The Numbers
Week of January 1-7, 2025
| Metric | Result |
|---|---|
| Total Signals | 47 |
| Wins | 30 |
| Losses | 17 |
| Win Rate | 63.8% |
| Total Edge Captured | $2,340 |
| Average Signal Edge | 7.2% |
| Best Trade | +$180 (Chicago, Jan 4) |
| Worst Trade | -$95 (Miami, Jan 6) |
City Breakdown
NYC: Consistent Performer 🏆
NYC continues to be our bread and butter. The dense network of nearby weather stations gives us excellent validation data.
- Signals: 14
- Win Rate: 71%
- Notes: Strong performance on the cold snap mid-week. HRRR nailed the timing.
Chicago: Cold Weather Edge
Chicago showed interesting patterns this week. When temperatures dropped below 20°F, our ensemble significantly outperformed the market.
- Signals: 12
- Win Rate: 67%
- Notes: Markets consistently underpriced extreme cold probability.
Austin: Learning Curve
Austin remains our most challenging market. The city's geography creates microclimates that models struggle with.
- Signals: 11
- Win Rate: 54%
- Notes: Need to adjust weights for Texas-specific patterns.
Miami: Tropical Troubles
Miami had our worst week. A persistent onshore flow confused all models.
- Signals: 10
- Win Rate: 50%
- Notes: Flat for the week. Considering reducing position sizing for Miami until we improve.
Key Insights
1. The Overnight Edge is Real
Our best signals came from acting on overnight model updates. Here's the pattern:
- 10 PM: GFS updates, market slowly adjusts
- 12 AM: ECMWF updates, most traders sleeping
- 6 AM: We act on fresh ensemble, market hasn't caught up
- 9 AM: Market reprices, we capture spread
Takeaway: Early birds get the +EV.
2. Cold > Hot for Edge
We found significantly more edge in cold temperature markets than warm ones:
| Condition | Average Edge | Win Rate |
|---|---|---|
| High < 32°F | 9.1% | 68% |
| High 32-50°F | 7.4% | 64% |
| High 50-70°F | 6.2% | 61% |
| High > 70°F | 4.8% | 58% |
Why? Hypothesis: Cold air masses are more predictable. Heating from solar radiation introduces more variance in warm conditions.
3. The Friday Effect
Friday markets showed unusual pricing inefficiency. Our theory: retail traders closing positions before the weekend, creating temporary mispricings.
- Friday average edge: 8.9%
- Other days average: 6.4%
We'll track this pattern going forward.
Biggest Win: Chicago January 4
Setup: Arctic blast moving through the Midwest. Models agreed temperature would plummet.
Market: Chicago High Temp Above 15°F priced at 72¢ Our Probability: 41% Edge: 31%
Trade: Sold (bet on Below 15°F)
Result: High temperature hit 12°F. Clean win.
Why it worked: Markets were anchored to the previous day's 28°F high. Our ensemble captured the cold front timing better.
Biggest Loss: Miami January 6
Setup: Onshore flow bringing moisture. Models disagreed on cloud cover impact.
Market: Miami High Temp Above 78°F priced at 55¢ Our Probability: 72% Edge: 17%
Trade: Bought (bet on Above 78°F)
Result: High reached only 76°F. Clouds persisted longer than modeled.
Lesson: When models disagree significantly, reduce position size or sit out. Our ensemble showed high uncertainty, but we traded anyway.
What We're Changing
Based on this week's performance, we're implementing three adjustments:
1. Miami Position Sizing
Reducing Miami positions by 25% until we improve tropical performance.
2. Model Disagreement Filter
Adding a new filter: if model spread > 4°F, reduce position size by 50%.
3. Weekend Liquidity
Increasing Friday position sizes to capture the weekend effect.
Signals to Watch Next Week
Looking ahead to January 8-14:
NYC: Major storm system possible mid-week. High volatility expected. We'll be watching for pricing inefficiencies as models converge.
Chicago: Polar vortex dip likely. If models confirm, expect strong signals on extreme cold markets.
Austin: Relatively quiet week. May sit out unless compelling setup emerges.
Miami: Continued onshore pattern. Staying cautious.
Community Callout
Shoutout to everyone who's been following the signals on our dashboard. Your feedback has been invaluable.
Quick question for you all: Would you prefer more frequent signal updates (hourly during active periods) or is the current cadence sufficient?
Drop your thoughts in our Discord or reply on X @arcticodds.
The Bottom Line
Week 1 was solid but not spectacular. Our edge is real, our methodology is working, but there's room to improve—especially in tropical markets.
We'll keep hunting. We'll keep learning. We'll keep showing our math.
+EV or die trying. 🐻❄️
See all signals in real-time on the Weather Dashboard.