2026-07-15 · category AI · budget $1000
3 picks · $250 deployed of $1000 · 4 logged rejects
| Ticker | Market | Side | Conviction | Consensus vs market | Size | Per-model |
|---|---|---|---|---|---|---|
KXTOPMODEL-26JUL20-CLAUT |
What will be the top AI model this month? | NO @ 7c | HIGH | 96.0% vs 7c edge 89.0c · spread 0.0 · 1/1 agree |
1072 ($75) | · gpt-5.6-terra NO 96 |
| Thesis: The current public Arena page identifies newer Claude Opus 4.8 rather than the exact 4.6-thinking string. Rationale: The contract requires an exact named model, while the public leaderboard now displays newer successor models; that makes 4.6-thinking unlikely to regain the top spot by July 20. The decisive tail risk is that Kalshi did not specify the underlying leaderboard in the primary rule, so the semantic source is disputable. | ||||||
KXTOPMODEL-26JUL31-CLAUT |
What will be the top AI model this month? | NO @ 23c | HIGH | 90.0% vs 23c edge 67.0c · spread 0.0 · 1/1 agree |
325 ($75) | · gpt-5.6-terra NO 90 |
| Thesis: The public Arena listing shows newer Claude Opus 4.8 rather than exact Claude Opus 4.6 Thinking. Rationale: A successor model is publicly listed, leaving the exact older model an unlikely top-ranked candidate at the snapshot date. The clean tail risk is an undisclosed intended leaderboard or a ranking-specific methodology that differs from Arena's public default. | ||||||
KXTECHRANKLISTAICODE-26JUL20-CHAT |
Top Coding AI this week? | NO @ 35c | HIGH | 70.0% vs 35c edge 35.0c · spread 0.0 · 1/1 agree |
285 ($100) | · gpt-5.6-terra NO 70 |
| Thesis: Code Arena's current #1 is specifically gpt-5.6-sol-xhigh, not a leaderboard entry named ChatGPT. Rationale: Arena's July 13 public Code leaderboard names gpt-5.6-sol-xhigh (codex-harness) as rank 1, so mapping the generic product label ChatGPT to a resolution candidate is not clean. A product-level Kalshi convention could nonetheless map OpenAI's entry to ChatGPT, which is the material tail risk. | ||||||
| Ticker | Market | Why rejected |
|---|---|---|
KXTECHRANKLISTAICODE-26JUL20-CLAU | Top Coding AI this week? | no edge after shrinkage (-4.0c < 2.0c) |
KXRTX5090WS-26AUG07-0.750 | Will the NVIDIA RTX 5090 compute per hour price be above $0.75 at 4 PM ET on Aug 07? | no edge after shrinkage (0.0c < 2.0c) |
KXGEMINI-GEMI35P-26JUL31 | Will Google release Gemini 3.5 Pro before Jul 31, 2026? | no edge after shrinkage (-1.0c < 2.0c) |
KXGEMINI-GEMI35P-26JUL24 | Will Google release Gemini 3.5 Pro before Jul 24, 2026? | no edge after shrinkage (-1.0c < 2.0c) |
Probabilities are subjective; contracts can resolve to zero. Not financial advice.