BTC 15m Observatory

BTC 15m Observatory SYNCING LIVE BTC...

OWNER REMINDERS · NO PROFIT GUARANTEE

ALWAYS CASH PROFITS · DO NOT HOLD AT 99 · WHALES DO EXIST

CASH OUT A LOSS IF THE BOT MAKES A CALL AND YOU SEE IT DROP SIGNIFICANTLY

Poly target $:

READ ONLY --:--:-- W/L Tracker

Locked-call winrate measures market direction, not trading profit.

Fees and late entries affect profit. Bigger positions mean bigger losses.

No main call yet · A direction is not an entry price

Market Coinbase -- Kraken -- Bitstamp -- BRTI -- Regime -- Live · BTC -- · Book -- · UI --

Paper · Read-only Polymarket US 15m ↗

Model right now

Syncing

Waiting for live data...

Time left SYNCING

--:--

Round --:-- – --:-- ET

Bot decisions · 15m Closes --:--

Each call fires once per round and keeps its own record. A call is a direction, not an entry price.

Reversal · Model % --%

Waiting for live data...

Next candles · ET

Sell guide · no call yet

Nothing to manage yet

Sell guidance appears once a call exists this round.

UP ↔ DOWN · Polymarket US 15m market --

UP · market price

--%

Live order book
midpoint

DOWN · market price

--%

Buy UP -- · Sell UP -- · Buy DOWN -- · Sell DOWN --

Last 3 min · UP pushed --% DOWN pushed --%

Waiting for Polymarket US trades this round

Which side traders pushed the price toward (Polymarket US doesn't publish trade sizes) · A market price is not the bot's confidence

Latest UP / DOWN trades · This market

Why BTC is leaning this way -- checks

Polymarket quotes can sit nearly flat while BTC moves · BTC movement alone does not set contract prices

Patterns · Hover the chart for candle details

BTC · Candles

BTC pressure now: --

Setup: --

Continuation: --

BTC Buyers ↔ Sellers · Coinbase · last 3 min --

BUY --% SELL --%

Last 60s · -- bought / -- sold

Raw BRTI · leads the shown price OFFICIAL

-- $--

Raw BRTI vs the shown 60-second average

Shown BTC = Polymarket US current price (60-second BRTI average, the settlement method) · Polymarket US and Kalshi both settle on BRTI · Volume: Coinbase

Next 1m candle: --

Eastern time · Continuous candle model

Past 10 Polymarket US markets · Newest first

Closed candle trend: --

$-- Polymarket US current price · --

Reading finish direction Glide on

Finish reading · not a lock

BTC VS TARGET

--

POLYMARKET US TARGET --

Hover a candle · Arrow keys inspect · White outline = forming

Arrows are context, not guaranteed candles.

Win / Loss tracker This browser

Original calls only, settled on official Polymarket US results. No-call rounds and pending results are excluded. Direction record, not trading profit.

Calls by day · Eastern time

Recent main calls · newest first

Paper trading · 15m · $20 per call

Not real money. Every call "buys" $20 of its side at the Polymarket US ask when it fired, holds to the official result, and pays an estimated fee (Polymarket US' fee coefficient × contracts × price × (1 − price)). The market-favorite row buys whatever the market favors halfway through each round: a strategy only has edge if it beats that benchmark.

Bot memory & learning · 15m

The server journals every round (what the model saw at fixed times left, then the official result) and retrains every hour. A learned model only takes over once it beats the formula on rounds it never trained on.

DEZ'S SAUCE 🔥

Everything on this screen, what the calls mean, and the money math. Two minutes and you're locked in.

THE ONE TO WATCH

5m SNIPER · the 5-minute main call

Fires once per round at 95%+ confidence with 4 of 6 checks agreeing, from 4:00 left.
30-day backtest: about 98% right, and the only call that beats just buying the market favorite.
The catch: it buys at about 93–97%, so wins pay small and one loss wipes out 17–35 of them.

