My crypto backtest had a sharpe of 13.4. Then I added trading costs
I was not trying to prove a grand theory about crypto. I was testing simple, falsifiable baselines and looking for one result worth a second experim…
AI & ML
⚑ Report a ProblemLatest AI & ML news from Tech News
All topics
AI
News
Tech
agents
ai
api
architecture
automation
backend
beginners
career
claude
database
devchallenge
devops
javascript
llm
machinelearning
mcp
opensource
productivity
programming
python
react
security
showdev
testing
tutorial
typescript
webdev
그래픽카드 세대별 로컬 LLM 속도와, 우리 프로덕션이 매일 처리하는 토큰량
모델명·서빙 설정·prefill/decode 실측치까지 전부 공개합니다 지난 글 (새 창)에서 그래픽카드를 5090으로 바꾼 얘기를 했습니다. 이번엔 훨씬 더 파고들어서, 실제로 어떤 모델을 어떤 설정으로 돌리고 있는지, 하드웨어별 prefill/decode 속도가…
Local LLM Speed by GPU Generation, and Our Production's Daily Token Volume
Model names, serving configs, and full prefill/decode measurements included In my last post (new tab) I talked about switching to the RTX 5090. This…
"[260817~0823] 발행 인프라를 새로 짜고, 조용히 사라지는 신호들을 잡았다"
블로그 발행 방식을 바꾼 김에, 표본에 속지 않는 판단과 소리 없이 죽는 파이프라인을 잡아내는 계측을 함께 정비했습니다 이번 주는 토요일과 일요일, 이틀 동안 작업이 몰렸습니다. 그 전 며칠은 개발일지가 조용했는데, 알고 보니 원인 자체가 이번 주의 첫 번째 주제였…
How I built an FVG trading bot for OKX and made 99% of its signals useless on purpose
How I built an FVG trading bot for OKX and made 99% of its signals useless on purpose If you trade crypto futures, you know the drill. You're starin…
"[Jul 27-29] A Silent Crash, a Missing Record, and 27 Hours of Signal Silence"
Three days spent chasing incidents where "nothing happened" rather than "money went out wrong" — a silent Monday-morning crash, a performance-record…
[260710] '조용히' 잘못되고 있던 것들을 하루에 세 개 잡았다
생성이 끝없이 폭주하던 문제부터, 아침에 고친 게 저녁에 조용히 속도를 반토막 낸 것까지 주말을 끼고 지난 며칠 사이 발견된 것들을 정리하고 마무리한 하루였습니다. 공통점이 하나 있었는데, 셋 다 "겉으로는 티가 안 나게 조용히" 잘못되고 있었다는 겁니다. 사고 1…
[Jul 10] Caught Three Things That Were Going Wrong Quietly
From a generation that spiraled out of control forever, to a morning fix that quietly halved throughput by evening This is the English version of a…
[Aug 17-22] We Swapped GPUs, Then Kept Finding Blind Spots in Our Own Checks
A hardware verdict flipped twice in one week, two verification tools turned out to be quietly missing errors, and a real brokerage account needed tw…
[260817~0822] 그래픽카드를 바꾼 뒤, 검증 도구들의 허점을 연달아 잡았다
하드웨어 교체 판정이 한 주 동안 두 번 뒤집혔고, 자동 검증 도구 두 개가 조용히 오류를 놓치고 있었다는 게 드러났고, 실계좌에서는 사람이 두 번 직접 개입해야 했습니다 지난 개발일지 이후로 조금 뜸했는데, 그 사이가 오히려 가장 바빴던 한 주였습니다. 월요일엔…
FlashAlpha vs Bloomberg Terminal 2026 - Options Analytics for Quants
Originally published at flashalpha.com . If you are comparing these two, you are probably not choosing between them. Most desks that run FlashAlpha…
Point-in-Time Fundamentals for Numerai Signals: Killing Lookahead in Your Feature Join
If you build features for Numerai Signals from fundamentals, the single most common way to silently overstate your live performance is joining on th…
I rebuilt a viral "3,888% return" TradingView strategy — it landed almost exactly at coin-flip odds
A trading YouTube channel ( Trading with DaviddTech ) posted a video claiming a "strange 63% win rate" TradingView indicator, turned into a full str…
How to Build a Point-in-Time Fundamentals Database from SEC EDGAR (and When Not To)
If you backtest with today's fundamentals, you are cheating yourself. The Revenue figure you pull for Q4 2019 was probably restated in 2021, and the…
Predictive Alpha: Pipeline Engineering for Real-Time Machine Learning Inference
Most retail algorithmic trading bots rely heavily on legacy technical analysis indicators—think RSI, MACD, or Bollinger Bands. While these indicator…
Defensive Algo Design: Error Handling, Backtesting, and Mitigating Simulated Slippage
Every quant developer knows the feeling: you write an algorithmic strategy, run it against a basic backtesting script, and the equity curve looks li…
Trading fee calculator that shows costs in R (units of risk)
Most trading fee calculators show you two numbers: the dollar amount and the percentage of notional. Both are correct. Neither is useful. Here's the…