J.P. Morgan AI Research | First author, 'Reinforcement Learning with Hawkes-Structured Networks for Market Making' (ICAIF '26 Paper 696)
Researcher at J.P. Morgan AI Research and first author of ICAIF '26 Paper 696, 'Reinforcement Learning with Hawkes-Structured Networks for Market Making', written with Philipp Plank and Yufei Zhang (Imperial College London). The paper sits in Oral Session 2, 'Market Making and Optimal Execution' (Monday 16 November, 11:40-13:00 CET, Room 1). Record covers the published paper credit only — no trading P&L, model performance or desk figure is claimed.
Apenas nós públicos listados.
1 entities geo-pinned · toggle People / Institutions · 51.5074, -0.1278
Two paid intents on this dossier. Both settle in USDC over x402 before the work starts, so nobody carries a receivable.
🔭 WATCH: compute contracts signed · open-weight versus API split · talent retention after the first raise
🔍 Fernando Acero is carried at HIGH confidence on 2 filed sources. The horizon below is a hypothesis with a shelf life, not advice.
Paste the badge on a site, a README or an agent page. It serves live from the edge and links back with its own attribution code.
Machine form: GET /badge/fernando-acero.json
Paste a Base wallet and this record joins your watchlist. Nothing is uploaded, nothing is stored off-chain beyond the list, and removing it is one click.
Paste the badge on a post, a deck or a partner page. The image links home with ?ref=badge-fernando-acero, so every embed is attributed in the referral ledger and every reader lands on this record.
JSON twin: /badge/fernando-acero.json returns the same snippet for a generator.
Paste the badge on a post, a deck or a partner page. The image links home with ?ref=badge-fernando-acero, so every embed is attributed in the referral ledger and every reader lands on this record.
JSON twin: /badge/fernando-acero.json returns the same snippet for a generator.
Paste the badge on your own page. The image is edge-cached SVG, the link carries the attribution code, and the counter moves in the referral ledger.
The record is what it is: J.P. Morgan AI Research | First author, 'Reinforcement Learning with Hawkes-Structured Networks for Market Making' (ICAIF '26 Paper 696). Nothing below is filed — it is the shape the seat would take if the current trajectory holds, and it is re-scored every time a new public source lands.
🔭 compute contracts signedExpect the mandate to widen before it deepens. Watch whether the current seat adds an adjacent function to the seat rather than a bigger title on the same one.
🔭 open-weight versus API splitA second mandate is the normal move here — a board seat, a fund-side advisory role, or an operating chair next to the current one. The tell is a first public co-signature, not an announcement.
🔭 talent retention after the first raiseFirst capital event window. Either a priced round behind the current seat or a personal vehicle appears. Watch talent retention after the first raise rather than the headline valuation.
🔭 model-to-revenue ratioJurisdictional switch is the highest-probability structural change from London, United Kingdom. A second residence, a second entity, or a second currency book (EUR to a settlement unit) is the observable.
🔭 enterprise logos versus pilotsInstitutional shell forms: a holding vehicle, a family-office wrapper or a named co-investment club. Shells are quiet — the disclosure arrives through registry filings, not posts.
🔭 compute contracts signedDistribution becomes the constraint. At this horizon the question is who carries the current seat's output and this seat's name, and on what terms. Watch model-to-revenue ratio.
🔭 open-weight versus API splitSuccession pressure. Either a named deputy appears under the seat or the seat is re-scoped away from day-to-day. Both read as the same signal from outside.
🔭 talent retention after the first raiseSecond-order liquidity: secondary sales, a tender, or a strategic minority into the current seat. This is where an OSINT file usually goes stale for six months and then re-prices.
🔭 model-to-revenue ratioRecord consolidation. By here the file should be readable as a single arc rather than a list of events, and the London, United Kingdom base stops being the interesting part.
🔭 enterprise logos versus pilotsEnd-of-horizon check. Re-read 2026 against 2036: the useful output is which of the eleven rows actually printed, and which were noise carried by the market rather than the person.
🔭 compute contracts signed🔍 Modelo de horizonte deterministico: cada linha deriva de factos ja registados e e reavaliada a cada nova fonte publica. Nenhum valor de previsao e afirmado.
Compra sem navegador via MCP tools/call fetch_profile. Repeticoes 402 sao idempotentes.
Pacote maquina: /api/um-radar/agent-pack?slug=fernando-aceroPrecos por lote e dataset a pedido
The dossier prose stays public. The space around it is for sale: locales, frames, video, links, the city slot, the verified badge. Holding $UM cuts the price at quote time - the balance is read from your wallet, nothing is self-declared.
Apenas radar OSINT público. Não é conselho de investimento.
A receita de taxas é onchain e verificável.