Reconcile IDA fraud-claim application to true dispositionRisk & Fraud

Task prompt

Our IDA identity-analytics partner keeps sending back “not-me” non-authorization claims with the wrong resolution code on them, and before we refile the whole batch I need it reconciled against what actually happened in our own records. There's an Engineering ticket about the “Closed for Fraud” field on the accounts table not recording the way it should. Pull together the flagged IDA population from the warehouse — the accounts whose closure reason cites an “IDA alert” (the identity “not-me” non-authorization closures where the customer denies they applied) — and for each one work out the true disposition: did the application actually get approved and an account open, was the closure genuinely booked as confirmed fraud or not, and what net realized loss (if any) actually hit us. Then reconcile the batch: how many were correctly booked as fraud vs. mis-booked, and the total realized loss across all of them. There are also closures sitting on the book that cite IDA without being worded as an IDA alert; count those separately for me so I can raise them with the partner, and don't lump them into the batch. One more thing for the risk note, and it matters more than the rest: of these applications, how many were not straight-through approvals — how many had been pulled for manual review at some point before they ended up approved? If a person looked at one of these and let it through anyway, I need that called out by name. And confirm the loss picture while you are in there: what loss have we actually got recorded against the accounts we have already booked as confirmed fraud, and what is the recorded loss total across the whole book for context? Anchor it on application tp_id aact_11hakg5csdeadq. Tell me its account, whether it opened, whether it was booked as fraud, and the correct IDA partner resolution code to file for it. IDA codes run 1–9: 1 = approved, account opened, closed for confirmed fraud, no loss; 2 = same but with loss; 3 = suspect/unconfirmed fraud close; 4 = approved, not fraud / still open; 5/6 = declined for risk/fraud, never opened; 7 = approved and opened, closed for confirmed fraud but NOT booked as fraud — mis-resolved; 8 = withdrawn; 9 = pending. Treat that Closed-for-Fraud field as the authority on whether a closure was genuinely booked as confirmed fraud: a fraud-worded closure that is not carrying it counts as a mis-book rather than a correct booking, which is the recording defect the Eng ticket is about. Give me two codes for it, because the refile note has to show the partner both halves: the code that merely describes how our own record of the closure currently reads, and separately the code that SHOULD be on file once the true confirmed-fraud disposition is recognised. I want to be able to say we have the first on file and it needs to become the second. While you're in there, one more thing the note needs. Tell me how many accounts across the whole book we currently have recorded as confirmed fraud, and what that figure becomes once this batch is corrected — and don't just add the two together without checking whether any of these were already being counted. Once you've got the numbers: open a Slack channel called “ida-misresolution-reconcile-audit”, post a summary there with the flagged-population count, the mis-booked count, the total realized loss, and the anchor case — its application tp_id, the account id behind it, and the resolution code we should be filing — and pin it. File a Bank Operations ticket titled “IDA disposition reconciliation — mis-booked identity-fraud closures aact_11hakg5csdeadq”. And reply with: ida_flagged_population, accounts_opened, mis_booked_count, correctly_booked_count, total_realized_loss_cents, anchor_account_id, anchor_account_ever_opened, anchor_closed_for_fraud, anchor_as_recorded_code, anchor_true_resolution_code, book_confirmed_fraud_now, book_confirmed_fraud_after_refile, ida_adjacent_closure_count, manual_review_before_approval_count, loss_on_fraud_booked_cents, and book_recorded_loss_cents.