What it means
An AI system can bring several kinds of stock information into one assessment, but the question it answers depends on how the tool is designed. It might summarize company reports and news, classify sentiment, estimate risk, or examine the likely cost of executing a proposed trade.
An AI stock picker typically focuses on which stocks may be attractive according to its criteria. AI pre-trade analysis instead examines a trade or basket someone has already proposed: its size, execution conditions, liquidity, risk and potential trading costs. Neither approach can establish a future return with certainty. A plausible explanation is not proof that an estimate is accurate.
How it is calculated
There is no universal AI stock analysis formula. Models differ in their inputs, targets and methods. One measurable input to a pre-trade check is the proposed order’s share of typical daily volume:
Participation rate (%) = order quantity (shares) / average daily volume (shares per trading day) × 100
- Order quantity is the number of shares proposed for that stock.
- Average daily volume is the average number of shares traded per trading day over a specified lookback period. The period matters because volume changes.
- Participation rate expresses the order as a percentage of one typical day’s volume. It is a screening measure, not a guarantee of how much will be available when an order is placed.
An AI tool may combine this measure with other inputs, such as price history, news or analyst ratings. Any overall score or probability is model-specific; it cannot be calculated from participation rate alone.
Worked example
Suppose, hypothetically, someone proposes buying 2,000 shares of a US stock at US$50 per share. The proposed value is US$100,000. Assume its average daily volume over the chosen period is 200,000 shares per trading day.
- Divide the order quantity by average daily volume:
2,000 / 200,000 = 0.01. - Convert to a percentage:
0.01 × 100 = 1%. - Interpret it narrowly: the order equals 1% of a typical day’s share volume, not 1% of shares necessarily available at US$50.
An AI stock picker might evaluate the company using its chosen criteria without knowing this proposed order size. A pre-trade analysis can use the order details to examine execution conditions. Neither the 1% figure nor an AI explanation establishes whether the stock’s price will rise or whether the proposed trade will succeed.
Why it matters before you trade
A stock thesis and an executable trade are different questions. Even if an analysis presents a convincing case about a company, the proposed quantity, bid-ask spread, volatility and market impact can affect the cost and risk of acting on it. News may also be incomplete or quickly outdated.
AI can help organize evidence and surface issues for review. It cannot verify every source automatically, remove uncertainty or know your personal circumstances merely from a ticker. Its outputs deserve particular scrutiny when data are stale, a model’s assumptions are unclear, or a confident-sounding explanation lacks support in the inputs.
How PreTrAIde uses it
PreTrAIde is an AI platform for pre-trade analysis of stocks, not a stock-picking promise or a broker. A user builds a basket with a buy or sell side, quantity and execution conditions for each stock. Before any order is placed, PreTrAIde analyzes the basket and each line, returning a GO, NO-GO or MODIFY verdict, a model-estimated success probability, a risk assessment, a suggested execution strategy and plain-language reasoning.
Its deterministic quantitative layer estimates measures including spread, participation in daily volume, market impact and volatility, and produces liquidity, volatility and execution scores. AI agents reason from those scores, recent news and analyst ratings. Guardrails can correct contradictory AI output in the conservative direction. The success probability is constrained to 5%–95%; it is neither a guarantee nor a forecast of returns. PreTrAIde does not receive, transmit or execute orders. Users trade, if they choose to, through their own broker.
Common mistakes
- Treating an AI verdict as an instruction. Decision support does not replace an investor’s judgment or constitute a recommendation.
- Confusing a success probability with expected return. A probability estimate does not specify a future price or profit.
- Assuming a polished explanation validates its inputs. Unsupported claims, stale information and model errors can still produce fluent text.
- Equating stock selection with execution analysis. A view on a company does not account for the costs and risks of a particular proposed order.