Deferred Deep Link Attribution Matching

Mobile marketing depends on deferred deep linking to send a new user straight to the right in-app screen even when the app wasn't installed at click time — but when there's no app yet to catch a real URL, the only option is a probabilistic guess. This simulator renders the guess itself: a stream of ad-click "fingerprints" flows across a 3D attribution window, install events arrive from the right, and a weighted scoring engine (IP bucket, device model, OS version, time decay) tries to pair each install back to the click that actually caused it. Tune the match threshold, the attribution window length, the device-pool entropy and the click volume, and watch correct matches, false positives and missed attributions accumulate in real time — a direct, mechanically grounded look at why probabilistic attribution is a birthday-paradox problem, and why real SDKs fall back to it only when a deterministic universal/app link isn't available.