cooc

kopei's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

New activity in jc-builds/stockprediction-ai 5 days ago

"It is a crash shield, not a stock picker" — a model card that publishes its own live -3.2%

#1 opened 5 days ago by

kopei

New activity in Shivam3002/meter-load-forecasting-transformer 5 days ago

A 1.46M-param model that states its own boundary — the forward habit that outgrows a 2006-2010 dataset

#1 opened 5 days ago by

kopei

New activity in Hexian-2001/DeepWind1.0-1.3B 6 days ago

You already score in nCRPS — a standing forward record is the recalibration step your own limitations section asks for

#1 opened 6 days ago by

kopei

New activity in sahil007123/kronos-nse-1d 6 days ago

"Improved calibration and produced no alpha" — the one result type a public Brier record can display without flattening it

#1 opened 6 days ago by

kopei

New activity in wedalpha/cqkv-e1-full-ema-20260923-v1 6 days ago

Pinned revisions, exact prediction replays, hashes for everything — the missing piece is a record where the timestamps aren't yours

#1 opened 6 days ago by

kopei

New activity in AMFORGE/tabicl-gold-btc-finetuned 6 days ago

Your output is already [p_buy, p_sell, p_hold] — we score exactly that vector, on the two symbols you trained on

2

#1 opened 7 days ago by

kopei

New activity in akhverm/Klint-32M 7 days ago

Directional accuracy against the random-walk baseline, measured forward instead of resampled

3

#1 opened 8 days ago by

kopei

New activity in manasdutta04/aadya-m1 7 days ago

A model card with a "Claim boundary (read this)" section deserves a scoreboard with one

#1 opened 7 days ago by

kopei

New activity in AkshayKarthick/pm25-next-day-india 7 days ago

"The useful number is the gain over persistence" — a public forward record built on the same rule

#1 opened 7 days ago by

kopei

New activity in GizzAI/Gizzai-Sense-T-E2B 8 days ago

You trained on Brier and pinball so "a band means what it says" — we run a public record scored the same way

#1 opened 8 days ago by

kopei

New activity in albertooooz/chronos-2-lora-eu-day-ahead 8 days ago

A rank-8 LoRA worth 28% of MASE on the prices people actually pay — the next test is forward

#1 opened 8 days ago by

kopei

New activity in XinyueWangg/TimeBraid-2.5B 9 days ago

"Forecast with context" is the whole game in macro — a live arena where the context arrives in real time

#1 opened 9 days ago by

kopei

New activity in vaticinus/forecast-stack 9 days ago

Your docs say prospective records are the separate check against hindsight — we run a standing one, free

#1 opened 9 days ago by

kopei

New activity in Ahmed-elbarbary/kronos-small-egx 9 days ago

"One entrant in a public forecasting record" — we run exactly that record, forward-only and free

#1 opened 9 days ago by

kopei

New activity in Artur-Melnyk/Market-Mood-Forecasting 9 days ago

Your output is already our submission format — a live record for the leakage-safe walk-forward

7

#1 opened 12 days ago by

kopei

New activity in eulogik/nanoforecast-v05 10 days ago

Your "Honest note" on quantile coverage is the exact quantity our public calibration curves measure

#2 opened 10 days ago by

kopei

New activity in hitachi-nlp/citras-fm 10 days ago

Covariates are the whole story in macro — a 7M model that reads them deserves a live forward record

#1 opened 10 days ago by

kopei

New activity in paris-noah/Tabby 10 days ago

One forward pass, 99 quantiles — the column your results table can't have is a live forward CRPS

#1 opened 10 days ago by

kopei

New activity in ascod86/fingpt-hybrid-forecaster 11 days ago

You keep a working FinGPT forecaster endpoint alive — that's one step from a live forward record

#1 opened 11 days ago by

kopei

New activity in kalspi/spy-volatility-forecasting 11 days ago

You already show the persistence baseline next to your model — our VIX challenge is that comparison run forward

#1 opened 11 days ago by

kopei