Fan Fury
A daily fantasy sports platform with live scoring — where thousands of contest entries settle correctly or the product dies on day one.

What they needed — in their own words
“Daily contests on real matches. Users draft teams, watch live scores move their rank, and get paid out the moment the match ends. Scoring has to survive a data-feed outage mid-match.”
The system they run on today
- Contest lobby, team drafting, and entry management
- Live scoring with rank updates during matches
- Automatic settlement and payouts at match end
- Score-feed failover with replay on recovery
Arrived, assessed, hardened
Every Neolen engagement follows the same path: an honest assessment of what's actually there, then the hardening work that makes it a system. Here is how it played out on this one.
What arrived
A contest platform whose live scoring consumed the provider feed directly: whatever arrived was the truth, and whatever didn't arrive never happened. No one had asked what a 40-second feed outage during a match would do to payouts.
What the assessment found
We replayed a historical match with a simulated mid-match feed outage. Scores diverged from the official result on recovery — which means wrong ranks, wrong payouts, and refund chaos on the platform's biggest night. A settlement-corrupting failure, invisible in every demo.
What we hardened and shipped
We rebuilt score ingestion as an event log with replay: on feed recovery the system reconciles the gap from the provider's authoritative record before any rank or payout is computed, and settlement runs only against verified final scores.
What the client runs on today
Fan Fury settles contests against verified final scores even through mid-match feed failures — the tail-risk scenario a demo never shows and a live real-money product cannot survive without.
Added to the playbook
The historical-outage replay test joined the audit harness for every product that consumes a live third-party feed.
Hit a wall with something you've built?
The audit is fixed-fee and standalone — a plain-language report of what's actually wrong, ranked by risk, yours whether or not you continue.