15m

Safe lock

From 12:00 to 0:20 left. Its bar floats between 90% and 97% (around 94%).
Bar goes up early in the round, when it's choppy or fast, or when Polymarket disagrees. It goes down late, when it's calm, or when Polymarket agrees. The screen shows the live bar and why.
Right about 94% of the time. Doesn't beat the favorite yet.

15m · WATCH ONLY

Badman Lock · 15m

The early 15m call: from 14:00 to 7:30 left at 70%+.
Not traded until it proves itself on paper: 100+ calls, profit after fees, and beating the favorite.
Right about 70% of the time, but buys at about 70%, so there's no edge yet.

5m

Badman Lock · 5m

The early 5m call: from 4:40 to 2:30 left at 70%+.
Right about 78% of the time, but it pays about 78% for it, so it's right a lot without being profitable.

THE REST

LOOSE Calls & the benchmark

LOOSE Calls: a looser second opinion. It fires earlier (72%+, 3 checks) and calls almost every round, with its own record. It roughly breaks even, so treat it as context, not something to trade.
Market favorite: halfway through each round it "buys" whatever Polymarket favors. That's the bar to beat. If a call can't beat it, it's not adding anything.

“Safe” and 98% are not guaranteed. Every call is a read on direction, not a promise. Each call type keeps its own W/L record.

READ IN THIS ORDER

The call → BTC vs the target → raw BRTI → time left

How a round is decided

Price = the 60-second average of CF Benchmarks' BRTI, the same number Polymarket US shows.
The yellow target line = that average when the round opened.
UP wins if the average at the close is at or above the target. Otherwise DOWN wins.
Because it's an average, a late spike barely moves it, and in the last minute most of it is already locked in.

The live numbers

Raw BRTI (orange): the real-time index. The average chases it, so raw above the price means the price is about to rise.
Polymarket UP / DOWN %: what the market is charging, not the bot's confidence. UP 70% = a $1 UP share costs 70% of $1.
RISING / FALLING: which way BTC is pushing right now. BTC can be falling while UP is still at 99% if it's far above the target.

The decision boxes

Bot decisions: one card per call type. Each shows Not open (window hasn't started), Waiting (live confidence vs the bar it needs), ▲ UP / ▼ DOWN (it called), or No call (never hit its bar). Each card has its own record.
A call is saved the moment it fires, with its time and the price it bought at. A later change never erases it.
Model right now: the bot's live lean. Lean is just a hint; Risky means the evidence turned against the call.
Sell guide: for someone already holding the round's call.
Switch 15m / 5m at the top. Each market has its own calls and record.

Records & freshness

Locked winrate = how often the calls were right on official Polymarket results. That's direction, not profit.
Avg lock = the average price paid, not confidence.
The top pill shows data age. Stale = something's lagging. The stream adds its own delay.

What a win actually pays

A $20 bet, held to the end (before fees):

Bought at	Win pays	Wins to cover 1 loss
70%	+$8.57	2.3
80%	+$5.00	4
90%	+$2.22	9
95%	+$1.05	19
97%	+$0.62	32

A loss is always −$20. To break even you have to win more often than the price you paid: buying at 90% means you need to be right more than 90% of the time.

The rules that matter

Cheap usually means wrong. A low price means the market disagrees with the call, and the market is usually right. In the backtest, SNIPER calls under 74% won only about half the time.
Being right isn't the same as making money. A 70% call bought at 70% goes nowhere.
Fees and late entries eat profit. Buying after the call at a worse price shrinks every win.
Bigger bets = bigger losses. Size so a losing streak doesn't hurt.

Paper trading

Every call "buys" $20 of its side at the Polymarket US price when it fired, holds to the official result, and pays an estimated fee. It's not real money. It's how we find out which calls actually make money before any real money goes in. Look for calls that beat the market favorite row over many days, not just one hot night.