Troop — BTC 15m Paper Desk

⚠ FEED STALE — showing last read

🔔 ALERTS page must stay open

No alerts this window.

–

LOADING…

connecting to Kalshi + Coinbase

PASS · waiting for data

📌 POSITION

💰 PAPER ARM paper only — no real money

Stake: $250.00

🧮 PROFIT CALCULATOR

Entry 50¢ Stake $250

📊 MARKET

—

🧾 GRADED CALLS fired signals vs official Kalshi settlement

💵 P&L

🎯 CAMPAIGN $1 → $1,000

📈 EQUITY CURVE

🏷 TIERS

🧍 DESK vs MANUAL

🔥 STREAKS

🧾 RECENT TRADES

📖 HOW TRADING WORKS

What you're trading. Every 15 minutes Kalshi opens one yes/no question: "BTC price up in next 15 mins?" If Bitcoin finishes the window at or above where it started, YES pays $1.00 — otherwise NO pays $1.00. You buy either side at the live price (say 65¢), and each contract pays $1 if you're right.

What the desk does. It watches the live Coinbase price and runs a model that estimates the chance BTC finishes above the window-open price. The big banner always shows the highest-probability direction — ▲ HIGHER or ▼ LOWER — with its confidence.

What the tiers mean. Not every read is worth buying:

STRONG (90%+ confidence) — the desk's best setup. Hit 97.1% in the 7-day backtest.
SCALP (2–5 min left, 65–90%) — late-window moves. Hit 90.8%.
EARLY (8–14 min left, 75%+, calm market) — get in before the crowd. New tier, being validated live.
DIRECTION (2–5 min, 55–65%) — informational only, a lean not a trade.

What to buy. Alerts say it plainly: ▲ Buy OVER means buy YES; ▼ Buy UNDER means buy NO. Each alert shows the live ask and a walk-away ceiling — never pay over the ceiling.

ENTRY CLOSED. When the signaled side hits ~98¢ (or the final 3 minutes arrive, or the book empties), Coinbase/Kalshi effectively lock that side. The desk stops alerts and arms there — the read keeps running, but there's nothing left to buy.

Paper only. Every arm here is pretend money. The desk never touches a real account — you place real trades yourself, by hand, on Kalshi or Coinbase Predict.

Settlement. Contracts settle on the CF Benchmarks Bitcoin index (an average across exchanges), not the Coinbase price you see here. The desk grades your paper trades against Kalshi's official result after each window closes.

⚙️ GATES

Side locked at (¢) Entry cutoff (min left) Min net payout (×) Auto-arm on tier signals off = manual only Sound on alerts

🏦 BANKROLL

Bankroll $ Default stake %

Changing bankroll resets the paper ledger. Paper only — no real money.

🎯 CAMPAIGN MODE $1 → $1,000 · opt-in

Campaign mode STRONG-only, last 5 min, ≤90¢ Campaign bankroll $ Base risk %

Kill-switch: 2 consecutive losses or −20% on the day = stand down. Banks 25% of profit at each doubling milestone.

📚 EVIDENCE

Model: P(up) = Φ(d / (σ√m)) — Brownian, zero drift. d = relative distance from the window-open price, m = minutes left, σ = trailing 5-hour per-minute volatility (floored at 2 bps/min). 7-day backtest (2026-10-02…09, ~672 windows):

STRONG (≥90%) → 97.1% hit (n=1,332) · SCALP (2–5 min, 65–90%) → 90.8% (n=993) · DIRECTION (2–5 min, 55–65%) → 70.1% (n=355). Below 55% is a coin flip — no alert.

Fee-aware 1.1× rule: net payout = $1 ÷ (entry + taker fee); true 1.1× after fees needs entry ≤ ~90¢. Taker fee = ⌈7·p·(1−p)⌉/100.

Contracts settle on the CF Benchmarks RTI average, not Coinbase spot — the ~gap cancels because only the window's change matters. Model is underconfident late, overconfident early. One regime sample — your paper ledger is the forward test.