About

ForecastCard tracks public market predictions made by well-known financial commentators, traders, and macro analysts to show who predicted what, when, for which asset, and whether they were right or wrong.

How predictions are added

We monitor a list of YouTube interview channels daily for new content. When a new video is published, we pull the full transcript and run it through a two-pass agentic extraction pipeline:

1. Extraction: An LLM reads the transcript and identifies specific, gradeable predictions. To qualify, a prediction must name a specific asset or event, include a directional call or price target, and have an explicit or implied timeframe. Conditional statements (“if X happens, then Y”), vague commentary, and probability hedges are excluded.
2. Verification: Each candidate prediction is checked against the original transcript to confirm it reflects what was actually said. A second LLM pass then filters out anything conditional, vague, or duplicative.

Only predictions that survive both passes are added to the site.

We also run a periodic deduplication pass that catches cases where the same prediction was made by the same person across different videos. For example, when a commentator appears on two shows in the same week and repeats the same call. We keep the clearest version and drop the rest.

Channels we monitor

• The David Lin Report

• The Julia La Roche Show

• Wealthion
• Thoughtful Money
• Kitco NEWS
• ITM Trading
• In the Money
• Palisade Radio
• Sprott Money

We deliberately track interview channels rather than the personal channels of individual commentators. Two reasons: first, many prominent commentators publish daily, so pulling from their own channels would mean processing thousands of videos a month, much of it casual commentary. Second, we want predictions made on record: when someone agrees to a sit-down interview on a well-known show, they’re making a deliberate, public statement that’s worth holding them to. That bar matters more to us than volume.

How predictions are graded

When a prediction’s target date passes, it’s due for grading. There are two approaches:

• Price-based predictions: For predictions tied to specific assets (stocks, commodities, indexes), we pull historical price data from Yahoo Finance covering the prediction period. We check whether the predicted price target or percentage move was actually reached at any point during that window, not just the closing price on the target date.
• Event-based predictions: For macro or political predictions (recession calls, policy outcomes, etc.), we research the actual outcome using public data and news sources.

Some predictions can be graded early. For example, if a price target was already hit well before the target date, or if the remaining time makes the prediction mathematically impossible.

Details

ForecastCard was started in March 2026 and is owned and operated by Scalebloom.

Contact

Have questions or feedback? Reach out at hi@forecastcard.com.