No, SaltyBet is not rigged. We tested every angle — side balance, scheduling bias, outcome manipulation, pot-size conspiracy — across 585,534 matches and 10,257 fighters [1]. Every test came back clean. Here’s the full evidence chain.
Before digging into the data, let’s define what “rigged” would actually look like. A manipulated system would leave fingerprints in at least one of these five areas:
SaltyTrack has recorded 585,534 matches since December 2021, including every bet amount, every outcome, and every tier assignment [1]. We’re going to test all five claims in order — and each one builds on the last.
If someone is rigging SaltyBet, the simplest way to do it is at the most fundamental level: side assignment. Load the coin — make one side win more often — and you can control outcomes without touching anything else. So that’s where we start. If the coin is loaded, nothing else matters.
The data [1]:
| Side | Wins | Win Rate |
|---|---|---|
| Red | 288,365 | 49.25% |
| Blue | 297,169 | 50.75% |
Blue has a 1.5-point edge across 585,534 matches. A perfectly fair coin flipped 585,000 times would routinely produce a gap this size or larger — this is textbook statistical noise.
The split holds across every tier [1]:
| Tier | Red Win Rate | Blue Win Rate |
|---|---|---|
| P | 50.3% | 49.7% |
| X | 49.6% | 50.4% |
| B | 49.5% | 50.5% |
| A | 49.2% | 50.8% |
| S | 49.0% | 51.0% |
No tier shows a side imbalance greater than 2 percentage points. The coin is fair.
A fair coin means that side assignment alone can’t explain why your favorites keep losing. So if something is rigged, it’s not at the coin level — it would have to be deeper. The next place to look: the matchmaking itself. Because you don’t need to load the coin if you can choose which fighters show up.
A subtler way to rig a system is through the schedule. Over-schedule fighters whose results are predictable, and you can steer outcomes without touching the fights themselves. Repeat “entertaining” matchups, and you can drive betting volume on fights you already know the answer to.
With 10,257 unique fighters in the roster, here’s how evenly SaltyBet cycles through them [1]:
| Appearances | Fighters | % of Roster |
|---|---|---|
| 1–10 | 1,021 | 10.0% |
| 11–50 | 2,245 | 21.9% |
| 51–100 | 800 | 7.8% |
| 101–200 | 4,964 | 48.4% |
| 201–300 | 1,227 | 12.0% |
Nearly half the roster (48.4%) falls in the 101–200 appearance range. The most any single fighter has appeared is approximately 294 times out of 585,534 matches — that’s 0.05% of all matches. No fighter is being force-fed into the schedule.
At the matchup level, it’s even more striking. A 10,257-fighter roster creates roughly 52.6 million possible pairs. Here’s how many times specific matchups have occurred [1]:
| Times Matched | Matchup Pairs | Total Matches |
|---|---|---|
| 1 (unique) | 547,378 | 547,378 |
| 2 | 18,287 | 36,574 |
| 3 | 514 | 1,542 |
| 4–5 | 10 | 40 |
93.5% of all matchup pairs have occurred exactly once. The maximum any two specific fighters have faced each other is 4 times across 585,000+ matches. The system cycles through an enormous variety of combinations — the opposite of a rigged schedule.
This matters because it closes a specific door. If matchmaking were biased, it could explain why certain outcomes feel predictable — the system could be feeding you the same “rigged” fights over and over. But it’s not. Each fight is drawn from a pool of 52.6 million possible pairs, and the system almost never repeats. That means if outcomes still feel scripted, it can’t be because the schedule is steering them — it would have to be the fights themselves.
This is where the rigging theory would get its teeth — if the fights were scripted. If MUGEN fights are deterministic — same fighters in, same winner out, every time — then the matchmaking wouldn’t need to be biased. Knowing the matchup would be knowing the outcome, and anyone with the fight history could predict every result.
Of the 18,811 matchup pairs that have occurred two or more times [1]:
That 73% sounds high until you do the math: most repeat matchups are only two fights. If one fighter has a 70% win rate against the other, there’s a roughly 49% chance they sweep both meetings. The “same winner” rate is consistent with statistical expectation — not evidence of scripting.
