No sign-in · Reading costs 0 G-Points
Real historical research example
Apple Inc. · AAPL · USD
Historical record · Target 10/06/2026 · Reading does not rerun analysis
Read this study first
Apple Inc.AAPL · US
Dates and times shown in America/New_York
- Research horizon
- 7 trading days
- Recorded target date
- 10/06/2026
- Generated at
- 09/26/2026, 02:15
- Last saved historical candle
- 09/25/2026
Current observation · Model scenario
From the recorded price of 341.07 USD, the model reference is 318.9247 USD, a difference of 6.49% down. This is the original model scenario; read it with the limits below.
Source: prices and model outputs in this record. Open the original evidence for its path, method, and saved candles. This summary adds no model call.
Main risks, disagreements, and evidence limits
- Model ranges and stability are not calibrated probabilities of success. Price proximity is not a probability of profit.
- The last saved candle only describes historical price coverage. The full data cutoff and availability times of other materials were not recorded.
- Important disagreement: the model reference points down while the recorded trend context points up, in opposite directions. One path is not a definite direction. This is disagreement within the record, not evidence of multi-model consensus.
- The recorded risk level is 3/5. This model field does not replace a complete risk assessment.
- A complete risk list was not provided. Missing risk notes do not mean there are no risks.
What happened after this study
Later price stored in the original record: 333.63.
The target-date price in this example was checked against an independent unadjusted history read. See the method and limits below. This does not establish that the model conclusion was correct.
Anonymous reading cost
Reading and downloading cost 0 G-Points and create no analysis task. Sign in to study a new asset and review its latest cost before running.
Historical prices checked
Later outcome · Retrospective check
Recorded summary reference: USD 318.92471; target-date observed close: USD 333.63. Direction comparison: consistent.
- Observed close relative to the recorded starting price
- -2.18137%
- SMAPE between reference and observed price
- 4.506991%
Metric version · terminal_direction_v1
Direction compares the reference and observed prices with the recorded starting price; unchanged prices count as non-up. SMAPE = 200 × |reference − observed| ÷ (|reference| + |observed|). It is neither a return nor a probability of success.
Selection: within the authorized AAPL history, choose the latest record with complete saved fields, a matured horizon, and a checkable target-date price. Success is not a selection criterion. One example does not represent overall performance.
Verification evidence and limits
Original platform review generated: 10/07/2026, 08:02 GMT-4 (America/New_York). Independent historical-price check: 10/24/2026, 11:21 GMT-4 (America/New_York). The observed close belongs to the target date, not either review timestamp.
Retrospective price-basis check: a separate Polygon unadjusted history read matches the latest 20 saved input closes and the stored target-date close. The last input close is USD 341.07. Separate checks found 0 split executions and 0 ex-dividend events from 09/25/2026 through 10/06/2026. This does not certify the adjustment basis of the entire input history.
The original record did not retain a full data cutoff, model-file version, or confirmed adjustment basis. These price and corporate-action or factor checks are retrospective; they have not been written back as information known at creation. They cover this example only and do not verify the basis of all platform history.
Complete reference path and later observed closes
The model path retains its original precision. Observed values are later-read unadjusted closes for each trading day. The columns belong to different time stages.
| Trading date | Model path close (USD) | Later observed close (USD) |
|---|---|---|
| 09/28/2026 | 324.42 | 338.4 |
| 09/29/2026 | 322.9 | 329.4 |
| 09/30/2026 | 321.67 | 333.02 |
| 10/01/2026 | 321.03 | 330.32 |
| 10/02/2026 | 320.3 | 333.69 |
| 10/05/2026 | 319.45 | 332.89 |
| 10/06/2026 | 318.92 | 333.63 |
Original evidence, method, and complete snapshot
- Method used at the time
- ensemble
- Model-file version
- Not recorded
- Recorded model stability score
- 92.584737%
- Saved input-price window
- 07/18/2025 through 09/25/2026, 300 rows; requested history 1y
- Full data cutoff
- Not recorded
- Confirmed adjustment basis in the original record
- Not recorded
Recorded method references and weights
Original method outputs and weights are retained, including zero-weight methods. These are neither current defaults nor a quality ranking.
- G-Prophet 2026V1
- USD 328.09
- Recorded weight 50%
- LSTM
- USD 333.019034
- Recorded weight 30%
- Random Forest
- USD 274.87
- Recorded weight 20%
- Transformer
- USD 297.488001
- Recorded weight 0%
The recorded starting price of USD 341.07 and last input close of USD 341.07 are retained separately. The summary reference and terminal path value also keep their original precision and may differ through rounding.
Model stability is not a success probability. Path OHLC ranges are not calibrated confidence intervals. Gaps in the original risk list remain marked as missing.
Download the complete anonymous snapshot (JSON)Includes all saved candles, the complete model path, and retrospective checks for offline calculation. Historical dates are fixed; refreshing creates no new analysis.
Research an asset I follow
Reading and downloading cost 0 G-Points and create no analysis task. Sign in to study a new asset and review its latest cost before running.