Kalshi settled 73 bitcoin 15-minute windows in the session that ran from 6:00 AM Pacific on September 20, 2026 through 12:00 AM Pacific on September 21. Bitcoin closed above the price to beat in 40 of those windows and below it in 33. Every window below carries its Kalshi ticker, the price it had to beat, the value it settled at, and the YES price in cents at the open, the low, the high and the close.
The day in counts
Seven numbers that describe the session before any strategy is laid over it.
Windows settled
736:00 AM PT to 12:00 AM PT
Closed up
40Settled above the price to beat
Closed down
33Settled at or below it
Total volume
195,456,498As Kalshi reports it across the session
Biggest single-window move
+$490.655:15 PM PT window: $81,158.49 to beat, settled $81,649.14
First target
$80,334.906:00 AM PT window
Last target
$81,456.9212:00 AM PT window
Betting one side all day, $100 base, doubling after a loss
Two mechanical runs over the same 73 windows. Each one buys a single side at 50 cents whenever 50 cents
actually traded inside the window, commits $100.00, doubles after a losing window and resets to
$100.00 after a winning one. The stake is the money that leaves the account: Kalshi's taker fee is
inside it, not billed on top, so a $100.00 bet costs $100.00 and buys the contracts
that is left over after the fee. A losing window loses exactly the stake.
8 windows never printed 50 cents, so both runs sat those out and carried the stake.
Always up
Buys YES in every window
Started
6:00 AM PTSeptember 20, 2026
Final
+$2,911.10
Bankroll needed
$800.00largest at the 10:45 AM PT window
Peak
+$2,911.1012:00 AM PT
Worst point
−$6.786:30 AM PT
Largest single stake
$800.00
Longest losing run
3 windows
Windows skipped
8 of 7350 cents never printed inside the window
Always down
Buys NO in every window
Started
6:00 AM PTSeptember 20, 2026
Final
+$1,817.46
Bankroll needed
$12,613.54largest at the 8:30 AM PT window
Peak
+$1,917.4611:45 PM PT
Worst point
−$6,213.548:15 AM PT
Largest single stake
$6,400.00
Longest losing run
6 windows
Windows skipped
8 of 7350 cents never printed inside the window
Always up finished the day ahead, +$2,911.10 against +$1,817.46 for the always down run. That is the whole of what the day says.
Doubling after a loss only survives a short losing run. The always-down run lost 6 windows in a row, and the next window in that run needed $6,400.00 on a $100.00 base. A run of 10 would have needed $102,400.00 on one window to chase the same $100.00. Both sides finished positive on this particular day because September 20 never produced a long one-sided run, not because doubling works. A day that does produce one takes the stake past what most people have, and past Kalshi's book, before the run ends. This is one day of settled windows. It is not a strategy, and it is not advice.
Both runs, side by side
Press play and the two runs walk the session together, one window at a time, from
6:00 AM Pacific. The left panel is the always-up run, the right is always-down, and
both are placing their bets into the same 73 windows. Both accounts start the day
with $20,000, the same number, so the two panels can be read
against each other. Each shows the running total and what is actually in the account. The
second one is the one that dies and recovers.
Why a window says sat out. A run that pays a flat 50 cents can only enter
a window where 50 cents traded. In 8 of the 73 windows it never did: the
price opened on one side of 50 and stayed there for the full fifteen minutes, so there was
nothing to buy at that price. Those windows are sat out and the stake carries to the next one,
which is not free, because the stake carried is a doubled one. Step onto a sat-out window and
the panel says which way that window's price missed. Switch to Buy at the open
and nothing is sat out, because every window has an opening price.
These windows do not open at 50 cents. Across the 73 windows the YES price opened at exactly 50 cents 10 times. The median open was 51 cents and the range ran from 33 to 64. 27 windows opened within a cent of 50 and 52 within five.
So the cards above are not taking the open. They pay a flat 50 cents, and only in the 65 windows where 50 cents traded at some point inside the window; the other 8 are sat out. That is an assumption, and it is doing work: it assumes a fill at a coin-flip price whenever a coin-flip price existed anywhere in those fifteen minutes.
Switch the module to Buy at the open and the same day is re-run paying each window's actual opening price, which makes every window playable. Always up finishes +$3,314.21 instead of +$2,911.10. The interesting one is always down: it finishes about the same, +$1,839.82 against +$1,817.46, but it demands $6,250.18 instead of $12,613.54, because it never sits a window out and its longest losing run is 5 rather than 6. Neither number is a recommendation. Both are what the settled prices say.
Every figure was settled when the page was built. Nothing is simulated in your browser,
and with scripting off the full ledger below is still the whole record.
This is a model, not a record of trades anyone placed. A 15-minute
market does not open at 50 cents just because two outcomes exist: on this day the YES price opened
at exactly 50 cents in 10 of the 73 windows, and the run buys at 50 cents anyway wherever 50
cents traded. Changing only that one assumption, and nothing else, moves the always-up result by
$403.11 and takes the always-down deposit from $12,613.54 to
$6,250.18. Real fills, the spread and the size of the book would move all of it
again. Treat the gap between the two modes as the size of the guess, not as a margin of error.
