Crypto Quant Signals for Algorithmic Traders, Built for Cursor, Claude Code, Codex

One more signal in your stack. One you can actually trust.

You're always hunting for the next uncorrelated input, and most of it turns out to be noise. Pearlixa is an independent quant model you can plug into the process you already run. Every call comes back with a calibrated confidence you can weight on, plus an entry, target and stop.

predict.tslive
import { Pearlixa, Horizon } from "@pearlixa/sdk"
const px = new Pearlixa({
  apiKey: process.env.PEARLIXA_KEY,
})

const signal = await px.getPrediction(
  "BTC", Horizon.MID,
)
returns
{
  action: "BUY",
  confidence: 0.87,
  entry_price: 94_250,
  take_profit: 98_400,
  stop_loss: 91_100,
}

Your bot already works. Your data, your models, your execution are dialed in. Pearlixa adds a second, independent quant analysis to weigh alongside your own.

A directional call with entry, target, stop, and a calibrated confidence, in one request.

Never a blind signal. You always see how sure the model is.

Every call ships with the model's own confidence, so you're never acting blind. A 0.87 is high conviction. A 0.55 is the model telling you it isn't sure.

You set the bar: take what clears it, and size the rest by how sure the model is. That number is what turns a tip into something you can gate, size, and blend on.

See the methodology
Every call carries a confidence
your bar · 0.75
BTCBUY
0.87
SOLBUY
0.78
ADABUY
0.64
ETHHOLD
0.55

You see every signal's conviction and act only on what clears your bar.

Confidence is calibrated to mean what it says. Live figures publish at launch.

An edge you add. Not a system you adopt.

Blending it in takes a few lines, not a rebuild. Here are four patterns teams use, from a light confirmation check to feeding the signal straight into a model.

your signalBUY
pearlixaBUY · 0.87
both agree →execute

Confirmation filter

Gate your own signals so they only fire when Pearlixa agrees and its confidence clears your bar.

0.62
0.74
0.87
confidence → position size

Position sizing

Let the calibrated confidence set the size, so a 0.87 setup gets more than a 0.62 one.

rsi_14
funding_rate
orderbook_imb
pearlixa.direction
pearlixa.confidence
→ your model

Ensemble feature

Feed the direction and confidence into your model as one more input, next to your own features.

your signals pearlixa · uncorrelated

Diversifying alpha

An independent signal to diversify the edges you already run. Low correlation, more robust.

Looks simple. Does a lot more.

The single call is the easy path. Batch your whole watchlist in one request, get a full setup back for every name, and rank them however your strategy wants. A market screener in a dozen lines.

core / screener.ts
1import { Pearlixa, Horizon } from "@pearlixa/sdk"
2
3const px = new Pearlixa({ apiKey: process.env.PEARLIXA_KEY })
4
5// one request scans the whole watchlist
6const { predictions } = await px.getBatchPredictions(
7 watchlist, Horizon.MID,
8)
9
10// your bar: keep the conviction, ranked
11const setups = predictions
12 .filter(p => p.confidence >= 0.80)
13 .sort((a, b) => b.confidence - a.confidence)
setups · rankedBTC0.91SOL0.86ETH0.82LINK0.80
  • Batch your whole watchlist in one request
  • Short, mid and long horizons
  • Stream new signals over WebSocket
  • Filter and rank it however your strategy wants

Built to drop straight into your bot.