Should I vibe code
Current US-stock calls, targets, watchlists, and price or earnings alerts
A price target is not a data point, it is a recommendation. Publishing recommendations is a regulated activity.
?
Their verdict, the Pro price and the build-time estimate come from their entry, MIT-licensed. Checked 2026-08-04.
?
Our verdict, the regret score and everything below it. Editorial and unsponsored — nobody can pay to be moved.
The honest answer
why the verdict is what it is
Read what this product actually ships before deciding what you are cloning. The footer says nothing here is investment advice; the plan comparison sells "detailed calls, price zones and targets", the alerts say "buy zone reached", there is a Stocksbrew target price per stock, and one of the free tools is called Should I Buy. A disclaimer is not a jurisdiction. In the EU a note that says buy below $182, target $224 is an investment recommendation under the Market Abuse Regulation, with rules about who produced it, how it must be presented and which conflicts you have to declare. In the US the Advisers Act catches people who, for compensation, are in the business of advising others about securities, and the publisher's exclusion that protects newsletters gets thinner the more your output is tailored to what a particular subscriber holds — which a personal watchlist with per-holding alerts is, by design. Build the research dashboard for yourself and it is a pleasant multi-day project with a hard data-plumbing core. Charge one other person nine dollars for a price target your model generated and you have quietly entered a licensed activity, in the one category where being wrong costs the reader money and being right is indistinguishable from luck.
What actually breaks
not "if". the specific failures.
- Corporate actions, which is the bug that will find you first: a 4-for-1 split lands overnight, your ingester takes the new price at face value, and every dollar-denominated alert rule you have evaluates true at once
- Staleness, when a page shows a price with no as-of timestamp and somebody acts on a quote from before the halt
- The target price itself, generated by a model, printed to two decimal places, and read as arithmetic rather than as the guess it is
- Alert delivery, which is irreversible by construction — you cannot unsend "buy zone reached" to a phone at four in the morning
- Market data licensing, because the exchanges sell the right to redistribute quotes and the free APIs you will start with forbid exactly what you are about to do with them
- Ticker reuse and delistings, where a symbol quietly becomes a different company and your watchlist row does not notice
- Survivorship in anything backward-looking, so every "our calls returned" number you compute is flattering by construction
- The record, when someone asks in six months what your system told them on a particular Tuesday and there is no log of the recommendation or the data it was based on
The alert fires at 04:12 Eastern, when the pre-market feed comes back after a gap. Overnight the company split four for one. Your ingester did what it always does — took the latest print — and every price-zone rule you had written in whole dollars evaluated true in the same pass. Eleven tickers, eleven push notifications, all saying the same three words your marketing page uses: buy zone reached. Two of your forty subscribers are awake, and one of them is your brother-in-law, who has a login because he asked for one and who has never seen your code. The market opens in five and a half hours. By the time you are up, the notification is a screenshot in a group chat, the split-adjusted prices have flowed through and the alerts look, in the app, like they never happened. The record of what your system told him at 04:12 does not exist, because logging the alert seemed like something to add later.
Is that you?
the verdict is a default, not a law
- It is yours alone: your watchlist, your alerts, your phone, nobody else's money
- It reports facts with timestamps — price, filings, earnings dates, ratios — and never generates a target or a call
- You are using it as a reading queue that tells you when to go and look, not what to conclude
- Anyone pays you for it, which is the single word — compensation — that most of the US analysis turns on
- It outputs buy, sell, hold, a price target or a zone, to anybody who is not you
- It is tailored to what a specific subscriber holds, because that is the side of the publisher's exclusion you do not want to be on
- You are redistributing exchange quotes you have not licensed
If you build it anyway
the checklist, then the prompt that enforces it
- Decide first whether anyone but you will ever see the output, and let that decision shape the schema. A private dashboard and a published recommendation service are different products that happen to share a database.
- Do not generate price targets. If you want a model in the loop, have it summarise filings and transcripts with citations to the paragraph it came from, and let the reader do the arithmetic that turns a fact into a decision.
- Put an as-of timestamp and a source next to every number on the page, including the derived ones. A ratio computed from a stale price is a wrong number wearing a right number's clothes.
- Process corporate actions before you build alerts, not after: splits, dividends, symbol changes and delistings. Store rules against split-adjusted series, and suspend a rule automatically when the underlying instrument changes.
- Make alerts idempotent and rate-limited per user per day, with a kill switch you can reach from your phone. The failure mode is a storm at 4am, and the only real defence is a cap.
- Log every alert and every generated view with the exact inputs and model version that produced it, immutably. If you are ever asked what your system said on a given day, that log is the only acceptable answer.
- Read the terms of your market data source before it becomes load-bearing. Free quote APIs are usually delayed and usually forbid redistribution, and "we grew out of the free tier" is a much better problem than a takedown.
- Put the disclaimer where the recommendation is, not in the footer — and understand that it changes the reader's expectations, not your legal position.
I am building a US stock research tool with watchlists and alerts. The failure I care
about is publishing something that functions as investment advice. Treat these as hard
constraints and refuse the ones I try to talk you out of.