73 of 73 windows · started 6:00 AM PT
12:00 AM PT close
· $81,456.92 to beat
· settled $81,631.08
· opened 33.0¢ on YES
· ▲ Up
Always up
Max bankroll needed
$800.00at 2:30 PM PT
Max drawdown
$700.00at 10:30 AM PT
Longest losing run
3windows
Always down
Max bankroll needed
$6,250.18at 10:00 AM PT
Max drawdown
$3,123.12at 9:45 AM PT
Longest losing run
5windows
Always upAlways downBreak evenBitcoin, $80,447 to $81,795
Running total, window by window, 6:00 AM PT to midnight. The two runs share one
scale, so the dip on one is the same size as the dip on the other. Bitcoin is in the band
underneath, on its own scale, because a price near $81,795 and a profit of a
few hundred dollars are not the same kind of number and putting them on one axis would make the
smaller one look like whatever the drawing needed. The two share the clock and nothing else. Move
your pointer across to read any window; the grid is half-hourly. The table below carries every
figure.
Oldest first, so the two runs read in the order they were placed. Paid is the price
that window was entered at. Stake is the money that left the account, Kalshi's taker fee included, so a
losing window loses exactly that and nothing more.
73 of 73
· 12:00 AM PT
· BTC $81,631.08
· ▲ Up
Always up
Done
+$3,314.21$23,314.21
running totalof $20,000 started with
33.0¢$200.00incl $8.96 fee+$378.91
Always down
Done
+$1,839.82$21,839.82
running totalof $20,000 started with
67.0¢$100.00incl $2.26 fee−$100.00
Why some windows say sat out
A run that insists on paying 50 cents can only enter a window where
50 cents actually traded. In 8 of the 73 windows it never did. The
price opened on one side of 50 and stayed there for the full fifteen minutes, so
there was no 50-cent contract to buy. Those windows are sat out.
Sitting out is not a rule anyone chose. It is what falls out of fixing the entry
price, and it is not free: the stake that was going to be placed carries to the next window, and if
the run was in a losing streak the stake that carries is a doubled one. That is most of why the
always-down run needed $13,000.00-odd at 50 cents and about half that at the open, where nothing
is ever sat out.
The always-up run crossed these marks during the session. A stop rule is a
number you pick before the day starts, not one you find afterwards.
+$500reached at 8:00 AM PT
+$1,000reached at 9:30 AM PT
+$2,000reached at 5:30 PM PT
The extremes: written off, then paid
In 11 windows the side that eventually won traded under 10 cents at some point inside the
window. A contract at 4 cents is the market saying this one is finished. These are the windows where it was
not. Nine of the 11 printed that low with four minutes or less on the clock.
Each card shows three real prints from that window: where the winning side opened, its cheapest print and
the minute it happened, and where it closed. Nothing is drawn between them, because nothing between them is in
the data. A print that lasted one minute is not a price anyone was promised.
6:45 AM▲ Up won
x27.8from 3.6¢ with 2 minutes left
Bitcoin finished $19.88 above the target. Target $80,487.24, settled $80,507.12.
In 22 of the 73 windows, the side that eventually won traded under 25 cents at some point inside
the window. In 11 of them it traded under 10 cents. The median cheapest print by the winning side was
36.0 cents. The deepest was 3.6¢, and it happened twice: the 6:45 AM window that settled up and the 6:30 PM window that settled down.
This is what the price did, not what anyone could reliably catch. The cheapest print is often one minute wide,
because one minute is the resolution of the candles these rows come from. A price that existed for sixty seconds
is not a price you can count on filling.
All 73 windows are inside this table. Scroll it down for the rest of the day, and sideways on a narrow screen for the full row.
Before you trade one of these
A 15-minute contract is a short-term bet. It settles one way or the other within minutes of the window
closing, and there is nothing to do with it after that. Kalshi's perpetual contracts work differently: they do
not expire, so a position can be held and sold later. That is a different risk shape, not a safer one. A
perpetual carries funding and leverage, and a position in one can be closed against you long before your view on
the price plays out. Read what a perpetual actually is before you decide
which of the two you are choosing.
Kalshi is the exchange every price and grade on this page came from.
Open KalshiPaid linkWe may earn a commission if you sign up. It never changes a grade.
Sign-up offers change; Kalshi shows the current terms on the sign-up page. 21+. Not available in every state.
PM Countdown is not affiliated with or endorsed by Kalshi.
How this page was made
Every number here comes from Kalshi's public API: the settled markets in series KXBTC15M and their
one-minute candlesticks. The rows were fetched on September 21, 2026 at 17:17 UTC. Kalshi can revise a settlement for seven days
after it posts, so a value on this page can still change until September 28, 2026. Every row carries its Kalshi
ticker so you can check it against the exchange yourself.
Prices are the last traded YES price in cents, taken from the candle price series rather than the resting
bid, because a bid that nobody hit is not a price anything traded at. KXBTC15M uses a deci-cent ladder,
so a price below one cent is a real price and not a rounding artifact. Volume is the figure Kalshi reports for
the window.
The two curves above are arithmetic on those settled results. Nobody placed these trades. Each played window
buys one side at 50 cents, so $100.00 buys 200 contracts. A winning window
returns the stake again and a losing window costs it, and both carry Kalshi's taker fee, which is
ceil(0.07 * contracts * price * (1 - price) * 100) / 100, because the exchange bills that fee when
an order fills rather than when it settles. A window counts as played only when 50 cents traded inside its low
to high range. A real fill would also pay the spread, which these curves do not, so treat them as the friendly
version.
21+. Prediction markets can lose you money fast. Only risk what you can afford to lose. Gambling
problem? Call or text 1-800-GAMBLER. Past results do not predict future results. Nothing on this page is
financial or betting advice.
The Morning Board · free
Tomorrow's plan, in your inbox by 8am ET
One market with the buy price, sell target and hold rule