Ethereum has already fallen to about $1,505 this year, its June low. Kalshi's floor ladder asks whether it goes lower before New Year's, and traders lean yes: as of August 12, a new low below $1,500 traded at 52 cents.
Six AI models read the same chart without seeing a quote and came back at 41 percent. They sit under the market on every rung, and furthest under on the crash levels, where they price the move at less than half of what traders pay. Both sides agree a retest of the June low is the live question. They split on how much downside insurance is worth.
The Board: Kalshi Versus The Models
Each rung pays YES if Ethereum's CF Real-Time Index trades below the level at any point in 2026, with settlement on January 1, 2027. Market prices are Kalshi marks as of August 12, 2026. The AI column is the equal-weight blend of six models that never saw them.
| 2026 Outcome | Move needed from spot near $1,880 | Market YES | AI blend |
|---|---|---|---|
| New Low Below $1,500 | about -20% | 52¢ | 41% |
| New Low Below $1,250 | about -34% | 35¢ | 14% |
| New Low Below $1,000 | about -47% | 16¢ | 5% |
| New Low Below $750 | about -60% | 8¢ | 2% |
Model estimates generated 2026-08-12, blended after a revision round. These are model estimates, not predictions of fact and not financial advice. Kalshi prices reflect real traders' money and are the better guide to how a market will resolve.
No rung has been touched. The June low stopped $5 above the top rung, so the first contract asks whether ETH completes a retest of a floor it already tapped. From a spot near $1,880, that is a drop of about 20 percent.
The deeper rungs are a different animal. Ethereum has not traded below $1,000 since 2023, so those contracts price a full bear-market breakdown rather than a nick below an existing low.
That is where the models pull away. On the first rung the panel is 11 points under the market. On the $1,250 rung the market pays 35 cents and the panel says 14 percent, and that ratio holds through the bottom two rungs.
Why The Market Pays More Than The Models
Pure volatility math on the quiet summer tape puts a touch below $1,500 near 27 percent. The models lift that to 41 for crypto's habit of moving violently once a selloff starts, and for a floor that sits so close it is unusually easy to complete. The market's 52 cents adds a further premium. This book is liquid, so that premium is real money, and the panel reads it as the price of protection rather than a probability. Traders are paying up to hedge an Ethereum crash, and the price-blind models think the crash itself is less likely than that price implies.
The Year That Built The Ladder
Per Coinbase daily candles, ETH ran to its year high of $3,405 on January 14, then slid through the spring into the June low, a peak-to-trough drop of about 56 percent.
Since June it has drifted back to roughly $1,880, where it sat on August 12. Summer has been a tight, low-energy range between about $1,800 and $1,980, on the quietest volatility of the year. That is what makes the floor board live: the market has watched ETH lose more than half its value once already, and it is pricing a real chance the slide is not finished.
Why A Retest Is Live And A Crash Is A Tail
The bear case is the chart. ETH has already made a move four times the size of the dip a retest needs, and the June low sits a rounding error above the first rung. GLM leaned on that geometry: with the low "sitting $5 above the top rung and 140 days left in a bear year," it read a completed retest as close to even money. Compressed summer volatility, on this view, is the quiet range that often precedes the next leg down.
The case against the deeper rungs is that ETH held $1,000 through the entire decline. Opus framed those contracts as regime breaks: "Below $1,000 requires a near-halving and below $750 a 2023-era 60 percent collapse." Fable added that a quiet range "is not itself evidence of imminent breakdown." Each leg below the June low needs a catalyst the current tape does not contain, which is why the panel's ladder decays so much faster than the market's.
No seat priced the crash rungs to zero. Sonnet kept weight there because "vol spikes during the very crash scenarios that drive deep touches, so constant-vol GBM understates tail risk." The kind of forced-selling cascade crypto produces every few years, an exchange failure or a stablecoin break, is all the $1,000 rung needs, and the panel leaves it on the board at roughly one in twenty.
