Esteemed family! The second half of April is in the books. 6,560 matches, $37.2 billion in Salty Bucks wagered, and one of the biggest upsets we’ve ever logged. Buckle up, friends — late April was wild. Three of the most chaotic days of the month all landed in this window, Sephiroth quietly went 8-0 in S-tier, and one MUGEN deep cut named Dies took down a beloved BlazBlue cat girl at roughly thirty-to-one odds. Let’s get into it!
The headline. Dies over Taokaka was a ~30:1 upset in S-tier — bigger than anything in our April first half or March second half reports. The crowd put 94% of its money on Taokaka. Dies won anyway. Welcome to SaltyBet.
| Stat | April 16–30, 2026 |
|---|---|
| Total matches | 6,560 |
| Unique fighters seen | 6,061 |
| Total Salty Bucks wagered | $37.2 billion |
| Average pot | $5.7 million |
| Red wins | 3,202 (48.8%) |
| Blue wins | 3,358 (51.2%) |
Combined with the first half (6,172 matches), April 2026 totaled 13,143 matches and roughly $72 billion in total wagering. Healthy month — slightly above the ~12,500/month average we’ve been tracking.
Red vs Blue continues its eternal non-rivalry: 48.8% vs 51.2%. Blue edges as always. We’ve run out of takes on this one. There is no Red/Blue advantage, and we’ll be saying that until the heat death of the SaltyBet universe.
6,061 unique fighters across 6,560 matches means the average fighter appeared just 1.08 times in two weeks. The roster is enormous and the matchmaker keeps spreading it thin.
| Tier | Matches | Avg Pot |
|---|---|---|
| A | 2,754 (42.0%) | $5.1M |
| S | 2,067 (31.5%) | $6.4M |
| B | 1,412 (21.5%) | $5.1M |
| X | 183 (2.8%) | $8.6M |
| P | 144 (2.2%) | $8.8M |
The usual distribution — A and S tier together account for 73.5% of all matches. P-tier and X-tier are where the big-money pots live, both averaging around $8.6–8.8M per match. P-tier only ran 144 matches in 16 days, but the average pot was $8.8M. The crowd bets big when they think they know who they’re watching. They don’t always.
For a refresher on what each tier means for your betting, we wrote the explainer back when we built the Explorer page.
These fighters went the entire second half of April without dropping a single match:
| Fighter | Record | Tier |
|---|---|---|
| Shimo EX3 | 9-0 | B |
| Sabretoothu | 8-0 | S |
| Sephiroth EX9 | 8-0 | A |
| Small chibi-moon | 7-0 | A |
| Subterranean rose | 7-0 | S |
| Arrange sion | 6-0 | A |
| Bobby drake | 6-0 | S |
| Geki de oso | 6-0 | B |
Shimo EX3 leads the pack at 9-0 in B-tier. Nine wins, zero losses, no real franchise tail to fall back on — just a MUGEN deep cut printing Salty Bucks for two weeks straight. We always love when a no-name takes the top spot. You’re a legend in our book, Shimo!
Sephiroth EX9 went 8-0 in A-tier. Final Fantasy VII’s iconic villain doing what Final Fantasy VII’s iconic villain does — winning. Two perfect SaltyBet runs in a row, depending on which Sephiroth variant you’re tracking, the EX9 build is the one to watch. One Winged Angel into eight straight wins. Check.
Bobby Drake (Iceman from X-Men!) at 6-0 in S-tier is the kind of name-recognition pick that the crowd should have caught — but didn’t. The Marvel deep cuts continue to overperform expectations. Subterranean rose joining the S-tier perfects at 6-0 is the other half of that story: two unfamiliar names, both shutting out S-tier competition.
Worth noting: of the eight perfects, three landed in S-tier. Going undefeated in S-tier means you’re beating the strongest pool of fighters SaltyBet routinely runs. These aren’t soft schedules. Salute.
Not everyone had a good two weeks:
| Fighter | Record | Tier |
|---|---|---|
| Buront | 0-6 | S |
| Tam tam | 0-6 | A |
| Akane yatogi | 0-6 | S |
| Gadis | 0-6 | S |
| Jecht EX | 0-6 | S |
| Kthanid | 0-6 | S |
Five of the six worst performers were S-tier. That’s a striking number — usually the struggle bus is at least half-populated by lower tiers eating losses against tougher opponents. This window, the worst slumps all happened in the upper tiers, which means a chunk of S-tier fighters had genuinely bad two weeks against their actual peer group.
Jecht EX (FFX boss) and Tam tam (Samurai Shodown) both went 0-6. Recognizable franchise names, terrible runs. The crowd had to keep adjusting on the fly — backing a name you trust, watching that name lose, and then doing it again three days later. As we covered in why some MUGEN characters always win, the franchise label on the select screen doesn’t predict the in-arena results. Big nerd-name energy, low actual SaltyBet output.
