I’ve tinkered with Claude Code over the last few weeks and I figured I’d throw the recent How many 6.5 or above earthquakes by January 25? market at it. I’m not skilled enough statistically to work out what the implied odds are based on recent data, and the opportunity cost of learning all that is quite high (I could be writing blog posts on AI prompts instead!). That’s why I’ve left it in the hands of Claude Code to tell me what to do. Will it be right? Will it be wrong? Is there a positive EV position here? Let’s find out!
Correcting Claude
The first issue I noticed was that Claude decided to look at historical data, but it focused only on the date range of the 19th to the 25th – which is what the market covers. Surely earthquakes aren’t more likely during certain times of the year, right? I did a quick Google and they seem random: nothing to suggest they’re more likely to occur during the day, night, or any particular month/year.
After I told Claude this, it ran some numbers and did the statistical analysis for me: very much appreciated for someone who’s really quite dumb when it comes to crunching data. Basically, it took the number of earthquakes in the last decade, calculated how many that equals per hour, then put that up against the 168 hours the market will run for. Note that fair yes/fair no indicate what those markets are statistically likely to resolve as. If the market price is lower than the fair price, it’s positive EV.
| Outcome | Fair YES | Fair NO | Market YES | Market NO | Best Trade |
|---|---|---|---|---|---|
| 0 earthquakes | 44.97% | 55.03% | 51.0% | 49.0% | BUY NO |
| 1 earthquake | 35.94% | 64.06% | 34.0% | 66.0% | – |
| 2 earthquakes | 14.36% | 85.64% | 14.0% | 86.0% | – |
| 3 earthquakes | 3.83% | 96.17% | 4.0% | 96.0% | – |
| 4 earthquakes | 0.76% | 99.24% | 3.0% | 97.0% | BUY NO |
| 5 earthquakes | 0.12% | 99.88% | 1.0% | 99.0% | BUY NO |
| >5 earthquakes | 0.02% | 99.98% | 2.0% | 98.0% | BUY NO |
The market is quite efficient, but this ignores the spread and for these smaller markets, there are margins we can eek out. The 0 earthquake outcome resolving “No” is a nice EV play, imo: a 6% edge over the fair probability is pure winning long-term (we just have to make sure we don’t get blown up). I guess one of the nice things about the market price being semi-close to the fair probability is that we can take either side of the trade with orders 3-5%~ off the fair probability. We can be the casino!
This market still has some time before it gets started, so I’ve loaded up a fresh wallet with a bunch of orders either side of all markets and I guess we’ll see how things go. Stay tuned!
