BTC/USD — CoinStats

CoinStats exchange prices via a small Cloudflare Worker · volatility + momentum model · practice only · not financial advice

Live BTC/USD

—

Simple view: hide the advanced cards (trend/streak, re-entry, scale-in, calibration, exchange and diagnostic tools). Nothing you've configured changes — this only changes what's shown. Switch back any time.

Quick setup, from what the data in this page has actually shown so far:

Connecting to CoinStats…

Manual override: use these only if sync status below looks wrong (e.g. after editing the token in the field below).

Kalshi 15-minute market

Waiting for the first reading…

read Kalshi’s market through the kalshi-btc15 Worker (public prices only: it cannot trade)

Refresh every Fee per contract ¢ (set this to what Kalshi charges you)

fill the strike automatically each window from Kalshi’s target (only if you have not set one for that window) also read the real-time BRTI over an authenticated feed, if you have set one up (see the server\u2019s README) — silently does nothing otherwise

use Kalshi’s live BRTI as the main price the page trades on, whenever the live feed is fresh and connected — falls back to the usual CoinStats price automatically and silently otherwise. The volatility estimate still comes from CoinStats either way; only the price level changes. Off by default.

Strike price

Type the BRTI you can see and press Enter the moment you read it. The price adjustment updates by itself.

news window: widen the odds (volatility × 1.5). Tick it around CPI, Fed and jobs releases. automatically widen the odds when Alpha Vantage shows a burst of Fed/rate/macro Bitcoin news

Trigger at within

Volatility reading

fat-tailed odds (Student-t) instead of a normal curve

Your own drift view (% per hour)

—

Enter a strike price to see the estimate.

closes undercloses over

—

Cross-exchange reference price

Waiting for the first read…

Exchange	Price	24h volume (BTC)

A volume-weighted average price across the public exchanges that make up most of CF Benchmarks’ BRTI — the index Kalshi’s 15-minute BTC contract actually settles against (Coinbase, Kraken, Bitstamp, Gemini, Bullish, itBit). LMAX Digital is a real BRTI constituent too but has no open public API, so it is not included; this is an approximation of BRTI, not a reproduction of it. One slow or unreachable exchange is simply left out of the average rather than failing the whole read — see how many sources answered above. This feeds the “Volume confirmation” check in Trade Readiness below; the page’s own tracked price and model are unchanged.

Trade Readiness

Enter a strike to see the checklist.

only show what's currently blocking

Every active gate for the currently favored side, checked all at once -- not just whichever one is currently blocking a trade. A signal can pass several gates comfortably and still fail on one; this shows the whole picture before a trade fires, whether or not auto-trading is on.

Six optional checks below are always shown, either way -- checking a box here is what turns that specific one from informational into something that can actually block a trade. All off by default.

Block on	Check	Threshold
	Fast vs slow indicators disagree	2-min change/Accelerator vs Stochastic/Bollinger/RSI/EMA/30-min
	Edge over Kalshi’s ask price, after fee	pts
	Strike far from the last recorded close	> $100
	Strike near a 24h high/low or round level	within 0.15%
	Intra-window consistency
	Strike near the volume point of control
	Volume confirmation (cross-exchange)	rise/min

Momentum

Indicator	Reading	Lean
Waiting for CoinStats chart history…

Momentum enters the odds as

Each indicator votes from −1 (down) to +1 (up). Readings that measure the same thing are grouped so they count once: RSI, Stochastic, Awesome and Accelerator are one momentum vote, the EMA-9/21 trend and 30-minute change are one vote, and the order book and taker flow are one vote. Bollinger, prior-candle momentum, the 1-minute EMA cross, the 2-minute change, the price against its hourly VWAP and the mark/index basis are one vote each, and the three volume readings (5-minute, 15-minute and signed) are one vote together, and the composite is the average of the votes. RSI and the 5-minute EMA still come from CoinStats' chart (refreshed every 2 minutes). Stochastic, the Awesome and Accelerator Oscillators, Bollinger and prior-candle momentum use real 15-minute candles from Crypto.com when its signals are on, falling back to bars built by joining three of CoinStats' 5-minute points otherwise (marked "approx.", with approximate highs and lows). The 30-minute change and a 1-minute EMA cross use real Crypto.com candles at those timeframes; the 2-minute change comes from this page's own live prices; and the order book, taker flow, volume and basis come from Crypto.com's trade and book data when ticked. It can only shift the odds a little; the size of the move needed to reach your strike still does most of the work. Abnormal volume (over 3 times the norm) widens the odds instead of pointing a direction. Momentum can enter the odds two ways: as a flat nudge of up to 8 points, or as a drift, which shifts where the price is expected to end up. A drift matters most near the strike and fades for strikes far away, which is how a real drift would behave.

Order flow (Crypto.com)

▾

use Crypto.com's order book, taker flow and volume as signals. The price itself stays CoinStats-only.

