FuturesAI Scoreboard: Is the Crypto 55.8% Just Luck?
Short answer: On November 3, 2026, the FuturesAI scoreboard showed crypto predictions right 48 times out of 86, or 55.8%. That is the best of three markets. It is also a small sample. A coin flip would reach 48 or more about 17% of the time, and picking the best of three markets makes a lucky-looking number more common. Treat it as a lead to keep tracking, not proof.
One number on a scoreboard looks better than the rest. The question is whether it earned that.
Below is the arithmetic, done by hand from the published counts, so you can repeat it on any signal you follow.
Not investment advice. Past hit rates do not predict future results, and trading with leverage can lose your entire principal. The scoreboard is a grading record, not a promise of profit.
What This Works Out
- What does the FuturesAI scoreboard show right now?
- How wide is the margin on each market?
- How often would luck alone produce 55.8%?
- Why does picking the best of three matter?
- How many predictions would settle it?
- A Four-Line Rule for Any Hot Streak
What does the FuturesAI scoreboard show right now?
The FuturesAI prediction scoreboard showed 138 correct out of 302 graded AI predictions, a 45.7% hit rate, with grading data through November 3, 2026 (futuresai.io scoreboard).
It splits that by market: domestic stocks 49 of 123, US stocks 48 of 108 and crypto 48 of 86. The page grades each call by comparing the price when the prediction was made with the price on the grading date, roughly 20 days later, and says fees are not deducted.
| Market | Hits / graded | Hit rate |
|---|---|---|
| Domestic stocks | 49 / 123 | 39.8% |
| US stocks | 48 / 108 | 44.4% |
| Crypto | 48 / 86 | 55.8% |
| All markets | 138 / 302 | 45.7% |
When I checked the page, I found that it labels the number a hit rate, not a probability of profit. A call can be โrightโ on direction and still lose money after fees, which is a separate question this post leaves alone.
How wide is the margin on each market?
A hit rate measured on 86 predictions carries a 95% margin of about ยฑ10.5 percentage points, because the margin is 1.96 ร โ(p ร (1 โ p) รท n) and n is small.
| Market | Hit rate | n | 95% margin | Plausible true range |
|---|---|---|---|---|
| Domestic stocks | 39.8% | 123 | ยฑ8.7 pts | 31.2% to 48.5% |
| US stocks | 44.4% | 108 | ยฑ9.4 pts | 35.1% to 53.8% |
| Crypto | 55.8% | 86 | ยฑ10.5 pts | 45.3% to 66.3% |
| All markets | 45.7% | 302 | ยฑ5.6 pts | 40.1% to 51.3% |
All three market ranges overlap. Cryptoโs lower end sits below 50%, and the all-market range straddles it.
How often would luck alone produce 55.8%?
If crypto predictions were pure coin flips at 50%, getting 48 or more right out of 86 would happen about 16.6% of the time, and at the all-market rate of 45.7% it would happen about 3.8% of the time (exact binomial, worked from the published counts).
| If the true hit rate wereโฆ | Chance of 48+ right out of 86 |
|---|---|
| 50.0% (a coin flip) | 16.6% |
| 45.7% (the all-market rate) | 3.8% |
Against a coin flip, 17% is not rare. Against the weaker all-market rate it is under 4%, which looks more interesting, with one catch: crypto is part of the 302, so that comparison is a little circular.
Here is the flip side. Domestic stocks sit at 49 of 123, and a fair coin would produce that few or fewer about 1.5% of the time. That one is further from a coin flip than crypto is, just in the unhelpful direction.
Why does picking the best of three matter?
When you look at three markets and highlight the best one, the chance that at least one looks as good as crypto does by luck is about 11%, not 4%, if the markets truly share the all-market rate.
The working is short: 1 โ (1 โ 0.038)ยณ = 0.110. I noticed that this is the step most scoreboard readers skip. You only ever see the winner, so the winner always looks special.
One reading in about nine, then, would produce a crypto-sized standout from nothing but chance.
How many predictions would settle it?
To pin a 55.8% hit rate down to ยฑ5 points at 95% confidence, you would need about 379 graded crypto predictions, versus the 86 on the scoreboard today.
| Graded crypto predictions | 95% margin around 55.8% |
|---|---|
| 86 | ยฑ10.5 pts |
| 150 | ยฑ7.9 pts |
| 250 | ยฑ6.2 pts |
| 379 | ยฑ5.0 pts |
The scoreboard keeps adding graded calls, so this number will move. Come back to it every few weeks and redo the table.
If you also want to see how streaks of misses behave at this hit rate, see FuturesAI Hit Rate: How Long a Losing Streak Gets, and for the general method, How Many Trades Before You Know a Signal Works?
For the filtering method itself, see FuturesAI Signals: Filter Them With Real Hit Rates.
A Four-Line Rule for Any Hot Streak
Write the count, not just the percent. Add the margin. Ask how many groups you looked at. Come back when n has doubled.
That is the whole filter. Do it before you size a position on any number that looks good.
Tips
- Always record hits and total, such as 48 / 86, next to any percentage.
- Compute the margin with the formula above before you compare two groups.
- Revisit the scoreboard on a schedule and keep dated screenshots of the counts.
- Check whether fees and funding costs would change the answer for your own trades.
Warnings
- A hit rate is not a return. The scoreboard says fees are not deducted.
- A 20-day grading window is not the holding period you will actually use.
- Leveraged trading can lose the full principal. This post is not investment advice.
FAQ
What is the FuturesAI scoreboard hit rate?
As of November 3, 2026 the FuturesAI scoreboard shows 138 correct out of 302 graded AI predictions, a hit rate of 45.7%. It grades the price on the prediction date against the price on the grading date, with no fees deducted.
Does a 55.8% crypto hit rate prove the signals work?
No. The crypto figure is 48 of 86 predictions. With a 95% margin of about 10.5 points on a sample that size, the true rate could sit anywhere from roughly 45% to 66%, which includes a coin flip.
How many predictions would make the crypto number believable?
If the true hit rate really were 55.8%, a margin of plus or minus 5 points would need about 379 graded crypto predictions, compared with the 86 on the scoreboard now.
Sources
Counts read from the FuturesAI scoreboard on November 3, 2026. All probabilities were worked out from those counts using the binomial formula. Not investment advice.
๐ Related guides
