Any stock tool can show you the calls that came true. The question that matters is how all of its calls did, including the misses, against something simple you could have done without it. This page explains how Ticker Council answers that question about itself.
The current result, first
As of October 12, 2026: it's too early to tell. About 960 next-week calls have been scored, over 3 market weeks. That isn't enough to judge yet: we don't say "better" or "worse" than any baseline until at least 4 market weeks are in (see step 3). 3-month calls start being scored in December 2026, and 1-year calls in autumn 2027.
The Council is also built to improve as the record grows: seats and models that keep getting it right will count for more. The record will be updated as more calls are scored, whatever it shows.
Step 1: every call is saved where it can't be edited
When a run finishes, its calls are written to a store that can't be changed afterwards: the chance it gave for each period, the price at the time, and when the call should be checked. Nobody, including us, can quietly delete a bad call or adjust one after the fact.
Step 2: each call is checked when its time is up
A week, 3 months or a year later, the price is checked. Did it end higher or lower than when the call was made? A price that didn't move at all isn't scored either way.
Step 3: one stock counts once a day
If thirty people run NVDA on the same morning, the market only gets to answer once: NVDA either rose that week or it didn't. Counting thirty "right" calls would make the record look far bigger and more certain than it is. So for the scorecard, each stock counts once per period per day, no matter how many times it was run.
The same logic applies across the market. Calls made in the same week rise and fall together (when the whole market drops, most "up" calls are wrong at once), so when we say whether the Council is better or worse than a baseline, each market week counts as one piece of evidence, not each call. We don't say "better" or "worse" until there are at least four such weeks.
Step 4: score the chance, not just the direction
"Right or wrong" throws away information. A call of 51% up and a call of 80% up are very different claims, and they should be judged differently. So the main score is the Brier score: the squared gap between the chance given and what happened (1 if the price rose, 0 if it fell), averaged over all calls.
- Saying 50% every time scores exactly 0.25.
- A perfect forecaster scores 0.
- Lower is better. Confident calls that turn out wrong are punished hardest.
For example, a call of 70% up on a stock that rose scores (0.7 − 1)² = 0.09. The same call on a stock that fell scores (0.7 − 0)² = 0.49.
Step 5: compare with baselines anyone could use
A score only means something next to an alternative. The Council is compared with:
- Always 50%: say 50% for every stock.
- The market's usual odds: the same chance for every stock, from how often the market rises over that period (about 54% for a week, 60% for 3 months, 64.5% for a year). If the Council can't beat this, its twelve seats aren't adding anything.
- The stock's own record: how often this particular stock rose over that period in its recent history.
We also check two other things. Calibration: of the calls that said about 60% up, did about 60% actually rise? And ranking: on the same day, did the stocks the Council rated higher tend to do better than the ones it rated lower? Ranking is scored with AUC, where 0.5 means no better than chance and 1.0 means perfect.
Why we publish it
Most "AI stock picker" sites show a highlight reel. We think the full record is the product: it tells you how much weight to give a call, and it tells us which seats and models are helping and which are hurting. Seats that keep getting it right will count for more over time, and only after the scored record shows it.
Not financial advice: this page explains how the record is kept. Past results, good or bad, don't tell you what a stock will do next.
Questions
Is Ticker Council accurate?
It's too early to tell. As of October 12, 2026, about 960 next-week calls have been scored over 3 market weeks, which isn't enough to judge: the record isn't compared with any baseline until at least 4 market weeks are in. 3-month calls are first scored in December 2026 and 1-year calls in autumn 2027.
What is a Brier score?
It's the average squared gap between the chance a forecast gave and what happened (1 if the event happened, 0 if not). Always saying 50% scores 0.25, a perfect forecaster scores 0, and lower is better.
Why does each stock only count once a day?
Because many people running the same stock on the same day are one outcome, not many. Counting each run separately would make the record look bigger and more certain than it really is.
Can old calls be edited or deleted?
No. Calls are saved unedited when the run finishes and scored later, so bad calls stay in the record.