earlier than WR; QB has the longest tail).
Roster context. When "🎯 For my roster" is on (default), rankings re-sort by GAP, how much each player would lift YOUR Nth-best at his position.
Custom Fantasy Intelligence Layer. For ~700/829 active players: leverage-weighted volume, custom YPRR, NGS skill components (separation, time-to-throw, CPOE), volume quality + sustainability scores. None of these are paywalled or premium-data signals. All derived from MIT-licensed nflverse + NGS.
Trajectory + breakout flags. Surface the players the engine thinks are about to over-perform or decline.

What it explicitly DOES NOT do: - Aggregate consensus rankings from expert-polling or consensus-aggregator tools. The engine is supposed to disagree with those when the math says so. - Trust beat-reporter "vibes" or coaching speculation. - Use any data source where commercial use isn't explicitly OK.

Things that work today

Feature	Status
Rankings (QB/RB/WR/TE/FLEX, mobile + desktop)	✓
Sleeper league import + roster sync	✓
Personalized rankings ("🎯 For my roster" toggle)	✓
Trade Central (trade analyzer + multi-trade)	✓
Recommendations tab (buy/sell + roster grades)	✓
League Overview (your contention window + competitor scan)	✓
Pick Valuator (rookie pick valuation)	✓
Keeper analyzer (if your league does keepers)	✓
Redraft Board (separate from dynasty)	✓
Multi-platform support	Sleeper only for now (ESPN, Yahoo, MFL planned)

Things to expect rough

No mobile-app polish. It's a web app. Looks fine on iPhone Safari but not as snappy as a native app.
Cold-start lag. First /rankings call after a deploy takes ~10-30 seconds while the engine warms its cache. After that, sub-second.
No live in-season updates yet. The engine reads through the 2024 season currently. Weekly in-season refresh ships before the 2026 season starts.
Market arbitrage is dormant. During beta the "vs market" signals are turned off (no commercial-licensed market data source yet). Your rankings are still the engine's; you just don't see "engine vs FantasyCalc" deltas.
Rookie projections are scaffold-quality. A new Bayesian rookie model (Layer 5) is built but not yet calibrated against historical data. The current rookies use the prior model. Sensible numbers, not yet world-class.

What I most want feedback on

Did the sign-in + league import flow work cleanly? If you got stuck or saw an error, that's the highest-leverage feedback.
Do the rankings disagree with consensus in ways that feel right or wrong? Look at a player you have an opinion about. If the engine has them way higher or lower than you'd expect, ping me with the player + your read. The engine should be defensible. If it's not, that's a bug.
What did you reach for that wasn't there? Anything you wanted to do and couldn't find an answer to.
Did anything look broken / weird / slow on your phone or laptop?

How to give feedback

Text me directly. Screenshots are gold.
Or click the user-menu → "Switch account" if you want to test multi-account behavior (each email gets its own isolated leagues).

Privacy notes

I store your email (for the magic link) and the Sleeper league IDs you connect. That's it.
No tracking, no ads, no analytics, no resale.
A formal privacy policy + ToS will be in place before any public/commercial launch. Friend-beta operates under a personal-use understanding.
You can email me to delete your account + all data at any time.

Heads up: this is alpha-beta

The whole platform is being deployed every few hours as I ship changes. If you hit a 502 or "Application failed to respond" page, it's likely a deploy in progress. Refresh in 30-60 seconds. If it persists, ping me.

Thanks for being a guinea pig. The point of the beta is for me to find bugs before launching to anyone else, and your feedback is the entire reason this stage exists.

Ryan