FuturesAI Scoreboard: Why 341 and 326 Calls Both Appear
Short answer: The FuturesAI scoreboard shows 341 graded predictions in its market breakdown but 326 in its headline box. The gap is 15 predictions and 7 hits, which matches the 15 user predictions and 7 user hits the page lists separately. Either way the hit rate is 45.7%. The average return of +0.03% per prediction is before fees. Scoreboard checked November 17, 2026. Not investment advice.
Two totals on one page. Both canโt be the whole scoreboard.
That is the kind of thing worth five minutes before you trust any track record. Below: the numbers, the subtraction that reconciles them, what the +0.03% average return looks like after fees, and a five-check routine for any scoreboard you read.
What This Checks
- What does the FuturesAI scoreboard actually show?
- Why do the 341 and 326 totals differ?
- What does the +0.03% average return look like after fees?
- How do you audit any prediction scoreboard in five checks?
- What can this audit not tell you?
What does the FuturesAI scoreboard actually show?
As of its last automatic check on November 15, 2026, the FuturesAI scoreboard reports a 45.7% hit rate, an average return of +0.03% per prediction before fees, and a market benchmark of +2.00% over the same period, a gap of โ1.96 percentage points (futuresai.io scoreboard). Each prediction is graded 20 days after it is made, against the closing price.
When I checked the page, it lists three markets: domestic stocks at 41.7%, US stocks at 46.9% and crypto at 50.0%. It also says wrong predictions are published, which is what makes a check like this possible at all.

Why do the 341 and 326 totals differ?
The 341 and 326 totals differ by 15 graded predictions and 7 hits, and those two numbers match the separate user prediction line (7 hits of 15) exactly. The scoreboard does not state this, so treat it as an inference from the arithmetic, not a confirmed explanation.
| Line on the scoreboard | Hits / graded | Hit rate |
|---|---|---|
| Domestic stocks | 55 / 132 | 41.7% |
| US stocks | 53 / 113 | 46.9% |
| Crypto | 48 / 96 | 50.0% |
| Market lines added up | 156 / 341 | 45.7% |
| Up calls + down calls added up (96 + 60) / (215 + 126) | 156 / 341 | 45.7% |
| Headline box | 149 / 326 | 45.7% |
| User predictions | 7 / 15 | 46.7% |
| Headline + user | 156 / 341 | 45.7% |
I found that the market lines (132, 113 and 96) and the direction lines (215 and 126) both sum to 341, and that both hit counts sum to 156. A coincidence is unlikely when two independent splits agree. The headline box then sits exactly one user group below that.
Honestly, that is a reassuring result. Nothing is double counted. The page just mixes two bases.
What does the +0.03% average return look like after fees?
An average return of +0.03% per prediction turns into between +0.018% and โ0.058% once you subtract round-trip fees at the effective rates listed on the FuturesAI payback page on November 17, 2026. The scoreboard states that its returns exclude fees and use equal-size trades.
| Cost assumption (per trade, both sides) | Fee on the position | Average return +0.03% minus fees |
|---|---|---|
| No fees (as the scoreboard reports) | 0.000% | +0.030% |
| Lowest limit-fee pair on the payback page (Gate, 0.006% each way) | 0.012% | +0.018% |
| Bybit limit-fee pair (0.016% each way) | 0.032% | โ0.002% |
| Bybit market-fee pair (0.044% each way) | 0.088% | โ0.058% |
This is arithmetic on two public pages, not a backtest. Real fills, funding and spreads would move it, and the scoreboard also says the market itself returned more over the period. The point is narrower: a very small average return leaves almost no room for fees.
This part surprises people.
A 45.7% hit rate sounds poor, yet the average return is slightly positive because the profit-to-loss ratio is 1.20. Hit rate alone does not decide the outcome. Our earlier post on where cryptoโs 52.2% sits after fees works through the break-even side of this.
How do you audit any prediction scoreboard in five checks?
You audit a prediction scoreboard by adding up its parts, testing a second split, hunting for a small group that explains any gap, and confirming which total each statistic uses. These five checks took about ten minutes here.
| Check | What to look for | Result here |
|---|---|---|
| 1 | Do the parts add up to the total? | Market lines give 341, headline says 326 |
| 2 | Can a second split reproduce the same total? | Up plus down also gives 341 |
| 3 | Is there a small group that explains the gap? | User predictions: 15 graded, 7 hits |
| 4 | Does subtracting that group close both gaps? | 341 โ 15 = 326 and 156 โ 7 = 149 |
| 5 | Which total does each stat use? | Return stats use 326 |
If you want a sense of how wide the uncertainty is on a sample this size, see how wide the error band is, and for a ledger-style tally method, how to tally a ledger by basis tag.
What can this audit not tell you?
This audit cannot tell you whether the calls will work from here, whether the 20-day grading window suits your own holding period, or what the user predictions were. It only checks whether the published numbers are internally consistent.
- The page shows a 100-entry ledger sample, which is not the full set, so its crypto line (a different rate from the 50.0% headline) should not be compared directly.
- The sample covers prediction dates from August 31 to November 11, 2026, a short window.
- Per-ticker rates rest on 5 to 7 calls each. Treat them as noise.
- Past results, even clean ones, are not a forecast.
Before you act on any score, write down the base it was computed on. If the page does not tell you, assume nothing.
Not investment advice. FuturesAI describes its scores as observation tools, not buy or sell recommendations. Trading, especially with leverage, can lose your full principal.
FAQ
Why does the FuturesAI scoreboard show both 341 and 326 predictions?
The page does not explain it, but the arithmetic fits one reading. The market and direction lines add up to 341 graded predictions and 156 hits, while the headline box shows 326 and 149. The difference is 15 and 7, which are exactly the 15 user predictions and 7 user hits listed separately.
Which number should I use for the hit rate?
Either one. Both 156 of 341 and 149 of 326 round to 45.7%. For the return statistics, use the 326 base, because that is the base the page says its return figures use.
Does a 45.7% hit rate mean the calls lose money?
Not by itself. Hit rate and return are different things. The page reports an average return of +0.03% per prediction before fees, and fees can turn that negative, as the fee table in this post shows. Nothing here predicts future results, and trading can lose your whole principal.
Sources
- FuturesAI prediction scoreboard (checked Nov 17, 2026)
- FuturesAI payback page, effective exchange fees (checked Nov 17, 2026)
Figures read from the live FuturesAI pages on November 17, 2026. The reconciliation and fee table are our own arithmetic.
๐ Related guides
