What this site is, in one paragraph
This is a stock and portfolio research tool. Give it a ticker symbol (like AAPL for Apple or NVDA for Nvidia) and it reads the company's financials, its recent price action, the news and some specialist market data, then hands you a plain-English read: Buy, Hold or Sell, with a confidence level, the reasoning behind it and the case against it. Give it a portfolio instead, or one of the sample books, and the same engine scores every holding and X-rays the whole book: concentration, how it moves with the market, and how far it sits from the risk level you chose. It is a second opinion, not financial advice, and it will refuse to tell you what you personally should do. The public demo covers 20 pre-run names and seven sample portfolios; a firm key opens the rest. Every number below is explained on this page.
What to do next: go to Analyzer in the sidebar, type a ticker or two, leave the mode on Balanced, and press Analyze. The result card uses the words defined below — hover the little “i” next to any term anywhere on the site for the same explanation in place.
The scores you'll see
Each one: what it measures · what high vs low means · where the data comes from · why it matters
Words on the page
Jargon translated to plain English
Buttons & modes
What each control actually does
New analysis
Enter tickers, choose depth, hit analyze
Presets
FundamentalsiIs the business healthy?
TechnicalsiWhat is the price doing?
News & sentimentiWhat are people saying?
Alt-data (Quiver)iInsider & institutional activity
Portfolio-aware sizingiSize vs. what you already own
Last run —
Results
No analysis yet
Enter tickers above and hit Analyze to get institutional-grade takes.
Backtestingi
Replays the tool's own weekly calls on past data · graded step-by-step through time
Average return per calli
—
run a backtest
Sharpe ratioi
—
return per unit of risk
Max drawdowni
—
worst drop, peak to trough
Win ratei
—
—
Equity curvei
Not yet connected to the backtest — no curve is drawn
Trade logi
Up to 20 calls, best return first
| Symbol | Entryi | Exiti | Heldi | Returni | Rec at entryi |
|---|---|---|---|---|---|
| Run a backtest to see trades | |||||
Sector Scani
Rank a sector's best-known names by RBWM score · find leaders and laggards
Constituents scannedi
—
select a sector
Sector sentimenti
—
—
Top RBWMi
—
—
Laggardi
—
—
Ranked universei
Sorted by RBWM score, best first
| #i | Symbol | Name | RBWMi | Fundi | Techi | Senti | Calli |
|---|
Portfolio
Your holdings, each scored by RBWM AI · a whole-portfolio X-ray below
Portfolio valuei
—
—
Positionsi
—
active holdings
Largesti
—
—
Unrealized P/Li
—
—
Holdingsi
Edit your holdings in the Advanced drawer · "Refresh" to score them
| Symbol | Name | Shares | Avg costi | Mkt valuei | Weighti | P/Li | Recommendationi |
|---|---|---|---|---|---|---|---|
| No holdings yet. Click "Edit holdings" to add positions, then hit Refresh. | |||||||
Portfolio X-ray i
The whole portfolio as one object: concentration, co-movement, sector lean, and a risk check against the level you choose
Your risk tolerancei
i
Try a sample portfolioi
Set your risk tolerance (what you want), then analyze. The X-ray measures what your portfolio actually is — its own risk level, what drives it, and whether it matches your tolerance. New here? Try a sample portfolio — Conservative, Balanced (55 stocks) or Aggressive — and see the X-ray instantly.
Crypto Marketi
Major coins · daily price, 24h change and volume via Polygon
AI market readi
AI read of the 24h snapshot — descriptive, not a trade signal
The AI market read appears here once the coin table has loaded.
Read the rest
Coins trackedi
—
—
Total 24h volumei
—
USD notional
BTC 24hi
—
Bitcoin
Top mover 24hi
—
—
Top coinsi
Daily crypto market stats via Polygon
| Coin | Pricei | 24h %i | 24h Volumei |
|---|---|---|---|
| Loading crypto market data… | |||
Watchlisti
Saved in this browser only · click "Analyze all" to score every symbol
i
| Symbol | Name | Lasti | RBWMi | Confi | Recommendationi |
|---|
Alertsi
Price, RBWM-score, recommendation, and confidence triggers · checked automatically every 5 min
Active rulesi
—
—
Triggered todayi
—
today (UTC)
Total triggersi
—
retained history
Most recenti
—
—
Recent triggersi
| Triggeredi | Symbol | Typei | Conditioni | Trigger valuei | Recommendationi |
|---|---|---|---|---|---|
| No triggers yet | |||||
Rulesi
| Symbol | Typei | Conditioni | Channeli | Last triggeredi | Statusi | |
|---|---|---|---|---|---|---|
| No rules yet · click "New alert" | ||||||
Deep Research
The full analysis, shown stage by stage — up to 60 days of news coverage from every publisher the feeds return, fundamentals, technicals, a devil’s-advocate audit and the written reasoning. Same engine and same run as the Analyzer; a name with a recent stored analysis returns in seconds.
