Built for traders who read the math

Seventeen models read the market.
You read seventeen models.

The Quantum Engine runs an ensemble of quant strategies against your watchlist every session, scores the agreement, and surfaces only the trades the math agrees on.

Free for 7 days. No card required. Cancel in two clicks.
Capital is at risk. Past performance is no guarantee of future results. Read the risk disclosure.
10

Grade A gate, ten checks

A signal becomes Grade A only when ten independent checks line up — sample size, profit factor, drawdown, regime alignment, change-point detection, latency, volume, hazard, factor exposure, walk-forward stability. We reject more than we publish.

A brain that updates from realized trades

Every closed trade teaches the ensemble. Equation weights move with realized P&L, not with backtest dreams. The numbers you see are the numbers we ran on.

In-sample and realized, side by side

We label every chart. In-sample is in-sample. Realized is realized. When they disagree, you see both.

Why most signals fail

Most retail signal services overfit, hide losses, and quietly retune.

Pick any newsletter. Look at the open positions. Look at the closed positions. Compare them. The gap is where the marketing lives.

The standard tricks are well-known. Cherry-pick winners. Retrain on the test set. Drop a strategy when it stops working and add a new one. Stop publishing the ones that lose. Quote in-sample Sharpe ratios above 3 with a straight face. Ask the regulator nothing about it.

Real edge is rare, small, and hard to keep. We have a track record because we built systems to defend against the ways edge dies — overfitting, regime drift, look-ahead leaks, false discoveries across thousands of symbols, and the human reflex to retune at the worst time. Each of those defenses lives in the codebase. None of them is marketing.

If a signal service tells you a win rate above 75% on liquid US equities, you are looking at survivorship bias. Walk back to your wallet.

What's inside the engine

Seventeen models, one ensemble, one gate.

The ensemble

Seventeen quantitative equations covering momentum, mean reversion, volatility regimes, factor exposure, change points, and microstructure. Each runs independently; the brain weighs them by realized P&L, not backtested dreams.

The Grade A gate

A signal earns a Grade only when ten independent checks pass — sample size, walk-forward stability, asset-class profit factor, drawdown ceiling, regime alignment, change-point absence, signal latency, volume confirmation, hazard plausibility, and factor non-redundancy. The gate fails-safe open on missing data; it fails-safe closed on bad data.

The brain

Equation weights are not static. Every closed trade — paper or real — feeds back into the ensemble. Equations that produced losses get downweighted. Equations that produced wins get upweighted. The brain saves a snapshot every day so the change is auditable.

The universe

Roughly 1,100 instruments across US large-cap, small-cap, ETF, FX, futures, indices, KRX, TWSE, ASX, and select crypto. We do not scan instruments where data is stale, exchanges are closed, or volume is too thin to trade.

The schedule

Thirteen scheduled scans run across the global session — Asia indices, India, Europe, US, FX London open, FX London-NY overlap, futures NY open, and metals AM/PM. Each schedule has its own gates and its own kill-switches.

The publisher posture

We are a publisher of financial information, not your adviser, not your broker. We do not know your circumstances and we do not give personalized advice. The trading decision is yours. Read the risk disclosure.

Realized, not in-sample

Read the realized scoreboard before you subscribe.

We publish a scoreboard. It includes every signal we ever issued, when it issued, when it closed, and what it closed at. Wins and losses are both there. The Sharpe ratio shown is the Sharpe of the realized track, not the in-sample backtest.

The realized number, today, is in line with our internal target. It moves around. Some weeks the brain has a brutal week. We publish those weeks too.

If we ever stopped publishing losses, the scoreboard would tell on us. The scoreboard exists because of that.

See the scoreboard →

What we are not

We're a publisher. Not your adviser. Not your broker.

Ikaay Capital is a publisher of financial information. We are not a registered investment adviser, broker-dealer, fund manager, or fiduciary. We do not know your circumstances. We do not give personalized advice and we do not respond to requests for it.

The signals on the platform are model output. They are not recommendations. The trading decision is yours. Capital is at risk.

If you want personalized advice, talk to a licensed adviser in your country. If you want to follow algorithmic signals with discipline, you are in the right place.

Coverage

1,100+ instruments. Real exchanges. Real session timing.

We scan US equities (NYSE, NASDAQ), US ETFs, US futures, FX majors and crosses, the major European indices, KRX (Korea), TWSE (Taiwan), ASX (Australia), India (NSE), Hong Kong, Japan, and a selected list of crypto pairs.

Each market is gated by its own session-hours rule. The platform does not generate fresh signals on a market while its exchange is closed. If a price is stale, the scan suppresses the symbol; we surface the suppression in the daily summary.

Who builds it

We trade what we publish.

The team behind the engine has worked across systematic equities, futures, and FX. Every signal that goes out is a signal we look at on our own books. The platform is the same platform we use.

If a strategy fails for us, we kill it. If a strategy works for us and the realized track confirms it, we publish it.

Plans

Start free. Upgrade when you need more.

Five tiers, monthly or annual. Annual is roughly seventeen percent cheaper.

Free
$0
Try the engine. One market, one scan a day, watermarked output. No card.
Lite
$29/mo
3 markets, 20 scans a day, daily email, paper trading.
Pro
$99/mo
All markets, full risk model, Telegram bot, watchlist alerts, scheduled scans.
Edge
$299/mo
Everything in Pro, plus factor exposure, weekly research, leveraged-ETF coverage, sector caps, alpha attribution.
API
$999/mo
Raw API, custom scan schedules, programmatic signals, white-label.