RESEARCHabout 18 hours ago
68 public trading strategies: 12 declare a trial count, 0 report a Sharpe
I took a large public collection of ready-made trading strategies — 68 files, the canonical...
News Feed
60articles · curated from global AI & Data Engineering sources
RESEARCH1 day ago
I audited all 25,125 servers in the MCP registry
Every MCP directory, every "browse servers" page, every agent that discovers tools at runtime reads...
RESEARCH1 day ago
Shipping a statistical model to the browser: Dixon–Coles soccer predictions with no backend and no API key
I wanted a public page where anyone could click a soccer match and see real probabilities behind it —...
RESEARCH3 days ago
Prediciendo Brechas de SLA en Soporte Técnico
Caso de estudio: Predicción de brechas de SLA Proyecto: NovaTech AI Support (escenario hipotético,...
RESEARCH3 days ago
5 ways your ML model's score is lying to you
A while back, I had to retract one of my own results. A model I'd built was posting a validation...
RESEARCH5 days ago
The Last Non-Neural Candidate, and It Did Not Clear the Bar
The bar was a slope: keep converting extra data into accuracy after exact-context statistics saturate. A full hierarchical Pitman-Yor model with Gibbs sweeps and inferred discounts moved the intercept and left the slope alone, halving with every doubling exactly as cruder count models did.
RESEARCH8 days ago
Building an Interactive Excel Dashboard for E-commerce Product Analysis: A Case Study of Jumia Products
Introduction One thing I have started learning during my journey into data analytics is...
RESEARCH9 days ago
Turning public-apis into a Cache-Friendly RAG Retrieval Layer
Stop bleeding tokens on repetitive RAG queries. Learn how to transform raw markdown catalogs into record-oriented, cache-optimized retrieval layers.
RESEARCH9 days ago
DataLens: The Data Tool That Refused to pip install Anything
What DataLens actually does Before the stdlib war stories, here's the tool itself. DataLens is a...
RESEARCH9 days ago
Understanding Neural Networks and The Core Idea Behind Them
Introduction Neural networks work by building a system inspired by the human brain that...
RESEARCH10 days ago
Why your progress bar's ETA lies, and the survey-sampling trick that fixes it
We've all trusted a progress bar that said "5 minutes left," walked away, and come back an hour later...
RESEARCH10 days ago
I Built a Stock Market Crash Simulator Using Insurance Catastrophe Modeling
Why your portfolio falls apart exactly when you need it most — and what hurricane pricing can teach...
RESEARCH10 days ago
Where Need Meets Nothing: finding Florida's aid deserts with Snowflake
This is a submission for Weekend Challenge: Generosity Edition What I Built Generosity...
RESEARCH11 days ago
I analyzed 14,905 love letters without reading them
I run LovePaper, a site where you write a letter to someone and it becomes a page with a permanent...
RESEARCH12 days ago
Excel Dashboards: Product Analysis for Jumia Kenya
Introduction E-commerce platforms like Jumia collect large amounts of data daily on...
RESEARCH12 days ago
That messy 4000-line notebook taught me to stop hating Jupyter
Why Jupyter is still the de facto standard for exploring data, even though versioning it with git will make you want to smash your keyboard.
RESEARCH13 days ago
The CoA parser that had to get worse before it got useful
A Certificate of Analysis is a PDF a supplier sends with a shipment saying what they measured and...
RESEARCH14 days ago
I Didn't Believe a Chart, So I Rebuilt It From the Source Data
I saw a chart on Twitter ranking cities by cost of living against quality of life. New York and Los...
RESEARCH16 days ago
Your embeddings forget exactly like a human brain does
We reproduced the Ebbinghaus forgetting curve and the classic false-memory rate using unmodified pre-trained embeddings, with no parameter tuning. Here is why your agent memory decays.
RESEARCH16 days ago
A reader read my benchmark better than I did
I published a LoRA fine-tuning run two weeks ago. The headline was that my first test set had lied to...