Research output
Enter a symbol and hit "Start research"
—
No research yet
Results will appear here with the call, scores, news coverage, the written reasoning, the devil’s-advocate audit and a suggested position size
Training Data
News stories paired with what the stock actually did over the next 7 daysi
Labeled examplesi
—
—
Stories held, not yet pairedi
—
waiting to be paired
From our own published picksi
—
counted separately
Stocks coveredi
—
—
How the examples spliti
Grouped by what the stock did over the following 7 days
Where the examples came fromi
Most examples first · top 15
| Sourcei | Examplesi | Sharei |
|---|
Most-covered stocksi
Most examples first · top 15
| Symboli | Examplesi |
|---|
Source Credibility
Per-source win rate and per-sector edge. Sources whose calls have tended to match the next move get up-weighted; sources that consistently miss get down-weighted.
Sources trackedi
—
with ≥10 articles
Articles scoredi
—
from the news archive
Best win rate (high-confidence)i
—
≥30 stance samples
Best win rate (raw)i
—
≥10 stance samples
Worst win rate (high-confidence)i
—
≥30 stance samples
Sources rankedi
| Source | Articlesi | Avg 7d returni | Stance win ratei | Stance counti | Bull / Beari | Edgei |
|---|
* n/a: fewer than 30 graded directional calls, too few for a verdict. A dash in Stance win rate means the outlet made no gradeable bullish or bearish call in the archive.
Per-sector edge matrixi
Top edges per source × sector combo (min 5 articles, min 3 stance samples). Green = source beats random by 10%+; red = coverage was followed by the wrong direction more often than not.
| Source | Sector | Win Ratei | Avg 7d reti | Samplei |
|---|
AI interpretationi
AI read of the per-source forward track record — measured history, not a trading signal
Load the scorecard to get an AI read of which sources have actually called the direction right.
Deep Research — pick scorecard
Realized hit-rate of Deep Research's own BUY/SHORT calls (not news outlets). A "hit" = beat a benchmark of similar-size stocks in the same sector, in the called direction, over 7 days. Scored automatically from the app's own logged picks.
Picks trackedi
—
—
Resolvedi
—
7d horizon
BUY hit-ratei
—
—
SHORT hit-ratei
—
—
Median return vs sectori
—
BUY picks, 7d
How this is graded
Every scoring method is graded the same hard way: on cases it never saw while it was being built,
with a 7-day gap either side so nothing leaks across, and always as a range rather than one flattering number.
i
Updates on its own · read-onlyi
Method in use todayi
Nothing in use yet — the first method to clear the bar takes over.
Loading…
Promises made in advancei
Share that passed: — · still waiting: —i
Read-onlyi
| What was written downi | Measurei | Promisedi | Actuali | Resulti | Writteni |
|---|---|---|---|---|---|
| No experiments yet. | |||||
Cases in the recordi
—
in eval pool
Methods gradedi
—
in leaderboard
Attempts madei
—
deflation factor
Top methodi
—
—
Scoreboardi
Ordered by the pessimistic end of each accuracy range. Anything whose range reaches down to the coin-flip line has not been shown to beat a coin flip.
| Methodi | Cases gradedi | Accuracyi | Accuracy rangei | Luck checki | Test slicesi | Verdicti |
|---|---|---|---|---|---|---|
| Loading the scoreboard… | ||||||
Is the confidence honest?i
When a method says it is 60% confident, does it win about 60 times in 100? The top bar is the confidence claimed, the bottom bar is what actually happened. Pick a row in the scoreboard above to switch method.
No confidence check available yet.
Where a method worksi
The same cases, split by the conditions they were made in — which is where a method that looks fine overall usually falls apart.
| Market conditioni | Casesi | Accuracyi | Accuracy rangei |
|---|---|---|---|
| No breakdown available yet. | |||
Health checksi
The last 20 scheduled checks on whichever method was on top at the time. A check is flagged when the pessimistic end of that method’s accuracy range has fallen to the coin-flip line.
No regression log yet.
