Most investors spend their research time reading. Earnings summaries, analyst upgrades, Reddit threads, CNBC segments. They read until they feel informed enough to act.
I do not read any of that.
Instead, every morning before markets open, a pipeline I built runs automatically, scans 1,835 tickers across stocks, crypto, and prediction markets, and produces a ranked list of signals with directions (BUY or SELL), conviction scores, and exit levels. By the time I look at it, the work is already done.
Today I am going to show you exactly how that pipeline works — what it checks, why it ignores analyst reports, and how you can build a scaled-down version using free tools and a few Python libraries.
The first design decision that matters: your pipeline should measure, not read.
Analyst reports are narratives. They are useful for understanding a business but nearly useless for timing. A stock can have 15 Buy ratings and fall 40% in a year. The narrative reflects what has already happened and what analysts are paid to believe. It is lagging and biased.
Price, volume, momentum, and macro data are measurements. They reflect what the market — millions of buyers and sellers with actual money on the line — believes right now.
My pipeline checks 40+ technical indicators for every ticker. The ones that do the most work:
How to build this yourself: Start with just two indicators — RSI and 52-week relative strength. Stocks with RSI under 40 that are outperforming their sector over the last year have historically been good mean-reversion candidates. You do not need 40 indicators to beat reading analyst reports. You need 2 good ones, consistently applied.
Individual stock signals are noise without macro context. A technically perfect BUY signal in a stock during a credit crunch is a trap.
My pipeline pulls real-time macro data from the FRED API (Federal Reserve Economic Data — free, updated daily) and checks:
When multiple macro signals are flashing warning signs simultaneously, the pipeline reduces conviction scores across the board — even for technically clean setups. A good stock in a bad macro environment is a coin flip, not an edge.
How to build this yourself: Register for a free FRED API key at fred.stlouisfed.org. With 10 lines of Python, you can pull the 10Y-2Y yield spread every morning and build a simple rule: if the spread is below -0.5%, flag all BUY signals as “macro headwind present.” This alone eliminates a common mistake — buying breakouts into deteriorating economic conditions.
Raw signals are not actionable. “RSI is oversold” is not a trade. A ranked, filtered list is.
My pipeline combines all indicator scores into a single conviction score from 0 to 100 for every ticker, then ranks them. The top of the BUY list is the strongest setup today. The top of the SELL list is the most technically vulnerable name.
The ranking is not static. The pipeline includes a self-correction layer called conviction adjustments (ADJs) — rules added over time based on what the data showed was working and what was not. Example: I found that SELL signals in a specific volume regime had a dramatically higher win rate than average. So I built an adjustment that boosts conviction for those setups automatically. Over time, 400+ of these adjustments have compounded into a system that is materially more accurate than the starting version.
The result: the system currently has 108 consecutive winning SELL trades. Not because of any single insight, but because hundreds of small corrections have compounded over months into something materially more accurate than where it started.
How to build this yourself: Start a spreadsheet. Every week, run your two indicators (RSI + relative strength). Note the signals. After 30 days, look back: which signals worked? Which did not? Find one pattern in the failures and add a filter rule. That is your first ADJ. Repeat monthly. In a year, your process will be unrecognizable from today’s version — and better.
Quick Prompt: The Daily Research Brief
Paste this into Claude each morning before you check prices. Fill in the macro data from FRED (takes 2 minutes):
I am reviewing my watchlist today. Here is current macro context: - 10Y-2Y yield spread: [X]% - VIX: [X] - HY credit spread: [X]% And here are 3 stocks I am watching: - [TICKER 1]: price $[X], RSI [X], up/down [X]% past month - [TICKER 2]: price $[X], RSI [X], up/down [X]% past month - [TICKER 3]: price $[X], RSI [X], up/down [X]% past month For each stock: 1. Does the macro environment support a position right now? Why or why not? 2. What is the technically strongest setup in this group? 3. What would change your mind? Keep answers under 4 sentences per stock. No hedging language.
This takes 5 minutes and replaces 45 minutes of reading. The macro context forces the analysis to be grounded. The “no hedging” instruction forces a recommendation, not a disclaimer-filled non-answer.
Start Today
You can replicate the principles with a spreadsheet and two free tools: Yahoo Finance (free price + RSI data) and the FRED API (free macro data at fred.stlouisfed.org). Start with 10 stocks, track two indicators, add one macro filter, build one ratchet exit rule. Run it for 30 days. Add one refinement. That is how every serious research process starts.
The 5-Minute Meeting Prep: The 5-minute meeting prep that makes you the most prepared person in the room. Three prompts that surface what actually matters before you walk in.