1. Build a stock basket
Start with one or more stocks. For each line, specify the ticker, whether you intend to buy or sell, the quantity and your execution conditions. The basket lets the analysis consider both individual trades and how the lines fit together.
Quantity matters. A hypothetical order for 100 shares and one for 10,000 shares of the same stock can represent very different shares of its daily trading volume. Basket composition matters too: several stocks can concentrate exposure in one sector or country even when no single line looks unusually large.
You do not submit an order to PreTrAIde. You are describing a possible trade for analysis, not authorizing a transaction.
2. Enrich each stock with available data
For each stock, PreTrAIde draws on about 57 fields from market data, including end-of-day prices, trading volumes and company profile information. It also uses recent news by ticker and analyst ratings.
These inputs serve different purposes. Prices and volumes support numerical estimates; company information supports basket breakdowns; news and ratings provide context for the AI layer. An analyst rating is an input to consider, not PreTrAIde’s verdict.
PreTrAIde covers US stocks and major European exchanges, including Euronext Paris, Xetra and the London Stock Exchange. It is available in English, French, Spanish, German, Italian and Portuguese.
3. Calculate pre-trade metrics
The quantitative layer is deterministic and reproducible. It estimates trading friction, risk and basket composition before the AI agents reason about them.
- Estimated spread: a liquidity-tier model with a volatility uplift and a one-tick floor estimates the bid-ask spread. The spread is a potential cost of crossing from the bid to the ask, not a broker commission or a guaranteed fill cost.
- Participation in daily volume: estimated order size is compared with daily trading volume. For example, a hypothetical 10,000-share order against 1,000,000 shares of daily volume represents 1% of that volume. The figure does not tell you how much volume will be available at the moment you trade.
- Market impact: an Almgren-Chriss-style model estimates the price effect associated with executing an order. This is a model estimate, not a prediction of an actual fill.
- Daily volatility: the system blends Parkinson, close-to-close, Yang-Zhang and simplified GARCH(1,1) estimators.
- One-day 95% value at risk (VaR): this estimates a loss threshold over one day at a 95% confidence level under the model’s assumptions. It does not cap possible losses; outcomes beyond the threshold remain possible.
PreTrAIde also scores data quality, assigns liquidity scores and tiers, and produces 0–100 liquidity, volatility and execution scores for each stock and the basket. Basket breakdowns cover country, sector, region, currency and market capitalization. A Herfindahl-Hirschman index (HHI) measures concentration by summing squared position weights: a hypothetical basket weighted 50%, 30% and 20% has an HHI of 0.5² + 0.3² + 0.2² = 0.38 on a 0–1 scale. A higher value indicates greater concentration.
This is pre-trade analysis, not chart-pattern technical analysis. Its market-microstructure references include Kyle (1985), Glosten-Milgrom (1985), Admati-Pfleiderer (1988) and Almgren-Chriss (2000); PreTrAIde adapts institutional pre-trade and transaction-cost analysis for private investors.
4. Add AI reasoning for each line and the basket
AI agents examine each stock and the whole basket, anchored on the quantitative scores, news and analyst ratings. The output explains the reasoning in plain language, so you can see whether a concern relates to a particular line or the basket as a whole.
The distinction matters. A stock might have an estimated execution cost worth examining on its own, while the basket could raise a separate concentration concern. PreTrAIde reports both levels rather than treating a collection of lines as one undifferentiated trade.
5. Apply guardrails before showing a verdict
Guardrails correct AI output that contradicts the data and only move verdicts in a more conservative direction. A suspended stock is always NO-GO. A GO paired with a success probability below 50% is downgraded. Probabilities are constrained to the 5%–95% range.
These rules do not make the estimates certain. They are checks on how the AI presents its conclusion, including whether a favorable label is consistent with its stated probability.
6. Read the result as decision support
The PreTrade analysis returns GO, NO-GO or MODIFY for the basket and for each line. Alongside each verdict are a success probability, risk assessment, suggested execution strategy and plain-language reasoning.
A success probability is a model estimate between 5% and 95%. It is not a guaranteed outcome, a forecast of investment returns or a prediction of the stock’s price. Likewise, GO is not an instruction to buy or sell, and NO-GO is not a prohibition on trading. The result is decision support for reviewing the proposed basket and its execution conditions.
Each analysis uses one analysis credit, also called a token. If you want to inspect the underlying ideas separately, PreTrAIde’s free tools include a position size calculator, trading cost calculator, portfolio concentration checker, value-at-risk calculator and pre-trade checklist. They require no sign-up.
7. Explore a modification when the verdict is MODIFY
A MODIFY verdict gives you the option to use Let AI Modify. It can rework the basket automatically, once per analysis, at no additional analysis-credit cost. Across up to five AI passes, it may trim, exit or replace lines, subject to these limits:
- No more than two new lines.
- At least half of your original lines remain.
- No single position exceeds 40% of the basket.
The feature presents a revised basket to examine; it does not place an order. Its constraints limit how far an automatic revision can depart from what you entered, but they do not remove trading risk.
8. Paper-trade the basket
After a verdict, you can open a paper-trading simulation and follow how the basket would have performed. The simulator is available on every plan and uses no real money.
A simulation can help you compare a proposed basket with later market outcomes. It does not establish that a future live trade would receive the same fills or produce the same result.
9. Log what you actually did
If you trade through your own broker, you can log whether a proposed trade was executed, partially executed or skipped, along with real fill prices. PreTrAIde measures the outcome using market data and builds PreTrAIde Memory for your account.
That memory personalizes later AI-layer analyses; the deterministic quantitative scores remain identical. Personalization levels unlock at 1, 5, 15 and 40 logged trades. You can pause or erase the memory at any time. Logging a trade is optional: PreTrAIde never receives, transmits or executes your order.
Getting started
The Basic plan includes a 7-day free trial with one analysis. Basic allows up to five stocks per basket and 10 analyses per month; other plans allow larger baskets and more monthly analyses. See the pricing page for current plan details. PreTrAIde is analysis software, not a broker, investment adviser or portfolio manager: you remain responsible for any decision and place any trade yourself with your own broker.