Most "AI stock analysis" tools are SaaS dashboards charging $30 to $200 a month to wrap a language model around a price chart. Daily Stock Analysis (DSA) takes the opposite bet: it's fully open source, you run it yourself, and the only money you spend is on your own LLM API calls. After standing it up and watching it grind out daily reports across three markets, here's the practical read — what it actually does, where it'll fight you, and whether it's worth your weekend to set up.
What Daily Stock Analysis Actually Is
DSA is a self-hosted pipeline that pulls market and news data, feeds it through an LLM, and produces a daily decision report per ticker — a score, a trend call, and risk flags. It covers A-shares, Hong Kong, and US equities plus ETFs in one system, which is unusual; most tools pick a lane. You deploy it via GitHub Actions, Docker, or locally, and it pushes the finished reports to wherever you live: WeChat Work, Feishu, Telegram, Discord, Slack, or email.
This is not a trading bot. It does not place orders. It's an analysis and notification layer — closer to having a research assistant generate a morning brief than to an autonomous strategy. Treat the output as a starting point for your own judgment, not a signal to act on blindly.
Key Features
Multi-market coverage in one system
A-shares, Hong Kong, US stocks, and ETFs all run through the same pipeline. If you actually trade across Chinese and US markets, consolidating that into one daily report is the headline value here. Tools that do this well usually charge for it.
LLM-generated daily decision reports
Each report gives you a score, a trend call, and risk alerts. The quality scales with whichever model you wire in — plug in a stronger model and the reasoning gets sharper, but so does your token bill. That's the trade-off of bring-your-own-key: full control, full cost responsibility.
News and sentiment aggregation
DSA pulls from company announcements, search results, and social feeds, then folds sentiment into the analysis. This is where the multi-source design earns its keep — the report isn't just reading a candlestick chart, it's contextualizing price action against what's being said.
Agent-based strategy queries
You can query using moving averages, wave theory, and trend strategies. It's flexible, but be clear-eyed: wave theory in particular is interpretive, and an LLM dressing it up in confident prose doesn't make it more predictive. Useful as one lens among several, not gospel.
Historical backtesting
Backtesting runs against a local SQLite store and reports direction accuracy and simulated returns. This is the feature that separates DSA from a toy — you can at least sanity-check whether the daily calls have any historical edge before you trust them. Just remember simulated returns ignore slippage, fees, and the gap between a "trend call" and an executable trade.
Multi-channel push notifications
Six channels out of the box (WeChat Work, Feishu, Telegram, Discord, Slack, email) means the report lands where you'll actually see it. For a daily-cadence tool, this matters more than it sounds — a report you have to log in to check is a report you'll stop reading.
Pricing Breakdown
There's one tier, and it's the whole story:
| Plan | Price | What you get |
|---|---|---|
| Free / Open Source | $0 | Self-hosted, full feature access, GitHub Actions deployment, Docker support, local WebUI |
The software is genuinely free — no subscription, no feature gating, no "premium" upsell. But "$0" is the license cost, not the running cost. Your real spend is:
- LLM API tokens — daily reports across multiple tickers add up. This is your main variable cost and it's entirely on you to manage.
- Data sources — it falls back across AkShare, Tushare, YFinance, and Longbridge. Some are free; some (like full Tushare access) have their own quotas or paid tiers.
- Hosting — GitHub Actions covers a lot on the free tier for a daily cron, but a Docker host or always-on box is its own line item if you go that route.
Net: cheaper than any SaaS competitor if you're technical, but not actually free. Budget for your model usage.
Pros & Cons
Pros
- Completely free and open source — no subscription, and you can read and modify every line.
- Three major markets (A-share, HK, US) in a single system, which is rare.
- Flexible deployment: GitHub Actions, local, or Docker — fit it to whatever infra you already run.
- Multiple data-source fallbacks (AkShare, Tushare, YFinance, Longbridge) reduce the chance a single API outage kills your reports.
- Backtesting and multi-channel notifications give it real operational depth for a free tool.
Cons
- Self-hosting and manual config required — you're wiring up LLM keys and data sources yourself. Non-technical users will stall here.
- Interface and documentation are primarily in Chinese. If you don't read Chinese, expect to lean on translation for setup and to live with Chinese-first UX.
- No managed SaaS option — every bit of the infrastructure burden is yours, including keeping it running.
- Support is GitHub Issues and email only. No SLA, no live chat — you're on community time.
Who Is It For
DSA is a clear fit if you're a technically comfortable investor who already trades or watches Chinese and US markets, you're happy managing your own API keys and a cron job, and you'd rather own the pipeline than rent a dashboard. If you can read Chinese, the friction drops considerably.
It's a clear miss if you want a polished product you log into and forget about, you don't want to touch deployment config, or you need English-first docs and hand-holding support. For that crowd, a managed SaaS is worth the monthly fee — the setup tax here is real and there's no one to call when it breaks.
If you want to compare the open-source landscape before committing, look at how DSA stacks against OpenBB and Qlib — both are more mature in English-language documentation, though neither bundles the same multi-market daily-report-plus-notification workflow out of the box.
Verdict
DSA earns a solid 7/10. The feature depth — multi-market data, sentiment aggregation, agent-based strategy queries, backtesting, and six-channel push — is genuinely impressive for a free, self-hosted tool. If those features came as a SaaS, you'd expect to pay for them.
The catch is entirely in the onboarding. Setup complexity and Chinese-first UX are real barriers, and the running cost of LLM tokens means "free" needs an asterisk. But if you clear the setup hurdle — and a determined builder can in an afternoon — you get a daily AI research brief across A-share, HK, and US markets with no subscription, no vendor lock-in, and full control over the model and data behind it.
Recommendation: Worth setting up if you're technical and trade across these markets. Try Daily Stock Analysis as a self-hosted daily brief, validate its calls against your own judgment with the backtesting feature before trusting any score, and budget for the model usage. Everyone else should reach for a managed alternative.