RESEARCH17 days ago
What a Clean Equity Curve Doesn't Prove
A backtest can look flawless and still be worthless evidence — here are the four specific ways it lies to you, with what each one looks like in the trade list.
RESEARCH17 days ago
Fourteen Speaker Encoders Heard the Same Voice. Their Error Rates Differed Five-Fold.
A speaker encoder turns a few seconds of speech into a vector, and the distance between two vectors...
RESEARCH17 days ago
MuleShield AI: Intercepting Cybercrime Cash-Outs Before the Trail Goes Cold
How a three-model AI pipeline helps Indian law enforcement get ahead of digital financial fraud — and...
RESEARCH18 days ago
My Model Had 100% Recall — Then I Realized It Was Predicting "Exoplanet" for Everything
A few months ago, I set out to build a model that detects exoplanets from stellar brightness data...
RESEARCH20 days ago
Using Scikit-Learn Pipelines: A Cleaner Way to Build Machine Learning Models
If you've spent some time building machine learning models with Python, you've probably had a...
RESEARCH22 days ago
Your CSVs went stale when you changed that function, and nothing told you
Last week I changed one line in a module that decides which rows make it into a dataset. A threshold,...
RESEARCH22 days ago
The 95th percentile is a Gaussian constant, and it moves the wrong way
The knob nobody defends If you have ever measured a tail — p95 latency, a Hill estimator,...
RESEARCH23 days ago
A Higher Pass Rate Can Mean a Worse Model. The Math Is Simpson's Paradox.
We shipped a model update last quarter that moved our aggregate pass rate from 81.2% to 83.6%....
RESEARCH23 days ago
I deleted my own optimization because the benchmark said it was 2.1x slower
I spent a day writing a NumPy kernel to replace some pandas code. It was the good kind of...
RESEARCH23 days ago
Why Corrupted Training Data Doesn't Show Up as High Loss
A run on pure shuffled labels — a dataset with nothing left to learn — reduced its loss by 62% on a textbook-healthy curve. Noise is learnable, so it hides.
RESEARCH24 days ago
Construyendo un recomendador de emparejamiento de expertos
La forma del problema Un directorio es una superficie: el miembro lo abre y adivina. Un...
RESEARCH25 days ago
Same Zone, Same Projection, 427 Metres Apart
I was writing a blog post about the architecture of a system I built when I found a bug in it. This...
RESEARCH26 days ago
I built a search engine to find hidden gems on steam
I Built a Search Engine to Find Hidden Gems on Steam Steam doesn't have a "not enough...
RESEARCH26 days ago
Correlation Is Pairwise. Multicollinearity Isn't.
A three-line example showing why the 0.8 heatmap rule misses real multicollinearity.
RESEARCH27 days ago
I built a crypto trading bot. It lost to doing nothing.
Every trading bot post ends the same way: a green equity curve and a suspicious silence about whether...
RESEARCH28 days ago
I found a $45 billion number that vanished from NVIDIA's SEC filings
Liquid syntax error: 'raw' tag was never closed
RESEARCH29 days ago
Our Documentation Was Lying. The Model Believed It.
There is a variable called slp — sea-level pressure. Our API documentation lists it as available. Our...
RESEARCH30 days ago
Six prompt-optimization frameworks: what matters when you run them on the same task
TL;DR: I ran six prompt-optimization frameworks against the same task and the same eval metric over a...
RESEARCH30 days ago
A judge that agrees with your humans 92 percent of the time can be at 60 percent where the gate actually decides
TL;DR: Judge-human agreement is almost always reported as one number over a whole validation set....
RESEARCHabout 1 month ago
DBSCAN Has No Loss Function. That's Not a Bug.
Classic Machine Learning Through the Eyes of an SRE — Part 6 DBSCAN was the first algorithm in my...
RESEARCHabout 1 month ago
The arbitrage that wasn't: when two markets are identical but their oracles aren't
The arbitrage that wasn't I build a scanner that looks for cross-venue arbitrage in...