This makes sense because MUGEN fights are not fully deterministic [2][3]. The engine’s AI makes real-time decisions based on distance, opponent action, and health. Different positioning at the start of a round leads to different decision cascades. The same two characters will not play the same fight twice. (For a deep dive into how this works, see How MUGEN Characters Work: AI, Moves, and Balance Explained.)
The coin is fair. The scheduling is unbiased. And the outcomes genuinely vary — which means there’s no hidden script anyone could exploit, including SaltyBet itself. We’ve now proven that the mechanics of the system are clean. But the most dangerous version of the rigging conspiracy was never about mechanics — it’s about incentives. Even a fair system could be corrupted at the output layer if someone is selectively tampering with results when it matters most. And “when it matters most” on SaltyBet means one thing: the big pots.
If SaltyBet wanted to drain the crowd’s Salty Bucks, the most efficient way wouldn’t be loading the coin or fixing the schedule — it would be engineering upsets on the biggest pots. Let the crowd pile money on the favorite, then flip the result. Maximum damage, minimum fingerprints.
This is the conspiracy theory people actually believe. And it makes a specific, testable prediction: crowd accuracy should drop as pot sizes increase. If the house is manufacturing upsets where the most currency is at stake, we’d see it in this table.
Here’s what actually happens [1]:
| Pot Size | Matches | Crowd Accuracy | Upset Rate |
|---|---|---|---|
| Under $1M | 58,290 | 67.1% | 32.9% |
| $1M–$5M | 151,254 | 61.4% | 38.6% |
| $5M–$10M | 268,639 | 70.6% | 29.4% |
| $10M–$25M | 93,425 | 70.5% | 29.5% |
| $25M+ | 13,890 | 72.7% | 27.3% |
The conspiracy predicts accuracy should drop at high pots. Instead, it rises. The bigger the pot, the less likely an upset — because bigger pots attract more bettors whose collective knowledge is more accurate.
The most upset-prone range? $1M–$5M at 38.6% — mid-stakes matches where the crowd is less certain, not high-profile ones being manipulated.
We can sharpen this further by looking at how lopsided the betting is [1]:
| Bet Ratio | Matches | Crowd Accuracy |
|---|---|---|
| Close (under 1.5:1) | 125,724 | 54.5% |
| Leaning (1.5–2:1) | 95,401 | 61.3% |
| Strong (2–5:1) | 278,479 | 71.6% |
| Heavy (5–10:1) | 71,309 | 81.8% |
| Extreme (10:1+) | 14,585 | 87.1% |
When the crowd overwhelmingly agrees on a winner (10:1 or more), that fighter wins 87.1% of the time. If the system punished confident bettors, this number would collapse. Instead, stronger consensus = higher accuracy — exactly what a fair system produces.
The mechanics are clean and the incentive theory is debunked. No side loading, no schedule bias, no scripted outcomes, no pot-targeted upsets. But every test so far has looked at a specific mechanism. What if there’s a form of manipulation we haven’t thought to test? The final check is the one that catches everything — because any manipulation, no matter how subtle, would leave a mark on the aggregate shape of the data.
Every form of rigging — side loading, schedule bias, outcome scripting, upset engineering — would ultimately distort the distribution of fighter win rates. It’s the fingerprint that catches everything, because it reflects the cumulative output of the entire system. If the curve looks unnatural, something is wrong. If it doesn’t, there’s nowhere left for manipulation to hide.
Here’s the actual distribution across 8,043 fighters with 20+ matches [1]:
| Win Rate | Fighters |
|---|---|
| 0–19% | 163 |
| 20–29% | 753 |
| 30–39% | 1,505 |
| 40–49% | 1,716 |
| 50–59% | 1,704 |
| 60–69% | 1,360 |
| 70–79% | 624 |
| 80%+ | 218 |
A near-perfect bell curve centered around 50%. The peak sits at 40–59% (3,420 fighters), tapering symmetrically in both directions. Some MUGEN characters have better AI, some have better moves, and the distribution reflects those real skill differences — not manipulation.
