Polymarket · Kalshi · short-cycle crypto

Trade when
there's edge.
Skip when
there isn't.

NXT Candle identifies when there is directional edge on BTC and ETH, and stays silent when there isn't. One indicator. Four models. Built for prediction markets.

4 models
BTC + ETH · 15M + 1H
~10 alerts/day
All models combined · full suite
Early warning
30–90 sec before candle close
5 stake modes
Fixed · Adaptive · Progressive + curves
The edge problem

Not every
candle is
worth trading.

On the 15-minute BTC timeframe, there are 96 candles per day. On the 1-hour timeframe, 24. The majority of those candles offer no meaningful directional edge, effectively 50/50.

Most traders lose not because the market is random, but because they participate in every candle regardless of conditions. Edge, when it exists, is situational.

NXT Candle models generate a signal only when the model detects a statistically meaningful directional bias.

When conditions are 50/50, no signal is produced.
Silence is also information.

The result is a low-frequency, high-selectivity output. A few signals per day on the 15-minute models. Sometimes just one per day on the 1-hour models.

OKX:BTCUSD 15 | NXT Candle Suite · signal activity
3 signals · 15 skipped
68,100
67,800
67,500
08:0008:1508:3008:4509:0009:1509:3009:45
▲▼ = signal · grayed = no edge detected 15M · BTC
Model design

One model.
One environment.

NXT Candle does not build universal indicators. Each model is constrained to a specific asset, timeframe, and exchange feed. This is a structural alignment decision, not an optimization shortcut.

Candle closes differ slightly between exchanges. On most candles the difference is negligible. But on candles that close near a key level, the difference between exchanges can determine the outcome of a prediction market position.

The 1H models use Binance, the same feed Polymarket uses to resolve BTC and ETH hourly markets.

The 15M models use OKX, selected for structural alignment with the Chainlink oracle and its tendency to reduce borderline false signals.

The exchange is not chosen for convenience. It is chosen to maximise structural consistency between the model's calibration environment and the market being traded.

AssetTimeframeExchange
BTC/USDT1 HourBinance
ETH/USDT1 HourBinance
BTC/USDT15 MinuteOKX
ETH/USDT15 MinuteOKX
What the models analyze
01Candle structure
02Volatility behavior
03Breakout & reversal patterns
04Market participation dynamics
05Short-cycle momentum shifts
The suite

Four models.
One script.

Select your model from the dropdown inside TradingView. Each model is independently calibrated for its asset, timeframe, and exchange. All four are included in a single invite-only script.

Model 01
BTC
1 Hour
Binance · USDT · Polymarket resolution match
Hourly Bitcoin model calibrated on the exact Binance feed Polymarket uses to resolve BTC Up/Down markets.
~1–2 signals per day
Confirmed close only
No repaint
Model 02
ETH
1 Hour
Binance · USDT · Polymarket resolution match
Hourly Ethereum model, independently calibrated from the BTC model for ETH-specific market behavior.
~1–2 signals per day
Confirmed close only
No repaint
Model 03
BTC
15 Minute
OKX · USDT · Chainlink oracle alignment
Short-cycle Bitcoin model on OKX, selected for closest structural alignment with the Chainlink feed used by Polymarket 15M markets.
~3–5 signals per day
Confirmed close only
No repaint
Model 04
ETH
15 Minute
OKX · USDT · Chainlink oracle alignment
Short-cycle Ethereum model on OKX. Independently calibrated for ETH volatility characteristics at the 15-minute timeframe.
~3–5 signals per day
Confirmed close only
No repaint
Money management

Built-in stake
simulation
engine.

Most traders focus entirely on signal quality. But how you size across a losing streak determines whether edge translates into profit. The suite includes a built-in stake simulation engine with five modes, each designed around the statistical losing streak behavior of the models.

FixedLow risk
Stake stays constant regardless of outcome. Predictable, controlled exposure. Recommended for conservative sizing or when starting out.
Stake remains constant at start value
AdaptiveModerate
Linear exposure adjustment suited for higher-frequency environments. Stake moves one unit at a time.
After a loss: stake increases by one unit
After a win: stake decreases by one unit (min = start)
ProgressiveHigh risk
Exposure escalates rapidly. Carries significant risk during losing streak phases, especially on faster timeframes.
×2After a loss: stake doubles
After a win: resets to start stake
Balanced CurveModerate
Stake follows a multiplier curve calibrated on historical losing streak probability. Smoother escalation than Progressive.
~Stake scales with the statistical likelihood of consecutive losses
Aggressive CurveHigh risk
Steeper version of the same curve. Higher upside, higher drawdown risk.
~Same statistical basis as Balanced Curve, more aggressive scaling
Max Accepted Losing Streak: caps stake escalation on all non-fixed modes. Set it to the streak length you are willing to size for.

Manual Stake Override: if your live position differs from the simulation (skipped signals, etc.), override mode aligns alert messages with your actual stake. Essential for webhook/bot integrations.