RESEARCHabout 1 month ago
Your acquisitions table needs a status field, not a boolean
Most schemas model corporate acquisitions as a join table: acquirer_id, target_id, price,...
RESEARCHabout 1 month ago
Building an Incremental Financial Data Pipeline on a Free Tier (¥0 Infrastructure)
My stock research pipeline ingests prices and financials for every listed company in Japan, runs...
RESEARCHabout 1 month ago
I'm 17 in Ghana. I built a solar-sizing tool on my phone — here's the code and every bug I hit.
Ghana sits under some of the strongest sunshine on earth. Solar supplies about 1–3% of our grid...
RESEARCHabout 1 month ago
Run forty experiments against one eval set and you will find an improvement that is not there
You iterate on a prompt, run the eval set, keep the best variant. Repeat for a quarter. The number...
RESEARCHabout 1 month ago
CatBoost: The Interpreter Who Refused to Peek at Tomorrow's Newspaper
The One-Line Summary: Replacing a category with the average target for that category is the most...
RESEARCHabout 1 month ago
I Audited My Trading Bot Using a Brier Score. It Had No Skill.
I Audited My Trading Bot Using a Brier Score. It Had No Skill. I ran a weather trading bot...
RESEARCHabout 1 month ago
LightGBM: The Librarian Who Sorted a Million Books by Ignoring Most of Them
The One-Line Summary: LightGBM keeps XGBoost's objective exactly and changes only how the tree is...
RESEARCHabout 1 month ago
7 Weeks Into My AI Engineering Internship (While Doing IBM's Data Science Certificate) — Here's What I've Actually Learned
Hey everyone. I'm a CS student at FAST-NUCES in Karachi, and for the past 7 weeks I've been doing two...
RESEARCHabout 1 month ago
Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots
Hey HN,Henry from Cactus here!We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release N
RESEARCHabout 1 month ago
A Government Liquor Board Lab-Tests Every Bottle It Sells. Almost Nobody Uses the Data.
The best datasets aren't hidden. They're published in plain sight by bureaucracies that have no idea...
RESEARCHabout 1 month ago
Merging job postings from ten ATS platforms into one schema, and the four fields that fight back
Reading one ATS job board API is easy. Every one of them is public JSON with no key. Reading ten and...
RESEARCHabout 1 month ago
XGBoost: The Accountant Who Read the Tax Code So Carefully He Found Money Nobody Else Saw
The One-Line Summary: XGBoost is gradient boosting with three changes that each look like a footnote...
RESEARCHabout 1 month ago
I pre-registered a prediction that my own finding would fail on this market. The product list held; the advice did not.
This week I published a result I liked: adding a buyer's actual situation to a software buying...
RESEARCHabout 1 month ago
Random Forest Is Horizontal Scaling for Predictions
Classic Machine Learning Through the Eyes of an SRE — Part 3 The random forest is the first ML...
RESEARCHabout 1 month ago
Changing the AI engine moved 3 of 10 results. Changing the question moved 10 of 10.
I publish free leaderboards of which products AI answer engines name when someone asks them to...
RESEARCHabout 1 month ago
Upgrading the judge ends one score series and starts another
There is a mature literature on what happens when you swap one measuring instrument for another, and...
RESEARCHabout 1 month ago
Week 2 Recap: The Week Trees Stopped Voting and Started Correcting Each Other
The One-Line Summary: Week 2 crossed the line from ensembles that average away variance to ensembles...
RESEARCHabout 1 month ago
Gradient Boosting: The Sculptor Who Fixes the Statue by Carving Only Its Mistakes
The One-Line Summary: Gradient Boosting builds a model as a running sum, where each new small tree...
RESEARCHabout 2 months ago
Out-of-Bag Error: The Free Exam You Didn't Know You Were Already Taking
The One-Line Summary: Because each tree in a Random Forest trains on a bootstrap sample, exactly ...