These are the matches where the crowd put massive money on the favorite — and lost:
| Winner | Defeated | Bet Ratio | Tier |
|---|---|---|---|
| Dies | Taokaka | ~30:1 | S |
| Debonea | Rassius fomalhaut | ~3:1 | A |
| Tengu sw | Tetsuo k9999 | ~2.5:1 | S |
| Kotoe | Ai momoyama | ~2.4:1 | A |
| Pepe EX3 | Zerozuchi boss | ~1.6:1 | S |
Dies over Taokaka is the upset of the half — and one of the biggest single upsets we’ve logged in any biweekly window. The crowd put roughly 94% of the pot on Taokaka (BlazBlue’s cat-girl with a million memes attached) and 6% on Dies (Dies Irae’s protagonist, considerably less well-known). Dies won anyway. At ~30:1, this is bigger than the 21.8x Clone zero-kofm upset from early April and bigger than the 24.3:1 C-xion upset from March’s second half. New benchmark.
The other four upsets in this list are modest by comparison. Three sit in the 2–3:1 range — meaningful crowd misreads but not the kind of math that makes you spit out your drink. Pepe EX3 over Zerozuchi boss at 1.6:1 barely qualifies as an upset, but it was still in the top five of the period. Translation: outside the Dies-Taokaka outlier, late April’s crowd was actually pretty close to right on the high-stakes calls.
How did SaltyTrack’s ML model perform on these 6,560 matches?
| Confidence | Correct | Total | Accuracy |
|---|---|---|---|
| Lock | 8 | 12 | 66.7% |
| High | 1,237 | 1,411 | 87.7% |
| Medium | 1,993 | 2,783 | 71.6% |
| Low | 1,312 | 2,354 | 55.7% |
| Overall | 4,550 | 6,560 | 69.4% |
High confidence held strong at 87.7% — basically right where we want it. That’s roughly 1,400 matches where the model said “yeah, this one” and was right almost 9 out of 10 times. That’s where the edge lives. If you’re betting with the Chrome extension and you see a High call, the data says lean in.
Lock dipped to 66.7% — but that’s 8 correct out of 12 attempts, which is small-sample noise more than a real regression. Twelve matches in two weeks isn’t enough to draw a trend line on. We’ll watch it.
Medium climbed to 71.6%, Low stayed at ~56%. Medium getting closer to the 75% target line is the quiet win here. As always, we’re transparent: Low confidence is a coin-flip-plus, don’t anchor heavy bets to it.
Crowd accuracy this period: data unavailable. A reporting gap on our end — we’ll fix it for the May 15 report. If you want to understand why the model and the crowd usually trade leads of 3–4 points, we wrote the deep dive on prediction methodology.
Late April was chaotic. Three of the period’s wildest days happened back-to-back-to-back-to-back:
| Day | Chaos Score | Upsets | High-Conf Wrong |
|---|---|---|---|
| April 23 | 276 | 115/410 (28.0%) | 18 |
| April 18 | 262 | 147/414 (35.5%) | 16 |
| April 25 | 251 | 139/409 (34.0%) | 15 |
April 23 hit 276 with 18 high-confidence prediction failures in a single day. That’s the most high-conf misses we logged in the entire second half. The model called these matches “we know this one” and got it wrong eighteen separate times. When that happens, you know the field itself was unusually unpredictable.
The Apr 18 → Apr 23 → Apr 25 stretch is the receipt — three ≥250 chaos days inside a single week is the kind of cluster that catches our attention. It also lines up with the same window where we logged the B→P tier migration cluster and the E-terry 71113 perfect streak — late April had a meta seam. Tiers were re-shuffling, confident calls were missing, and nominal underdogs were taking over. We don’t always know why a chaos cluster forms, but we know one when we see one.
Combining both halves:
| Stat | Full Month |
|---|---|
| Total matches | 13,143 |
| Total wagered | ~$72.0 billion |
| Crowd accuracy | (data gap — TBD) |
| SaltyTrack accuracy | ~69.4% (period average) |
| Most chaotic day | April 23 (score: 276) |
| Undefeated highlight | Shimo EX3 (9-0, B-tier) |
| Biggest upset | Dies over Taokaka (~30:1, S-tier) |
April was a slightly bigger month than March (13,143 vs 11,663 matches) and brought the biggest single-window upset we’ve logged this year. Worth watching whether the late-April chaos pattern persists into early May or unwinds back to baseline. That’s tomorrow’s data; we’ll have it for the May 15 mid-month report.
May’s first-half meta report drops in two weeks. In the meantime:
See these stats live during matches. The SaltyTrack Chrome Extension shows you win rates, head-to-head records, and AI predictions right on saltybet.com. Free, no account needed.
SaltyBet uses virtual currency only. No real money is wagered or exchanged. SaltyTrack is not affiliated with SaltyBet.