Last 15 minutes

▾

Paper trading (practice only)

Balance$1,000.00

Contracts

Never risk more than of balance on one bet

A hard cap, applied to every bet -- manual, auto, Kelly-sized or re-entries -- right before it is placed. If the requested size costs more than this share of your current balance, it is quietly reduced to fit (never increased). This is the single rail every sizing method has to pass through, independent of how the contract count was arrived at.

Stop loss Take profit

A bet's value is its contracts at the model's current odds for that side. Stop loss closes it once it is worth that much less than you paid, and take profit once it is worth that much more. Closing sells at the current price, with no fee or spread, after the move has held for 2 seconds on a fresh price. A contract cannot be worth more than 99¢, so a bet bought at 75¢ can gain about 32% at most, and a higher target will never trigger for it (the row says so).

scale the stop with volatility: wider when the market has gotten more volatile since entry, tighter when it has calmed down (0.5×–2× the stop above) trail the stop once a bet is ahead, instead of only measuring from entry

Start trailing once ahead by give back up to

trail the take profit too: once a bet reaches the take-profit gain, let it keep running instead of closing right away

give back up to

Allow cancelling

Cancel on an open bet refunds what it cost, as if it had never been placed. That is different from closing, which sells at the current price. The Track record shows whether cancelling helped.

Bet	Expires	Left at entry	Contracts	Odds	Result
No trades yet.

Each contract costs the model's odds for your side, in cents, and pays $1 if you are right. Over at 75% costs 75\u00a2 and gains 25\u00a2 per contract if it wins, and loses the 75\u00a2 if it does not. Prices are rounded to the cent and kept between 1\u00a2 and 99\u00a2. Bets settle a few seconds after expiry on this page's combined price. One open bet per expiry. When you type the BRTI at expiry into a bet's row, that number also becomes your strike for the next window. If you didn't bet that window, type your strike yourself. Fake money, real logic.

Auto-trading

place bets automatically when the signal is strong enough

Auto-trading is off.

Auto-trading: entry rules

▾

Minimum signal from 50%

Edge needed before momentum so momentum cannot create a bet on its own

also require momentum to agree in the bet's direction by at least

Indicator votes agreeing

Only bet when the side costs to

Only place bets

Require the signal to hold for before betting (filters a brief flicker from a stable reading)

Size

Kelly sizing uses

Stake = balance × this fraction × (model odds − contract price) ÷ (1 − contract price) — the standard formula for a bet that costs its full stake on a loss. Only used when Size above is set to Kelly-sized; still capped by the maximum risk per trade set in Paper trading, and still throttled down after a losing streak like the other sizing modes.

Trend & streak rules

▾

Trend gate — checks the closes recorded in Window history, not the oscillators, so it can catch a real multi-window move the indicators keep reading the wrong way. One thing this is not evidence for: a check across 918 recorded 15-minute windows found that one window's direction does not predict the next window's direction — continuation sat at 48.4% (95% CI 45–52%), a coin flip, even after a large prior move. This gate looks at several consecutive closes together, which is a different, untested question; it is not invalidated by that finding, but it is also not supported by it.

watch for a bet fighting the recent trend

Look back at least

When a bet fights it extra points

Same-side loss streak — stops betting one direction after it keeps losing, independent of the trend gate above.

pause a side after it loses in a row

Losses in a row within

Re-entry after a stop loss

▾

Re-entry after a stop loss — if a bet gets stopped out and the price keeps going the other way, this allows one more trade in the same window on the opposite side, instead of sitting out until the next window. A take-profit close never re-enters, only a stop loss. This also applies to bets you place by hand, not just auto-trading. Needs the stop loss itself switched on above to ever trigger.

allow a re-entry after a stop loss, on the opposite side

Re-entries per window

Extra edge per re-entry

Minimum time left to re-enter

halve the contract count on each successive re-entry (floor of 1)

It bets at most once per expiry, only after the signal has held for 3 seconds, never within 15 seconds of expiry, and only when you have set a strike for this window (typing one, or entering the BRTI at the last close), so it cannot trade on an old strike. Its size shrinks after losses: half after 3 in a row, a quarter after 5. If your balance falls 20% below where it started the day, it switches itself off, and ticking the box again starts a fresh baseline. It uses the same contract pricing and settlement as manual bets.

Scale into a winning position

▾

Adds more contracts to an OPEN bet once the model's odds for its side have moved further in its favor, at the new (now more expensive) price, blended into an average cost. This is a manual button on every open bet's row regardless of this setting -- the toggle below only controls whether the auto-trader also does it by itself. It increases risk on a position already committed, so it is capped and off by default.

let the auto-trader add to an open bet once it is more certain

Add once certainty rises by

Add size max additions

Minimum time left to add

Track record

Settled bets0

Correct calls0

Hit rate—

Average model odds—

Net result—

Auto-trade bets—

Stop-loss closes—

Take-profit closes—

Edited bets—

Momentum vs the bet	Bets / hit rate
Accumulating — check back after a few bets.

Time left when the bet was placed	Bets / hit rate	Avg odds
Accumulating — check back after a few bets.