Five tests. Five clean results. No fingerprints anywhere in the system — not at the coin level, not at the schedule level, not in the outcomes, not in the money, and not in the aggregate shape of the data.
| Test | What Rigging Would Look Like | What the Data Shows | Result |
|---|---|---|---|
| Side balance | One side winning 55%+ | 49.25% red / 50.75% blue | Clean |
| Scheduling | Fighters appearing 10–100x more than others | 48.4% of roster in same bucket; max 294 appearances | Clean |
| Determinism | Same matchup = same winner every time | 27% of repeat matchups have split outcomes | Clean |
| Pot manipulation | Accuracy drops on big pots | Accuracy rises from 67.1% to 72.7% | Clean |
| Win rate shape | Flat, skewed, or clustered distribution | Textbook bell curve centered at 50% | Clean |
The data is clear, but the feeling is real. If the system is fair, why does it feel so unfair so often?
Confirmation bias. You remember the 3 upsets that cost you your bankroll. You forget the 7 favorites that won quietly. After a bad beat, you’re primed to notice the next one — and SaltyBet gives you a new match every few minutes.
The A-tier problem. A-tier makes up 41.9% of all matches but has the highest upset rate at 34.3% [1]. If you’re watching an extended session, most of what you’re seeing is A-tier — where the crowd is wrong one in three times. A-tier is where fighters are still being evaluated, still moving between tiers. The data is noisy, and so are your bets.
Three upsets in a row feels like a pattern. But with a 32% upset rate, three consecutive upsets happen roughly once every 30 matches — multiple times per day on a 24/7 stream. It’s not manipulation. It’s probability.
Sample size illusion. You’ve watched 20 fights tonight. The dataset is 585,000 deep. Your session is not statistically meaningful, even though it feels like it should be.
We wrote a full deep-dive on this: The Psychology of Betting on AI Fights. If you’ve ever felt like SaltyBet was personally targeting you, that article explains why — with academic research to back it up.
For more on how tiers and matchmaking actually work, check our SaltyBet FAQ. And for the broader dataset these conclusions are drawn from, see We Analyzed 570,000 SaltyBet Matches.
Is SaltyBet rigged?
No. Across 585,534 matches tracked by SaltyTrack, five independent tests show: near-perfect side balance (49.25% red, 50.75% blue), unbiased scheduling across 10,257 fighters, non-deterministic outcomes, higher crowd accuracy on bigger pots (not lower), and a natural bell curve of win rates [1].
Does side matter in SaltyBet?
Barely. Blue has a 1.5 percentage point edge overall, which is statistical noise for a 585,000+ match sample. No tier shows a side imbalance greater than 2 points [1].
Are MUGEN fights deterministic?
No. MUGEN’s AI makes real-time decisions based on opponent state, so the same two characters can produce different outcomes in different fights [2][3]. Of matchup pairs that occurred 2+ times, 27% had split results [1].
Why does SaltyBet feel rigged?
Confirmation bias and pattern-seeking. The overall upset rate is 32% — roughly one in three favorites loses. In A-tier (41.9% of all matches), upsets hit 34.3% [1]. Three upsets in a row feels like manipulation, but it happens multiple times per day by pure probability.
Does SaltyBet rig upsets when the pot is big?
No — the data shows the opposite. Crowd accuracy rises from 67.1% on small pots to 72.7% on large pots ($25M+). Bigger pots attract better collective knowledge, not more manipulation [1].
Want to bet smarter instead of wondering if it’s rigged? The SaltyTrack Chrome Extension gives you AI-powered predictions, win rates, and head-to-head records on every SaltyBet match. Our ML model beats the crowd’s 67.9% accuracy — and we’ve got the data to prove it.
SaltyBet uses virtual currency only. No real money is wagered or exchanged. SaltyTrack is not affiliated with SaltyBet.