A full research desk
on one ticker.
Thirteen stages. Four analysts read the same asset in parallel, argue about it, propose a trade, and then have it reviewed by three risk perspectives before a portfolio manager commits to a call. You read every stage.
What you put in
Three fields. No prompt engineering, no system messages, no describing what you want in the abstract.
Ticker
AAPL, BTCUSDT, GLD, EURUSD — whatever your broker or exchange uses.
As-of date (optional)
Leave it empty for the latest data, or pin a date to run the desk as it would have seen the asset then.
Analyst team
All four by default. Deselect any you do not want — the debate and every later stage re-derive from whatever made it in.
You pick the asset class
The pipeline is identical; what changes is the data it reads and the rules it judges against. Kawai never guesses the asset class from your symbol — you choose it, so a proxy ticker or an unusual pair still gets the right treatment.
Stocks
The base case the desk was written for. Fundamentals are real fundamentals — revenue and earnings trajectory, margins, leverage and liquidity, cash generation, earnings versus estimates, and insider transactions. Price history, sector context, crowd sentiment and the week's news fill in around it.
Crypto
There are no earnings, so the fundamentals desk reports on market structure instead: market cap and rank, supply dynamics, where liquidity concentrates, and where the asset sits relative to its all-time high. Every role is also told the norms it must judge against — 24/7 markets with no circuit breakers, drawdowns of 30–50% inside historical range, fragmented liquidity, and liquidation cascades that overshoot. A stop that would survive a stock's bad day gets swept by a routine wick here, and the desk is built to know that.
Commodities
No earnings, no balance sheets, no insiders — so "fundamentals" becomes macro and supply: real interest rates as the opportunity cost of holding a non-yielding metal, the dollar trend, inflation expectations, and the supply events that actually move these markets. Precious metals are read against real yields, energy and industrial metals against growth and inventories, and where you trade a proxy like GLD the tracking error and carry cost are called out rather than ignored.
Forex
A pair is the relative price of two economies' money, so every position here is long one side and short the other — there is no standalone bull case without naming the other leg. Rate differentials and their expected path lead; inflation differentials, trade balances, terms of trade and safe-haven flows follow. Central-bank intervention is treated as a real tail risk, and where price history comes from an exchange-traded proxy its volume is read as ETF flows rather than spot volume.
The 13 stages
Six waves. Stages inside a wave run at the same time; a wave waits for the one before it. Deselecting an analyst removes their stage and shortens the desk — the later debates still run, reading whatever is left.
The analysts read the asset
in parallelEach analyst fetches its own sources and writes its own report. They do not see each other's work — that is what makes the next stage a real debate.
candlestick_chart Market
Trend and key levels, momentum, the volatility regime, and whether volume confirms price. Indicator readings come from a verified computed block — the desk interprets the signals rather than re-deriving arithmetic from raw candles, and never quotes a number that contradicts it.
forum Social
Crowd mood quantified: bullish versus bearish counts, dominant themes, tone shifts, and — importantly — where crowd sentiment diverges from price. Hype cycles get treated as context, not confirmation.
newspaper News
The last week of headlines, split into asset-specific developments versus sector and macro themes, with scheduled catalysts called out — earnings dates, policy decisions, regulatory calls.
account_balance Fundamentals
For a stock: the financial picture. For everything else, the equivalent — market structure, macro drivers, or the two economies behind a pair. When a source genuinely has nothing for the asset, the report says so instead of padding.
The debate
bull and bear, in parallelBoth sides read all four analyst reports and build the strongest case they can for opposing conclusions. The bear has to argue the risks; the bull has to answer them.
trending_up Bull
Growth potential, competitive advantage, positive indicators — and an explicit rebuttal to the bear's points rather than ignoring them.
trending_down Bear
Overvaluation, leverage, competitive and regulatory threats, deteriorating numbers — and an explicit rebuttal to the bull.
The research manager rules on it
Reads both arguments and issues a rating on a fixed five-point scale — Buy Overweight Hold Underweight Sell — with the reasoning for the choice, not a summary of both sides.
The trader writes the proposal
Turns the rating into a transaction: direction, size, and entry and stop-loss stated as absolute price levels in the instrument's quote currency. Percentages and vague ranges are not an acceptable answer at this stage — a stop you cannot place is not a stop.
Three risk perspectives argue
in parallelEach reads the trader's proposal and the original analyst reports, then argues against the other two. They are instructed to debate each other directly, not to write three neutral memos.
Aggressive
Argues that fear-driven caution destroys more value than well-sized risk, and pushes for larger exposure where conviction is high.
Neutral
Weighs upside against risk on the evidence, concedes where each extreme is right, and corrects both.
Conservative
Argues for asset protection: what this sizing exposes you to, and where a more cautious alternative secures the same return.
Risk verdict, then the decision
two stages, sequentialThe risk manager weighs the three stances by the quality of their evidence — not by counting speakers — and returns a verdict (acceptable / acceptable with adjustments / unacceptable as sized), the concrete downside exposures, and the required changes to sizing, entry, stop or hedging.
The portfolio manager then commits: one of the five ratings, sized by how decisively the case won. The desk is told that a debate always contains conflict and that conflict alone is never a reason to sit on Hold.
One deliverable on top of thirteen reports
A final writer merges all thirteen into a single answer that leads with the decision card — action, confidence, entry and stop levels where the trader supplied them, key risks — then the reasoning behind it. Every individual report stays readable behind it, and the whole thing exports to PDF or DOCX.
What this is, and what it is not
A desk that cannot say no is not worth reading. These are the boundaries, stated plainly.
check_circle What it does
- —Reads live market data, news and sentiment for the asset you name, and shows its sources.
- —Runs the same sequence of specialists every time, so two runs are comparable.
- —Shows every intermediate report, including the ones that disagree with the final call.
- —Says when a source was unavailable instead of filling the gap with a plausible number.
- —Writes the whole run in your language while keeping tickers, prices and ratings untranslated.
cancel What it does not do
- —Place, modify or cancel any order. The desk produces analysis and nothing else.
- —Predict the future. Every stage is reasoning over retrieved data, not a forecast anyone can promise.
- —Replace a licensed data feed for execution-grade prices or audited financials.
- —Tell you the debate was unanimous. It is built to disagree with itself.
- —Count as investment advice. It is a research tool that shows its work.
Put a ticker through it
Pick the asset class, name the symbol, and read all thirteen stages — including whichever analyst disagrees with the call.