Shares of the window are used, so this works for any expiry: 60% of a 15-minute window is 9 minutes. It shows whether bets placed later, with more of the window already played, are better calibrated than early ones.

If bets where momentum agreed keep beating the ones where it disagreed, the indicator is earning its place. If the two look alike, it is not adding much. This needs a real sample before it means anything.

Kalshi official values

▾

Kalshi publishes the index value each 15-minute window was settled on. That is the number you otherwise type in as the BRTI at every close, so it can check your typed values, show how far this page’s price really was from the index, and (if you switch it on) stand in for the BRTI you would have typed. It never touches a bet you already gave a BRTI for.

Nothing loaded yet.

Closed	Target	Official	Result	Your BRTI	Diff

keep this up to date automatically (one small request every few minutes) use Kalshi’s official value as the BRTI for closes I did not type one for, from the moment this is switched on (it settles those bets, feeds the comparison log and the price adjustment; older bets are left alone)

Indicator calibration

▾

Checks your own trade history: for each saved indicator reading, how often you won when it agreed with the bet's side versus when it disagreed. This is informational only -- it never changes the live model's weights by itself. Needs at least 15 bets on each side of an indicator before saying anything about it. Agreeing and disagreeing bets are compared at similar price leads, because a bigger lead wins more on its own and would otherwise make an indicator look better or worse than it is.

Indicator	Agree (n, win%)	Disagree (n, win%)	Read

Calibration drift — compares your most recent bets against everything before them. A model that's quietly gotten worse (or better) since you last checked should show up here before it shows up as a losing streak.

Recent window

Cancelled bets

▾

No cancelled bets.

Bet	Expires	Left when cancelled	Refunded	Holding would have been	BRTI check
No cancelled bets.

Cancelled bets are refunded in full and kept out of the Track record: they do not count in the hit rate, the odds check, the time-left table, the net result or the auto-trade count. This box is only a log, so you can see whether cancelling helped.

Window history

▾

Recording the price at every 15-minute close.

Exhaustion study: after a run of windows that all moved the same way

each at least

Every 15 minutes the page saves the price at the close, how big a move is normal for that window, and the BRTI if you typed one for it. Windows it was open for come from live prices. When it was closed, the last 24 hours are filled in from CoinStats' 5-minute chart (approximate: interpolated to the close, and marked as chart data). The study only counts runs of consecutive windows from the same source, and measures each move in standard deviations of that window's own volatility. This is a gauge of how much evidence exists, not a rule yet.

Exchanges

▾

bring each exchange's price up to date using Crypto.com's trades. Only the movement since that price was set is used, never Crypto.com's own level.

Extra data delay (seconds)

Exchange	Last trade	Age	Up to date	Use
No data yet.

Price adjustment ($)

Added to the combined price everywhere: the header, chart, odds and settlement. Use it when this page sits above or below the BRTI by about the same amount every time.

Averages against the BRTI

▾

set the price adjustment by itself each time a BRTI is entered also poll Alpha Vantage's live BTC/USD rate as a 6th candidate below (uses your Alpha Vantage plan's request quota)

This page's price, made from	Average difference	Spread	Readings
No comparisons yet.

Recent readings. Remove any that were typed wrong.

Time	BRTI typed	Page vs BRTI
No readings yet.

Save and sync

▾

keep my data saved to the paper-store Worker, so it follows me to other browsers and computers

Starting…

Sync token

Paste this exact token into the same field on another browser or device to share this paper-trading history there. Anyone with this token can read or overwrite this history, so treat it like a password.

Bring data in from another browser without using the token: paste an export here.

Synced: your paper account, ledger, cancelled bets, BRTI readings, price adjustment, settings, and whether the session is open or closed. Not synced: whether live fetching and auto-trading are switched on (each browser keeps its own, so two devices never trade at once), the credit counter, and live prices. Closing the session on one device closes it everywhere once the other device syncs, so remember to press Start session there too. Unlike the Claude-artifact version, this standalone build does not detect conflicting changes from two devices at once -- the last one to sync wins -- so use one device at a time. Importing replaces this browser's data.

Live fetching

▾

Fetch prices from CoinStats (3 credits each) on the schedule below until you turn it off or the credits run out. Pauses while this tab is hidden.

Schedule

Fast rate

Before the last 6 minutes

Credits on your plan per month

How it works: the price is a combination of last-trade prices on the BRTI's USD exchanges that CoinStats lists. The odds assume no drift by default (you can add your own drift view, or let momentum act as a drift) and use how far Bitcoin has typically moved in 5 minutes over the last hour and the last 24 hours (whichever is larger), plus the age of the combined price, the larger of that and the last 30 minutes' moves on this page is used, widened for the news window or abnormal volume, and fat-tailed by default; then momentum adds a bounded nudge. To see how close this page is to the BRTI, type the BRTI at expiry into a settled bet's "BRTI check" box; it re-settles that bet on the BRTI and records the difference. CoinStats may not list every BRTI exchange, and these are last-trade prices, not order books.