Fixed and Adaptive modes work. But the Curve modes are designed specifically around how often these models produce consecutive losses. Using them means your stake sizing reflects the actual statistical behavior of the model you're running.
NXT Candle Suite · Dashboard ● WAITING
Accuracy
64.1%
Total Positions
284
PnL
+$46,151
EV per Position
+$162.50
Current Losing Streak
2
Max Losing Streak
7
Max Drawdown
−$2,856
Current Stake
$571
All metrics on closed-candle evaluation only.
Backtest any period directly in TradingView.
A note on losing streaks. A max losing streak of 7 is not a malfunction. It is a statistical reality of any edge-based system. Before subscribing, verify the historical losing streak distribution in the TradingView backtest and size your positions accordingly.
How access works

Invite-only
TradingView script.

The suite is distributed as a single protected TradingView indicator. One alert covers everything: UP, DOWN, WIN, LOSS. Webhook delivery supported for automation. Payment via Stripe or crypto.

01
Subscribe
Complete payment via Stripe or crypto. All four models included. Monthly or yearly.
02
Send username
Provide your TradingView username. Required to grant script access to your specific account.
03
Receive access
The script appears in your TradingView invite-only scripts library. All four models in one script.
04
Activate alerts
Add to a chart, select your model from the dropdown, create one alert using "Any alert() function call." Webhook supported.

Requires TradingView Essential plan or higher to activate alerts on invite-only scripts.

Pricing

One suite.
One price.

All four models included. Monthly or yearly. Cancel anytime.

Monthly Yearly 2 months free
NXT Candle Suite
Early adopter pricing, locked in for life
$89
per month
or switch to yearly · 2 months free
  • All four models: BTC + ETH, 15M + 1H
  • Model selector via dropdown
  • Early warning alerts (30–90 sec before close)
  • Five stake simulation modes
  • High activity mode for 15M models
  • Webhook delivery supported
  • Full backtest in TradingView
Subscribe now
Secure payment via Stripe
What's included
BTC 1H, Binance
ETH 1H, Binance
BTC 15M, OKX
ETH 15M, OKX
Early warning alerts
5 stake simulation modes
High activity mode (15M)
Webhook delivery
TradingView backtest
Subscribe now
One script · model via dropdown · no bundles
Other options

Not ready to subscribe?
We've got you.

Request a free trial week, or subscribe using crypto instead of a card. Both are handled manually and usually sorted within 24 hours.

Free trial
Try it for a week

Leave your TradingView username and we'll give you access to the full suite for 7 days. No card required.

We'll confirm your access here if you leave one.
Crypto payment
Pay with crypto

Prefer to pay in crypto? Leave your TradingView username and email address. We'll send you payment instructions within 24 hours.

We'll send you payment instructions within 24 hours.
FAQ

Frequently asked
questions.

Is this one script or four separate scripts?
One script. You select the model from a dropdown inside TradingView: BTC 1H, ETH 1H, BTC 15M, or ETH 15M. Each model has its own calibration, exchange lock, and signal logic. One subscription, one script, all four models.
What is the early warning alert and how does it work?
The early warning alert fires 30, 45, 60, or 90 seconds before candle close when the model detects a forming signal. It is a preparation alert, not a trade signal. The confirmation entry alert fires after candle close. The early warning gives manual traders time to navigate to Polymarket or Kalshi and prepare their position before the signal is confirmed.
Does the model signal on every candle?
No. The model generates a signal only when it detects a statistically meaningful directional bias. When conditions are close to 50/50, no signal is produced. On the 15M models roughly 3–5 signals per day in standard mode. On the 1H models, sometimes just one. Most candles are skipped intentionally.
Does the indicator repaint?
No. Signals are generated only after a candle has fully closed. The model produces no intrabar signals and does not modify historical output. Each signal is fixed at candle close and does not change.
Why Binance for 1H and OKX for 15M?
The 1H models use Binance because Polymarket resolves its hourly BTC and ETH markets on the Binance spot feed. For the 15M models, Polymarket resolves via Chainlink. During development, multiple exchange feeds were evaluated. OKX demonstrated the closest structural alignment with Chainlink's aggregated pricing across 15-minute settlement windows, and produced fewer borderline false signals. The exchange choice is a deliberate calibration decision.
What is a losing streak and should I be concerned?
Even a model with consistent historical edge will produce sequences of losing trades. This is a statistical inevitability. If you start on a given day, you may encounter a run of incorrect signals immediately. This does not mean the model has stopped working. Edge is evaluated over a large sample of trades. The dashboard tracks your current and maximum losing streak. Verify the historical streak distribution in the TradingView backtest before subscribing.
Can I verify historical performance myself?
Yes. Once you have script access, run the backtest across any historical period directly in TradingView. The backtest shows every signal, its direction, and the outcome. A backtest remains a backtest. Past performance does not guarantee future results.
Can I use this for automated trading or prediction market bots?
Yes. TradingView alerts support webhook delivery. One alert configuration covers entry UP, entry DOWN, and optionally WIN/LOSS results. The alert messages include stake, stake model, and losing streak data, everything a bot needs to execute and manage position sizing automatically.
Is this financial advice?
No. NXT Candle provides analytical software tools. Model outputs are directional predictions derived from historical market data. They do not constitute financial advice or trading guidance. You are solely responsible for all execution decisions.

Trade when
there's edge.
Skip when
there isn't.

NXT Candle Suite: BTC and ETH across 15M and 1H. Built for Polymarket, Kalshi, and short-cycle crypto markets. Not every candle. Just the ones worth trading.