FuturesAI Signals: A 3-Gate Filter Before You Trust One
Short answer: Run each score group through three gates: at least 100 graded calls, a 95% range whose low end clears the break-even line, and a traded-as-predicted return that beats the market over the same windows. On FuturesAI’s public scoreboard read on November 6, 2026 (141 hits in 308 graded calls, 45.8%), none of the six groups I checked passes all three. Crypto, at 55.2% on 87 calls, passes none of the hard ones.
A signal feed is easy to trust and hard to test. The scoreboard makes testing possible, because it publishes the misses next to the hits.
Below is a filter you can reuse, applied to the live numbers. It is arithmetic on published figures. Nothing here is a forecast or a recommendation.
The Gates
- Gate 1: is there enough sample?
- Gate 2: does the range clear break-even?
- Gate 3: does it beat just holding the market?
- How do the six groups score?
- How many more calls would crypto need?
Gate 1: how many graded calls are enough to judge a hit rate?
For this filter, a group needs at least 100 graded calls before its hit rate is worth testing, because below that a hit rate near 50% carries a 95% range wider than 20 percentage points.
That is a rule I picked, not a FuturesAI rule. With 100 calls and a 50% hit rate, the standard binomial margin is about plus or minus 10 points. With 87 calls it is a bit wider. Pick your own cutoff, but pick it before you look at the table.
Gate 2: does the 95% range clear the break-even hit rate?
A group passes Gate 2 when the low end of its 95% range sits above the break-even hit rate, which is 1 / (1 + payoff ratio), or 46.7% at the scoreboard’s overall payoff ratio of 1.14.
When I checked the scoreboard on November 6, 2026, it listed a payoff ratio of 1.14, defined as money made when right divided by money lost when wrong, with UP calls averaging +1.79% (191 calls) and DOWN calls averaging -3.45% (117 calls). I used the overall 46.7% line for every group because the page does not publish a payoff ratio per group. A group with its own ratio could have a different line.
The range uses the Wilson interval, a standard way to put a 95% band on a hit rate. It is my calculation, not a FuturesAI figure.
Gate 3: does a trader who followed the calls beat the market?
A group passes Gate 3 when the average return of trading every call in it, at equal size, is higher than the market’s average over the same windows.
The scoreboard gives this for the three markets: Korean stocks -0.60% against +3.01% for the market, US stocks -2.24% against +0.43%, crypto +2.97% against +6.34%. The page notes that it uses the price at the time of the call and the price at the scoring date, with no fees subtracted. Fees would only widen the gaps.
How do the six groups score against the three gates?
Of the six groups on the November 6, 2026 scoreboard, none passes all three gates: five clear Gate 1 but fail Gate 2, and crypto fails Gate 1 as well.
| Group | Hits / graded | Hit rate | 95% range (own calc) | Gate 1 (100+) | Gate 2 (low end > 46.7%) | Gate 3 (beats market) |
|---|---|---|---|---|---|---|
| All graded calls | 141 / 308 | 45.8% | 40.3% to 51.4% | Pass | Fail | Not published |
| Korean stocks | 51 / 127 | 40.2% | 32.0% to 48.9% | Pass | Fail | Fail (-3.61 pts) |
| US stocks | 49 / 109 | 45.0% | 35.9% to 54.3% | Pass | Fail | Fail (-2.67 pts) |
| Crypto | 48 / 87 | 55.2% | 44.7% to 65.2% | Fail | Fail | Fail (-3.37 pts) |
| UP calls | 96 / 205 | 46.8% | 40.1% to 53.7% | Pass | Fail | Not published |
| DOWN calls | 52 / 118 | 44.1% | 35.4% to 53.1% | Pass | Fail | Not published |
Source for hit counts and market gaps: FuturesAI’s public scoreboard, last auto-check dated November 4, 2026, read November 6, 2026. Ranges and gate results are my own arithmetic on those figures.
This table can disappoint in both directions. A fail is not proof the signals are useless. It means the published sample cannot show they work yet.

How many more calls would crypto need to pass Gate 2?
If the crypto group kept a 55.2% hit rate, it would need about 150 graded calls for the low end of its 95% range to clear 46.7%, roughly 63 more than the 87 it has now.
| Graded crypto calls | Hits at 55.2% | 95% range | Low end above 46.7%? |
|---|---|---|---|
| 87 (now) | 48 | 44.7% to 65.2% | No |
| 150 | 83 | 47.3% to 63.1% | Yes, barely |
| 200 | 110 | 48.1% to 61.7% | Yes |
| 300 | 166 | 49.7% to 60.9% | Yes |
That is a what-if. The hit rate may fall or rise as calls accumulate, which is exactly why the gate exists. Even a pass here would leave Gate 3, the market gap, open.
For the same crypto number tested against luck, see FuturesAI Scoreboard: Is the Crypto 55.8% Just Luck?. For the earlier four-question version of this filter on October data, see FuturesAI Signals: Filter Them With Real Hit Rates.
Here is where people get tripped up: a single strong row feels like proof. I noticed that the crypto row is the one most people quote, and it has the smallest sample on the page. Check the sample under it first.
Tips
- Set your gate thresholds before you read the table, then do not move them.
- Re-run the gates when the scoreboard updates. Its auto-check date is printed at the top.
- Use the break-even line from the payoff ratio, not 50%.
- Compare returns with the market over the same windows, not with zero.
Warnings
- The ranges and gate verdicts are my own arithmetic. They assume each graded call is an independent trial, which is a simplification.
- The scoreboard computes returns without fees and with equal-size trades. Real results differ.
- Leveraged futures can lose the full principal. Nothing here is investment advice, and FuturesAI states its signals are for information, not a buy or sell recommendation.
FAQ
Why use 46.7% instead of 50% as the line?
Because the scoreboard’s payoff ratio is 1.14, so a winning call earns slightly more than a losing call costs. The break-even hit rate is 1 / (1 + 1.14) = 46.7%, before fees. With fees it moves up.
Can a group fail Gate 2 and still be worth watching?
Yes. Failing means the published sample cannot rule out break-even or worse. A group that is close, like UP calls at 46.8%, may pass later with more graded calls, or may not.
Does Gate 3 matter if the hit rate is high?
It does. A high hit rate can sit next to a weak return if the losses are larger than the wins. Crypto shows a 55.2% hit rate on the scoreboard and still trails the market by 3.37 points.
Sources
- FuturesAI AI Prediction Scoreboard (read November 6, 2026; last auto-check November 4, 2026)
- FuturesAI payback page (read November 6, 2026)
🔗 Related guides
Not investment advice. Leverage can lose the full principal. Hit counts, payoff ratio and market returns are from the FuturesAI scoreboard as of November 6, 2026 and update automatically. Past hit rates do not guarantee future results. Intervals and gate verdicts are the author’s own calculations. Fact-checked based on public sources as of November 6, 2026.