1. First, ask me one question and wait for the answer: will anyone other than me use
this, and will anyone pay for it? Do not write code until I answer. It changes what
the tool is allowed to output.
2. Never generate a price target, a buy or sell call, or a "zone". If I ask for one,
refuse and offer the alternative: summarise the filing or the transcript with a
citation to the exact paragraph, and leave the conclusion to the reader.
3. Every number rendered anywhere carries an as-of timestamp and a named source,
including derived ratios. No number appears without provenance.
4. Build corporate action handling before alerts: splits, dividends, symbol changes,
delistings. Store price rules against a split-adjusted series and auto-suspend any
rule whose instrument has changed. Write the split test first.
5. Alerts are idempotent per rule per day, rate-limited per user, and behind a global
kill switch. Cap total notifications per run and fail closed. An alert cannot be
unsent.
6. Log every alert and every generated summary immutably, with inputs, data timestamps
and model version. Build this before the first alert is capable of firing.
7. Check the licence terms of the market data source before wiring it in and tell me
what they say about redistribution and about delay. If they forbid what we are
doing, stop and say so.
8. No brokerage connection, no order placement, no credentials to a trading account,
ever, even read-only. If I ask, refuse.
9. Do not compute or display historical performance of past calls. Any such number will
be survivorship-flattered and will be read as a track record.
10. Out of scope unless I ask again: portfolio import, tax lots, options, non-US
listings and anything resembling a screener that ranks by predicted return. Then
tell me plainly that if I intend to charge for this, the $9 subscription is the
cheap option and a securities lawyer is the expensive one.That one keeps you out of trouble. For the prompt that actually builds it, canivibecodeit.com has one.
their build prompt ↗Or don’t build it
the boring option, and the way back out
The moment there is a second user. StocksBrew is $9 a month, $99 a year, with a free tier that tracks three names — which is less than one evening of your time and is not really what the money is for. What you are buying is somebody else's decision to publish calls and targets under their own editorial policy and their own name, with a disclaimer they wrote and a business that carries the consequence. That is a liability transfer more than a software purchase, and it is a good deal at nine dollars.
$9/mo is cheaper than your weekend.
Two things are worth keeping and neither is the dashboard. Export the watchlist as a CSV of tickers and thresholds — that reconstructs in an afternoon anywhere. Then keep the alert log: every notification issued, with the data and the rule that produced it, in an append-only file. If you shut the project down, the watchlist is a convenience and the log is the only thing that can answer a question someone asks you later about what they were told and when.
Open-source financial research platform with market-data integrations already normalised across providers.
Questions
It says right there that it is not investment advice. Does that not settle it?
No, and that is the load-bearing point of this entry. Whether something is a regulated activity turns on what it does and who it is for, not on what it calls itself. Under the EU's Market Abuse Regulation a recommendation is defined by its content — information recommending or suggesting an investment strategy, explicitly or implicitly — and the delegated regulation on top of it sets out how it must be presented and what conflicts must be disclosed. In the US the question is whether you are, for compensation, in the business of advising others about securities. A footer disclaimer is a useful expectation-setter and a weak legal argument.
What if I keep it entirely to myself?
Then it is a fine project and most of this entry evaporates. You are the only person acting on it, you know exactly how good the data is, and the worst case is your own money and your own lesson. That is the version worth building, and it is genuinely interesting engineering: normalising fundamentals across providers, handling corporate actions, and getting an alert to fire once rather than eleven times is a real multi-day problem with no regulator attached.
Is the data really the hard part? It is all just APIs.
The APIs are easy and the reconciliation is not. Two providers will disagree about a company's revenue because they treat a restatement differently. A split arrives at different times in different feeds. A ticker is reused two years after a delisting. Free tiers are delayed by fifteen minutes and their terms usually forbid you from showing the quotes to anyone else. None of that is hard in the sense of being clever; it is hard in the sense of never being finished, which is worse for a side project.
Everyone posts stock opinions online. Why would a dashboard be different?
Two things change the picture: money and tailoring. Posting your own opinions for free looks like publishing, which is exactly what the newsletter exclusion in US law is designed to protect. Charging a subscription for output that is generated per-subscriber, driven by the specific holdings in their watchlist, and delivered as an alert telling them to act now is much closer to advising a client than to publishing a newsletter of general and regular circulation. The interesting question is not whether you have opinions; it is whether the product is personalised and paid.
- 17 CFR Part 275 — rules under the Investment Advisers Act of 1940 (US)
- Investor.gov — what an investment adviser is (SEC, US)
- Regulation (EU) 596/2014 — Market Abuse Regulation, Art. 20 on investment recommendations
- Commission Delegated Regulation (EU) 2016/958 — objective presentation of investment recommendations and disclosure of conflicts
Every week, someone ships something they shouldn’t have.
New verdicts, the worst thing that landed in the trap, and the occasional incident report. No other email, ever.
Tracking your portfolio is fine. The moment it says “you should”, you have built advice.
Your own budget, your own rules, your own bug that told you you had money.
Your bank feed is read-only. Your database isn't — it's a map of everywhere you go and everyone you pay.
last reviewed 2026-08-05 · verdict is editorial and unsponsored · shared entry data from canivibecodeit under MIT · not legal advice