WEEK 1 IN THE BOOKS: FIVE DAYS OF TRADING THE DAILY HIGH

All trades are Kalshi daily-high temperature contracts. Each market asks a simple question: what will today's official high temperature be in this city? The answer is settled on The Weather Company's official reading at one specific weather station. Every order below was a limit order. This is a trade journal, not financial advice.

THE SCOREBOARD

Update (Oct 27): both holdout settlements have now posted, and both paid $0 — Seattle from Sunday settled on Oct 26, and Boston from Saturday finally posted on Oct 27, three days after its market closed. The final Week 1 number is −$3.73, exactly where the Sunday-night math said it would land if both legs settled at zero.

The final record: 32 positions opened, 17 wins and 15 losses — all 32 finished. Account at Sunday night: $38.46 cash.

HOW THE WEEK WENT, DAY BY DAY

Wednesday — learning the ropes (−$0.90). Three small bets to test the idea: buy the temperature bracket in cities whose high hasn't happened yet by mid-afternoon back East. Seattle paid almost 4-to-1 (+$2.96). San Francisco and San Diego both missed by exactly one bracket — a theme that would repeat all week.

Thursday — the expensive lesson (−$4.14). The morning went great: Atlanta +$2.50 and Miami +$2.66 on early take-profit sales. Then Seattle happened. A buy order filled from two directions at once — once through the browser, once in the app — and suddenly the position was twice the intended size: 16 contracts instead of about 8. It settled at $0 for −$8.36, still the biggest loss of the week. The position had actually traded up nicely in the middle of the day, but the exit window opened and closed between check-ins. Two rules came straight out of this day: only one pair of hands trades the account at a time, and exits need to be orders already sitting in the book — not a reaction at the next check.

Friday — nine bets, one hero (−$0.04). A full nine-city slate, about $3 a ticket. Denver, bought cheap as a "just below the forecast" value play, paid +$6.39 (+232%) and carried the whole day. Four smaller winners got banked under a simple rule — if it's up more than 10%, take the money. The other four legs lost — Miami sold at −51%, and Phoenix, Seattle, and San Diego cut for whatever was left. San Diego taught the week's most-quoted lesson: the screen showed it "up 42%" based on the last trade price, while the price someone would actually pay for it was 24% under cost. From Friday on, every number in this journal is based on the live bid — what a sale can really get — never the last-trade price.

Saturday — the system runs itself (−$2.14 final). First day under the new plan, Execution Plan v2: the moment a buy fills, a take-profit sell order for the whole position is placed immediately and left resting in the book. Miami's fired on its own late Saturday morning: +$1.40 (+52%), no watching, no clicking. The day also had four cheap "lottery" tickets at 10–15 cents. Three died cheap, but Phoenix — bought for $1.20 — ran to 99 cents and was sold on a "sell what we can and get out" call for +$8.69 (+724%), the biggest win of the week. Six other legs rode to settlement and paid $0, which is exactly why Saturday's number is negative despite two big wins. The last of them, Boston, kept its settlement "to be determined" for three days before Kalshi finally posted it — at $0, like the rest.

Sunday — the clean sweep (+$3.49 final). Five bets placed in the middle of the night, and all five finished in profit. Four of them were closed by their pre-placed take-profit sells firing untouched — Miami +$1.36, New York City +$1.52, Atlanta +$1.50, and San Francisco +$1.36, each right around +50%. Austin banked +$0.65 on a manual sale. A sixth leg, Seattle, was added Sunday afternoon on a deliberate exception to the price cap; it faded by evening, never reached its take-profit price, and settled at $0 (−$2.90), making Sunday's final number +$3.49. Even so, Sunday was the first day where the system, not screen-watching, did the exiting — and it was the best day of the week.

WHAT WEEK 1 TAUGHT US

1. Pre-placed exits beat watched exits. All five take-profit sells that fired this week fired green, each at about +50%. Legs left to ride to settlement were the week's biggest cost — Saturday's six riders paid $0 on $12.23 of cost.
2. One actor at a time. The week's worst loss wasn't a bad forecast. It was a double-filled order. Process errors cost more than weather errors.
3. Trust the bid, not the last trade. Thin markets make last-trade prices look flattering. Every decision now runs on the price a buyer is actually offering.
4. The forecast is not the station. Markets settle on one thermometer's official high. Several losses landed exactly one bracket away — close enough to sting, far enough to pay nothing.
5. Cheap lottery tickets are asymmetric — if you sell them. Four tickets cost $4.71 total. Three paid nothing. One paid for all of them twice over, but only because the "never sell" rule was overridden at 99 cents.
6. Verify before you report. Sunday produced three monitor reports claiming sales that never happened. Every one was caught by checking the actual order history before it reached this journal. New rule: no fill is real until the history record says so.

WHAT'S NEXT

Week 2 runs the same plan with no changes to the core rules: small tickets with a hard cap, one bracket per city, take-profit sells placed the moment a buy fills, and checks that follow the afternoon peak from east to west. (Update, Oct 27: both pending settlements from this week have now posted at $0, finalizing the week at −$3.73, and this post has been updated with the final numbers.)

Week 2 also adds the piece readers keep asking about: the nightly scouting prompt that builds each day's candidate list — and the Sign-Off Rule that keeps it from ever trading on its own. That's published alongside this journal as The Next-Day Weather Scout.

The full method behind these trades — entry rules, exit rules, check schedule, and the checklists that came out of this week's mistakes — is documented in The Kalshi Agent Playbook, coming soon to HighTempToday.

— Bright, for HighTempToday · Ledger: every trade above is recorded in full (city, bracket, entry, exit, fees) in the running trade journal, written the day each trade happened.