Market Brief AI vs Santiment: on-device FinBERT & HMM regimes vs social-volume dashboards
Santiment Sanbase focuses on cloud-crawled crypto social volume, GitHub developer commit counts, and on-chain token age metrics across web charts. Market Brief AI runs entirely on your iPhone or iPad, scoring public GDELT headlines locally with Core ML FinBERT, feeding that sentiment alongside BLS CPI and Fed tone into four Hidden Markov Models on Binance USDT pairs, and replaying closed-bar backtests against buy-and-hold—for educational research, not financial advice.
1. Local Core ML FinBERT news scoring vs cloud social-mention scraping
Santiment scrapes Telegram, Reddit, X, and forum chatter in the cloud to plot social dominance and crowd euphoria spikes alongside GitHub activity. While crowd chatter highlights retail attention, raw mention counts can surge on spam campaigns and still require a quantitative framework to translate into regime probabilities.
Market Brief AI takes an auditable, local-first route: it fetches asset headlines from the public GDELT news index and runs FinBERT—a finance-tuned transformer model—directly on your iPhone or iPad through Apple Core ML. You can even open the built-in analyser and score any custom sentence on demand with zero cloud AI transmission.
| Feature | Market Brief AI | Santiment (Sanbase) |
|---|---|---|
| Sentiment model & runtime | Finance-tuned FinBERT running locally on device via Apple Core ML | Cloud-processed social sentiment & mention-volume indices |
| Text source | Public GDELT global news index + interactive custom sentence scorer | Social media channels (Telegram, Reddit, X) & crypto forums |
| Coupling to regime math | Sentiment score feeds directly into on-device Hidden Markov Models | Visual chart overlays and Sanbase screener filters |
| Macro & natural context | Kronos (CPI), Zeus (S&P 500), Demeter (Fed), Poseidon (quakes), Aiolos (NOAA) | Primarily crypto-native social, dev-activity & on-chain metrics |
| Cloud AI data exposure | None—zero text or settings sent to cloud AI providers | All queries and watchlists processed on Santiment cloud servers |
2. Probabilistic HMM backtests vs indicator exploration
Observing that sentiment turned negative on a chart does not automatically prove whether stepping into cash beat passive holding after accounting for whipsaws. Market Brief AI connects context inputs to Metis (input-aware HMM), Hestia (duration-aware HMM), and Athena (factorial HMM), then tests the resulting closed-bar strategy bar by bar.
Every run plots simulated purple strategy equity over blue buy-and-hold for any start date and test capital ($100 to $1B), while Pro adds four drawdown guards (flash-crash, EMA crash, CUSUM drift, and hard stop) plus a rule-by-rule execution log.
| Capability | Market Brief AI | Santiment |
|---|---|---|
| Hidden Markov Model suite | 4 models: Hera (Free), Metis & Hestia (Plus), Athena (Pro) | Not included |
| Closed-bar buy-and-hold backtest | Native one-tap replay on every Binance USDT pair | Requires external SanAPI Python notebooks for formal backtests |
| Interactive custom text test | Type any headline into the on-device FinBERT analyser offline | Not available as an offline local model |
| Account & tracking policy | No login, no email, no analytics SDK, no ad network | Account registration required for full Sanbase workspaces |
| Scope & advisory boundary | Educational simulation only; no live trading or financial advice | Research data platform; no order execution |
3. Mobile affordability vs web research subscriptions
Santiment’s full real-time Sanbase tiers target dedicated web researchers at roughly mid-tier SaaS pricing (~$49+/month), with free accounts experiencing data delays or metric caps. Market Brief AI never time-delays public Binance candles or watermarks charts on its Free tier.
Upgrading to Market Brief AI Plus ($7.99/month or $79.99/year) or Pro ($12.99/month or $129.99/year) unlocks deeper mathematical models and execution transparency rather than removing artificial paywall delays.
| Analytical need | Better choice | Why |
|---|---|---|
| Tracking GitHub commit frequency or social-channel mention spikes | Santiment | Dedicated cloud crawlers for developer repos and social chatter |
| Scoring GDELT news with FinBERT privately on iPhone/iPad via Core ML | Market Brief AI | Bundled on-device finance language model with zero cloud AI calls |
| Feeding sentiment & CPI/Fed context into an input-aware HMM | Market Brief AI (Plus/Pro) | Metis and Athena integrate context directly into state estimation |
| Verifying closed-bar performance vs buy-and-hold on a commute | Market Brief AI | Fast native iOS/iPadOS simulation that works offline once synced |
| Keeping monthly research software under $13/month | Market Brief AI Pro | $12.99/month or $129.99/year with a 3-day free trial |
Does Market Brief AI send news headlines or my custom text to OpenAI or cloud servers?
No. Market Brief AI fetches public GDELT headlines and scores all text locally on your iPhone or iPad using a bundled Core ML FinBERT model.
Does Market Brief AI track GitHub commits or Telegram chat volume like Santiment?
No. Market Brief AI focuses on Binance USDT price returns, GDELT news sentiment, BLS CPI, Federal Reserve tone, S&P 500 context, USGS/NOAA reports, and seasonal cycles.
Which tier includes the input-aware Metis Hidden Markov Model?
Market Brief AI Plus ($7.99/month or $79.99/year, starting with a 7-day free trial) unlocks both Metis and Hestia, plus 10 saved forecast cards per asset.
Is Market Brief AI a crypto signal service?
No. It is strictly an educational strategy-simulation and regime-analysis tool; nothing it shows is financial advice or a recommendation to trade.