Track Recordi
Per-recommendation forward returns · shows whether the system works (or doesn't)
i
Total recsi
—
—
Win ratei
—
overall
Avg returni
—
per rec
Pendingi
—
too new to score
By recommendationi
| Recommendation | Ni | Winsi | Win ratei | Avg returni |
|---|---|---|---|---|
| No data yet | ||||
By modeli
| Modeli | Ni | Win ratei | Avg returni |
|---|---|---|---|
| — | |||
Recent recommendationsi
| When | Symbol | Reci | Confi | Price at calli | Returni | Resulti |
|---|---|---|---|---|---|---|
| No data yet — analyze some symbols, come back later. | ||||||
Feedback Loop
What the record has changed so far · every past call is graded 7 days later, and those results feed back into how much each kind of evidence counts and how confidence is statedi
Calls gradedi
—
—
Confidence checki
—
stated → realized
Strongest evidence typei
—
highest win rate
Top-weighted settingi
—
last 30 days
How much each kind of evidence counts
Where each one started, where the graded record has moved it since, and — unless you switch one of the three off, or type your own three numbers in the Advanced drawer — the split every new analysis actually usesi
| Evidence typei | Starting weighti | Weight nowi | Changei | Win ratei | Calls gradedi |
|---|---|---|---|---|---|
| Reading the graded record… | |||||
Does the confidence hold up?
What RBWM AI said it was sure of, set against how those calls actually turned out · where confidence was overstated it is trimmed on new calls, and where it was understated nothing is raised
| Confidence bandi | Calls gradedi | Saidi | Came out righti | Measuredi | What it changesi |
|---|---|---|---|---|---|
| Reading the graded record… | |||||
Scoring settings — last 30 days
Operator view · how much each internal scoring setting counts toward the combined call
| Settingi | Calls gradedi | Win ratei | Vote weighti |
|---|---|---|---|
| Reading the graded record… | |||
Signalsi
Every recommendation, broken into the individual signals behind it — each graded on its own
Registered signalsi
—
testable in isolation
Signals with edgei
—
verdict = EDGE
Best signali
—
—
Worst (anti-edge)i
—
contra signal
Per-signal edgei
Checked against graded calls in Track Record · at least 10 bullish or bearish calls before a verdict
| Signali | Pillari | Ni | Bull WRi | Bear WRi | Avg ret (bull)i | Avg ret (bear)i | Verdicti |
|---|---|---|---|---|---|---|---|
| Loading… | |||||||
Position Sizeri
Kelly-criterion sizing · tells you exact share counts to hit target weights
Uses the holdings in the Advanced drawer; the Edit holdings button above opens it.
Cash targeti
—
unallocated
Kelly fractioni
—
—
Max weight capi
—
per position
Positionsi
—
in portfolio
Sizing plani
| Symbol | Current sharesi | Current weighti | Target weighti | Target sharesi | Δ sharesi | Δ $i | Actioni |
|---|---|---|---|---|---|---|---|
| Add positions on the Portfolio page (Edit holdings), then click "Compute sizing". | |||||||
Tax-Loss Harvesti
Scans the holdings in the Advanced drawer, measured against the cost per share you entered, for losing positions worth selling for tax savings · 31-day wash-sale rule
% min lossi
Candidatesi
—
loss positions
Total lossi
—
unrealized
Tax savings (24%)i
—
est. middle bracket
Tax savings (37%)i
—
est. top bracket
Candidatesi
| Symbol | Shares | Avg costi | Pricei | Lossi | Loss %i | Wash-sale checki | Suggestioni |
|---|---|---|---|---|---|---|---|
| Click "Find candidates" to scan your portfolio. | |||||||
Earnings Calendari
Upcoming earnings releases · via Finnhub
Next 14 days
i—
| Date | Houri | Symbol | EPS esti | Rev esti |
|---|---|---|---|---|
| Click "Load" to fetch the calendar. | ||||
AI interpretationi
Enter a symbol and click Load for an AI read of its earnings setup.
—
Options Flowi
Put/call ratio + top open-interest contracts · live options-market data via Polygon
AI interpretationi
AI read of the put/call balance, open-interest concentration & implied volatility — crowd positioning, not a prediction
Pull flow for a symbol to get an AI read of the positioning.
Read the rest
Contractsi
—
—
Call OIi
—
open interest
Put OIi
—
open interest
Put/Call ratioi
—
—
Top by open interesti
| Typei | Strikei | Expirationi | OIi | IVi | Lasti |
|---|---|---|---|---|---|
| Pull flow for a symbol to see the top 5 contracts. | |||||
Settingsi
Appearance · API keys · notifications · model · loaded from the server config
Appearance
ThemeLight or dark
Modeli
Primary modeliTechnical model name (e.g. gpt-4o)
Temperaturei0.0 – 1.0
Max tokensi
API keysi leave masked to keep current value · paste a new key to replace
OpenAIi
Anthropic (Claude fallback)i
Financial Modeling Prepi
Polygon.ioi
Alpha Vantagei
Finnhubi
Quiveri
NewsAPIi
Whale Alerti
Notifications (Email)i
Alert recipienticomma-separated
SMTP hosti
SMTP porti
SMTP useri
SMTP passwordi
Behaviori
Disable Google Custom Searchi"true" or "false" — true switches the extra search off
Get the monthly research findings
Each month's picks and the reasoning behind them, plus the post-mortems when something we ran didn't work. No hype, and you can leave in one click.