Quant
Ask the market anything. Get the receipts with it.
Quant reads price, volume, structure and the calendar in a single pass, then answers in plain language and shows every bar it used. It will never tell you what to buy — it will tell you exactly what happened, how often, and where to check.
Fifteen questions a month on the free plan. No card, no expiry.
NVDA beat on both lines last night and the tape sold it anyway. What changed in the structure?
The print was strong — 1.53 against 1.42 consensus, a 7.7% beat, on revenue of 57.1B against 54.6B expected. The reaction was not. NVDA is -2.90% at 188.60 on 177.3M shares, and that is the lowest close in the thirty-session window.
Structure has been rolling since the 208.83 close: only 14 of the last 29 sessions finished higher, and today puts price 9.7% under that high on the heaviest volume of the run. Good news being sold is a distribution signature, not a momentum one.
Cited: NVDA daily bars, 30 sessions to 27 Aug 2026 · Q2 report, 26 Aug after close.
That describes what happened. It is not a forecast and not advice.
Screen the rest of megacap semis for the same shape — beat, then a close at the low of the range.
Illustrative conversation. Nothing here is a recommendation.
Method
Three passes stand between your question and the answer on screen. Nothing is generated from the model’s memory of what markets usually do — every claim is measured against the bars in front of it.
It starts from the chart you are already looking at
Quant inherits your context: the symbol, the timeframe, the visible range and every indicator sitting on the pane. It then pulls the thirty prior sessions, the traded volume behind them and any calendar event inside the window.
- Chart context, not a blank prompt box
- Thirty sessions of history loaded by default
- Earnings, macro releases and listings inside the range
It tests the claim before it writes the sentence
Every condition Quant is about to describe gets evaluated across the loaded history first, and the hit rate comes back with it. A statement that does not survive its own count never reaches the answer.
- Conditions evaluated bar by bar over the window
- Confidence expressed as a frequency, never as a mood
- Evidence that cuts the other way is reported, not dropped
It shows the bars it used to get there
Each figure in an answer is a link. Click it and the chart jumps to the exact bar range behind the number with the condition highlighted, so you can disagree with the reasoning instead of trusting the tone.
- Bar-range citation attached to every number
- One click to the highlighted range on the chart
- One more click to a full backtest of the same rule
What you can hand it
Quant is not a chat window bolted onto a chart. Each of these runs against the same data layer the screeners, backtester and alerts read, which is why an answer can turn into a running rule without leaving the page.
Natural-language screening
Describe the setup in a sentence and Quant compiles it into a real screener query across stocks, ETFs, crypto, forex and commodities. The generated filter stays editable, so you can tighten it by hand afterwards.
Structure detection
Swing highs and lows, ranges, breaks of structure, failed breakouts and volume shelves are labelled on the series rather than eyeballed, which is what lets the answer talk about the shape of a move instead of just its size.
Calendar awareness
Earnings, IPOs and macro prints inside the loaded window are part of the context. Quant knows when a move happened the night of a report, and it will say so rather than attributing it to the indicator you asked about.
Citation of bars
No figure appears without the range it came from. Hover a number for the bar dates, click it to highlight the range, and export the citation list with the answer when you send it to somebody else.
One-click backtest handoff
Any answer that describes a repeatable condition carries a “test this” action. It writes the rule set, opens the backtester with commission and slippage already modelled, and leaves the parameters in your hands.
One context, every asset class
Ask about a stock, then about the currency it reports in, then about the commodity that drives its input cost. The session keeps the thread, and cross-asset comparisons resolve against the same bar clock.
Every answer has an exit into code
When a Quant answer describes something repeatable, it writes the rule out for you. Take it to the backtester, or take it off the platform entirely — AlgoBeamTS runs the same logic anywhere JavaScript runs, against the same bars.
- Answers export to AlgoBeamScript or plain TypeScript
- Identical series semantics on the platform and in your runtime
- Point it at any provider — exchange feed, REST endpoint or a local CSV
import { AlgoBeamTS, Provider, indicators } from 'algobeam-ts'
// provider, symbol, timeframe, how many bars of history to load
const algobeam = new AlgoBeamTS(Provider.Binance, 'BTCUSDT', '1h', 100)
async function main() {
const bars = await algobeam.load()
const last = bars.at(-1)
if (!last) throw new Error('provider returned no bars')
const rsi = await algobeam.run(indicators.rsi({ length: 14 }))
const ema = await algobeam.run(indicators.ema({ length: 50, source: 'close' }))
console.log('bars loaded ', bars.length)
console.log('last close ', last.close.toFixed(2))
console.log('rsi(14) ', rsi.last.toFixed(2))
console.log('ema(50) ', ema.last.toFixed(2))
console.log('regime ', rsi.last > 55 && ema.rising ? 'expansion' : 'neutral')
}
main().catch((error: unknown) => {
console.error('[algobeam] run failed', error)
process.exitCode = 1
})What Quant will not do
An analyst that hedges everything is useless, and one that promises anything is dangerous. Here is the line we drew, stated plainly enough to hold us to it.
It does not predict prices
Quant describes what the loaded history did and how often a condition held. It has no view on tomorrow, and any answer that reads like a forecast is a bug we want reported.
It does not give advice
No buy, sell, hold, position size or allocation. It will tell you what a rule did over ten years of bars; deciding whether to trade it is entirely yours.
It does not guarantee results
Every performance figure on this platform is computed on historical data with modelled costs. Past behaviour carries inherent limitations and never promises future performance.
It does not bluff
A question the loaded window genuinely cannot answer comes back as “not enough data in range”, with what it would need. That answer is on purpose, and we would rather ship it than a confident guess.
It does not touch your broker
Quant has no order routing and no execution path. It reads market data and the workspace you point it at; nothing it produces reaches an exchange without you doing it yourself.
It does not read what you did not attach
Positions, balances and private notes are outside the context unless you explicitly attach them to the session, and attached context is never used to train a shared model.
Everything people ask about Quant.
The short version: it reads your chart, it counts before it claims, and it shows you where every number came from.
Still stuck? Read the documentation.
The symbol and timeframe on screen, the visible bar range, the indicators on the pane, thirty sessions of history around that range, and any earnings or macro event inside it. Nothing else — it cannot see your positions, your broker or your other tabs unless you attach them to the session yourself.
Ready when you are
Put a question to the tape.
Open a chart, ask Quant what actually happened, and follow the citation to the bar. Fifteen questions a month are free forever, and no card is involved.
- No card required
- Cancel anytime
- 14-day Ultimate trial