Where The Panel Changed Its Mind
The panel runs two rounds. In the second, each model reads the others' anonymized numbers and reasoning and may revise. On this board the round pulled both outliers toward the center.
- Claude Opus moved up, 30 percent to 42 on the first rung. It opened as the lowest seat on textbook diffusion math and then conceded the geometry: "The $5 proximity of the June low means the barrier is essentially already at the tested floor, and a mild bear-year negative drift plus vol clustering make a marginal new low more likely than symmetric diffusion implies."
- DeepSeek moved down, 48 percent to 42. The most bearish opening seat trimmed its first rung and its middle rungs to sit inside the consensus band, describing the summer as "a low-volatility consolidation near $1,880 that leaves a fragile floor just above the $1,500 rung."
- Claude Fable held at 40 percent. "The peer set is tightly clustered and their shared anchor matches mine; no argument changed the core barrier math of a roughly 1.1-sigma touch with fat-tail widening below." Fable's hold is why the final blend is a real consensus rather than two outliers averaging out.
After the round the seats were tight. Claude Sonnet was lowest at 37 percent on the first rung; Claude Opus, GLM 5.2 and DeepSeek were highest at 42. Kimi K3 sat at 40 with Fable, and every seat was within two points of the others below the first rung. Sonnet stayed lowest because its own math was "a legitimate floor," and it named the open question for the whole board: "whether the low-vol summer consolidation is a real floor or pre-breakdown compression." Two seats, an OpenAI model and a Google model, were unavailable at run time, and we report that rather than backfill it.
Model estimates generated 2026-08-12. These are model estimates, not predictions of fact and not financial advice. Models are frequently wrong; the market price reflects real traders' money.
What Would Move The Numbers
- A daily close below the $1,505 June low. The cleanest trigger. Fable said a confirmed break would push the first rung above 50 percent. It would also, in Fable's words, "roughly double the $1,250 and $1,000 rungs."
- Losing the summer range floor near $1,800. Sonnet flagged this as the momentum tell, along with the 50-day average just above it. A decisive break on rising volume raises the odds of undercutting $1,500 well before year-end.
- A macro or liquidity shock, or a spot-ETH ETF outflow cascade. The seats agreed the deep rungs are gated on a systemic event. This is the one development that lifts the $1,000 and $750 contracts off their low-single-digit floors.
- A sustained hold above $1,800 with volatility staying compressed, or an ETF-inflow surge back toward $2,400. Every week ETH holds its range rebuilds the cushion above the first rung. A move toward $2,400 pulls the whole ladder down.
Where You Can Trade It
Kalshi is a CFTC-regulated exchange, its contracts are 18+, and state availability changes often enough that a list frozen into an article would go stale. Check eligibility on the platform, then read our Kalshi legality guide for the state-by-state detail. For how these index-settled crypto contracts are built, our guide to Kalshi's Bitcoin and Ethereum price markets covers the mechanics this page assumes.
More on this: How High Will Ethereum Get In 2026? Kalshi's Year-High Ladder · How Low Will Bitcoin Get In August? The Floor It Hit Twice · How High Will Bitcoin Get This Month? Kalshi's Monthly High Ladder · Will Bitcoin Top $200,000 By 2027? Kalshi Odds Vs The Models
The Bottom Line
The number to watch is $1,505. Break it and the bears have their retest, and by the panel's own account the deeper rungs roughly double. Hold it and the models were right that the market paid too much for the crash. Every number here gets graded when the market settles on January 1, 2027, on the full graded scoreboard.
Everything here is analysis for information, not a recommendation to trade. These are model estimates, not predictions of fact and not financial advice. Kalshi event contracts trade on a CFTC-regulated exchange, you must be 18 or older and in an eligible state to participate, and crypto prices move fast, so read the August 12 marks as a snapshot, not a live quote.
Prices on this page are Kalshi's book. If you also trade on Polymarket, code OS4 gets new users a $50 trading bonus on a $10 deposit — affiliate link; terms as stated by Polymarket; 18+, availability varies by state.
