It's All About Conversions: Why Chicago Midway Settled 88°F When the CLI Said 87°F

If you traded the KXHIGHCHI (Chicago Midway) market on June 9 and got whiplash, you're not alone. The thread was full of it.

The market settled 88–89 YES. The report a lot of people were staring at said 87. Both were "right." Here's the actual breakdown, because this exact trap is going to keep happening.

Trap #1: The report you're reading is preliminary

NWS issues the Daily Climate Report (the CLI — the thing Kalshi actually settles on) several times a day. The preliminary afternoon issuance on June 9 was stamped:

VALID TODAY AS OF 0400 PM LOCAL TIME.
  MAXIMUM   87   2:59 PM

That "AS OF 04:00 PM" is the whole ballgame. It's the max through 4 PM only. On the 9th, Midway kept climbing after that report — it hit its high in the 5–7 PM window. The preliminary report never saw it.

The final CLI (issued later, covering the full day) is what settles the market. On the 9th that was 88. So anyone anchoring to the 4 PM "87" was reading a half-finished box score and calling the game.

Takeaway: if the CLI says "VALID TODAY AS OF [some afternoon time]," it is not the settlement number. The high can still go up.

Trap #2: Whole-°C to °F rounding (the "conversions" thing)

This is the part the thread half-figured-out. ASOS reports temperature in whole degrees Celsius in the METAR body. So the 5-minute obs you see on the time series are already rounded:

• 31°C → 87.8°F, which displays/rounds to 88
• 30°C → 86.0°F

But the precise values live in the remarks. Two groups matter:

• the T-group (Tsnttt) = exact temp to 0.1°C, but only hourly
• the 6-hour MAX group (1snttt) = the precise 0.1°C high for the period — this captures sub-hourly peaks that the hourly readings miss

Here's Midway's 6:53 PM METAR:

KMDW 092353Z ... 31/22 ... RMK ... T03060222 10311 20289

Decode it:

• T0306 → current temp 30.6°C = 87.0°F
• 10311 → 6-hour MAX 31.1°C = 87.98°F → 88

So the hourly precise readings topped out at 87.0, but the 6-hour max group recorded a brief 31.1°C — and 87.98 rounds to 88. That's your settlement. The "88 on the board" people saw on the 5-min feed wasn't a fluke; the official precise record backed it up.

Why this combo is so nasty

The two traps stack:

1. The afternoon report gives you 87 and feels official.
2. The time series shows a wall of 88s that look like rounding noise you should discount.

So your instinct says "87, the 88s are conversion artifacts" — and you fade the 88–89 bracket.

Then the final CLI lands on 88 because a real 31.1°C got recorded after the preliminary cutoff, and you're holding the wrong side. It's a near-perfect setup to talk yourself out of the correct answer.

How to not get caught

• Check the "VALID AS OF" timestamp on any CLI before trusting the max. An afternoon stamp = preliminary.
• Watch the trend after the preliminary report. If it's still 86–88°F at 5–6 PM, the prelim high is stale.
• The 6-hour max group is your friend. A 1snttt reading above the hourly T-groups means a sub-hourly spike got logged — and the CLI uses that record, not the rounded body obs.
• Remember the rounding boundary. A string of "88"s on a 31°C day can settle 87 or 88, depending on whether the precise reading crossed 31.05°C (87.89°F). It's genuinely a coin-flip-looking 0.1°C, and the 6-hour max group is what breaks the tie.

Markets settle on the final NWS CLI — built from the precise 0.1°C record — not the preliminary report and not the rounded 5-minute obs you're eyeballing.

---

Full disclosure: our model is built for these markets, and we spent the day digging through KMDW's raw METARs to figure out exactly what happened. We figured the breakdown was worth sharing, since half the thread got burned by the same thing.