Quantitative methodology

AIM combines market analysis with its own quantitative engine. The engine generates thousands of possible price paths for the asset and the horizon of the analysis and counts how many reach the scenario’s target, how many reach the invalidation first and how many reach neither level. The result describes the scenario; it does not tell you what to do.

• Reference prices for the asset (5-minute, 1-hour and 1-day candles) from a real-time market data source.
• Only candles closed before the moment of the analysis are used; each result records the time of the quote used.
• The asset’s reference spread at that moment. Your broker’s prices may differ.

The engine uses a filtered historical simulation: it takes the asset’s recent historical moves, normalizes them by the volatility of each moment (exponential moving average) and reuses them in consecutive blocks, rescaled to current volatility. This preserves real market features, such as fat tails and volatility clustering, without assuming a normal distribution. The current model version is identified as fhs-pathwise-entry-v2and is recorded with every analysis.

• The default horizon is 24 hours and can range from 1 to 240 hours depending on the scenario’s timeframe.
• Up to 48 hours the simulation runs in 5-minute steps; for longer horizons, in 1-hour steps.
• If the scenario has a pending entry zone, only paths that reach it count, and from that point only the remaining time of the horizon is used.
• If the entry depends on a candle close, the simulated close of that timeframe is used, not the level as if it were filled exactly there.

The reference spread is deducted and assumed to be fixed. Commissions, swaps and slippage are included only if they are explicitly declared; otherwise they are not assumed. For a position that is already open, only the additional costs from that moment on are taken into account.

• Favorable scenarios (X of N): how many of the N simulated paths reach the target before the invalidation within the horizon.
• Invalidation: the level that, if reached, voids the scenario.
• Expected value (R): the average outcome of the simulated paths, expressed in multiples of the distance between entry and invalidation.
• High uncertainty: shown when favorable and unfavorable scenarios are practically balanced.

These figures describe the model’s behavior, not the real probability that something will happen nor a win rate. A scenario with many favorable cases may still not play out.

Each AIM model simulates a different number of scenarios per analysis: AIM Flash 1,000, AIM Pro 5,000, AIM Ultra 10,000 and AIM Apex 25,000. More scenarios reduce the noise of the simulation itself and add resolution; they do not make the market more predictable or guarantee better results.

• The model is evaluated chronologically and out of sample: it simulates with the data available on each past date and compares the result with what happened afterwards, without using future information.
• Quality is measured with calibration metrics (such as the Brier score) and intervals that quantify simulation noise.
• Results are separated by model version, and cases that are ambiguous within a single candle are excluded.
• Each analysis is recorded with its asset, horizon, data time, model version and number of scenarios so it can be reviewed later.

• It does not account for every market closure, holiday or sudden change in future volatility.
• It does not foresee news or unexpected events; a macroeconomic release can invalidate any scenario.
• It works with mid prices and a fixed spread; it does not reproduce your real execution or your broker’s order book.
• Favorable historical results of the model do not certify future returns.

Scenarios are information to help you analyze the market. AIM does not tell you what to buy or sell or how much to risk: those decisions, and their consequences, are yours.