Check the record. Wins, misses, and the math behind them.
A win rate needs context too. See which seasons AVRA tested, what it compared against, and which results passed the statistical checks. Results that fail those checks stay separate so you can judge the record for yourself.
Open beta access is free. Pro and All-Access are available now.
What AVRA's calls did against the experts.
right
When AVRA disagreed with the public rankings in 2025 testing, AVRA was right nearly 2 out of 3 times (65.04%) across 25,786 calls.
The number above has the strict statistical check applied before it reaches this page. A provisional badge means AVRA cannot yet certify that check, so AVRA does not claim it.
Draft day sets the roster. Trades keep changing it.
better side
How often AVRA picks the better side of a real league trade, measured on the actual trades managers made and on what those players went on to score.
Anyone can look right after the games. AVRA locks its call before kickoff.
Every pre-registered call is locked the moment it is made, with a tamper-proof timestamp that cannot be edited once the ball is in the air. The record can never be quietly cleaned up after the fact.
The first verdict was locked on August 4, 2026. Each one carries a hash of its contents, so any later edit would break the seal.
The number is only as good as what you do to try to break it.
AVRA applies a strict statistical check.
The disagreement rate above only appears once a stricter check clears it: AVRA discounts repeat decisions and the many looks it takes at the data, so a hot streak cannot pass itself off as skill.
AVRA shows what does not clear the bar.
AVRA also ran a Superflex lineup-construction gate. It did not survive the strict statistical check, so it is marked provisional and not claimed. A tool that only shows you its wins is hiding the rest of the table.
AVRA holds back the seasons it cannot trust.
Earlier seasons whose projections were built with information that only arrived later sit in a separate, flagged block. They never pool into the headline number, because that would quietly inflate it.
Every number passes an automated check.
The rates, the statistical check, and the direction of every result are rechecked by continuous integrity gates before anything is allowed to reach this page.
AVRA measures the whole lineup, not just single calls.
Across 51,840 team-weeks in 2025 testing, AVRA's lineup calls reached 91.25% of the mathematically optimal lineup and added +3.77 percentage points of championship probability across 115,200 simulated seasons.
Questions about the record
How does AVRA check the number on this page?
AVRA counts every call where it disagreed with the public rankings and how often AVRA was right. Before that number appears, a stricter check discounts repeat decisions and the many looks AVRA takes at the data, so a hot streak cannot pass itself off as skill.
What is the frozen-before-kickoff wall?
Every AVRA call is locked the moment it is made, with a tamper-proof timestamp that cannot be edited once the ball is in the air. The wall shows how many calls are frozen and how many have been scored against the final stats, so the record cannot be quietly cleaned up after the fact.
Does AVRA only show the results that worked?
No. AVRA also ran a Superflex lineup-construction gate and it did not survive the strict statistical check, so that result is marked provisional and not claimed. A tool that only shows its wins is hiding the rest of the table.
How is AVRA's trade evaluation accuracy measured?
It counts how often AVRA picks the better side of a real league trade, measured on the trades managers actually made and on what those players went on to score.
See what AVRA does for your own league.
AVRA makes the call on every lineup, trade, and waiver, in your scoring, with your roster, against your opponent. AVRA picks a side. The call is always yours to make.
Open beta access is free. Pro and All-Access are available now.