How Quanta AI's Time Machine Historical-Analogue Workflow Works

Market researchers often ask whether a current setup resembles earlier ones. Similar appearance does not guarantee a similar outcome, so the matching method and the full range of analogue results matter.

Given a stock's recent price action, the Time Machine searches a versioned historical index for ranked matches. It then summarizes what happened after those analogue patterns and reports a relative model-ranking score with outcome context.

Think of it as a research index that retrieves comparable precedents under a documented matching method. Its results still require human review.

The Time Machine represents the current price sequence and related market features under a versioned research method.

The exact inputs, preprocessing, and supported horizons can change with a model revision. Read the live result, timestamp, and current methodology instead of relying on a permanent indicator list or lookback claim in an article.

This representation is used to retrieve comparable indexed cases. It does not establish that the current setup will follow the same path.

The engine compares the current setup with a versioned historical index using sequence and feature similarity.

The returned cases are ranked under that method. A close mathematical match is evidence of resemblance in the chosen representation, not proof that the future outcome will repeat.

The model score summarizes relative model evidence for ranking. It is not a calibrated probability of a profitable outcome.

Once comparable historical patterns are identified, the Time Machine organizes the evidence:

Forward Paths — It shows what happened after the matched historical patterns at supported horizons. Those analogue outcomes describe historical cases; they do not constitute a backtest of a tradable strategy.

Model Score — The score summarizes relative model evidence used to rank results. It must not be read as the probability that a signal will win.

Outcome Range — The range of paths that followed the selected analogues supplies descriptive context. It is not a portfolio-risk forecast or an automatic sizing rule.

Outcome Summary — Descriptive statistics summarize the retrieved analogues. For forward model evaluation, use the separate, matured signal cohort on the [Results page](/track-record).

Model descriptions are more useful when readers can inspect outcomes. Quanta's [public high-model-score Results ledger](/track-record) reports matured directional signal observations with the current win rate, sample size, confidence interval, average direction-adjusted move, as-of date, and methodology.

That ledger is signal-level evidence. It does not model position sizing, compounding, slippage, fees, or a user's execution, so it should not be described as a portfolio return.

Legacy paper-portfolio simulations are retained for internal research while their temporal split, execution assumptions, and provenance are rebuilt. They are not used as public performance claims or presented as live capital.

A historical comparison is meaningful only with a declared chart timeframe and outcome horizon. Check the live analysis for the currently supported choices and keep evidence from materially different horizons separate.

Where more than one timeframe is available, agreement or disagreement can add context, but neither configuration guarantees an outcome.