What the algorithm actually does.

Most retail systems assume markets behave like the equations on the page. CORE APEX assumes the opposite. Below is a sketch of the edge, the principles, and the trade-offs we made along the way.

The edge, in three sentences

Markets are nonlinear, and most retail systems assume otherwise. CORE APEX reads liquidity structure — where size has to transact, when sessions overlap, how pivots get reclaimed — and only fires when those mechanics align. The result is roughly two to four trades a day across seven instruments, holding minutes to hours, never overnight, never overleveraged.

What we actually look at

Liquidity zones

Every market has price levels where significant interest has accumulated — prior session highs and lows, options strikes near expiry, levels where stops cluster. The algorithm classifies these zones daily and watches how price behaves when it returns to them.

Session-overlap timing

London open, NY open, Asia close, and the London/NY overlap each have characteristic volatility signatures. We don't trade through them blindly; we have explicit rules for what configurations qualify in each session window.

Pivot reclaims

Most failed breakouts give a clean reversal signature — price pushes through a level, fails to hold, and reverses sharply when it comes back. The strongest single-trade pattern in our universe. About 40% of trades come from this setup.

What we deliberately ignore

News and headlines. We don't trade NFP, FOMC, or earnings. The algorithm goes flat 15 minutes before, stays flat for 30 minutes after.

Most indicators. No moving-average crossovers, no RSI overbought signals, no MACD divergence. None of these encode liquidity behaviour reliably enough to size against.

Predictions. We don't try to predict where price will go. We classify what configuration is in front of us and assign it a known historical edge. If the edge is positive after costs, we trade.

The four principles

The product is the algorithm and the service. Source stays with us.

Liquidity is the primary signal. Price is the byproduct. Our engines read structure, order flow, and session dynamics — the mechanisms that move markets before they move.

Adaptive by design, not by overfitting. Regimes shift. The algorithm classifies, reweights, and defers — but the rule set hasn't changed materially in 18 months. Adaptation happens at the parameter layer, not the strategy layer.

Infrastructure, not source. We deliver through managed signals and execution. Subscribers receive capability; they do not receive code. This is deliberate — for their protection and ours.

The same algorithm we run. No premium tier, no faster signal feed for higher-paying subscribers. Every subscriber sees the same trades at the same time as our own capital does.

What happens when it stops working

It will, eventually. Strategies decay; edges close. When that happens, we'll tell you. There's no "model rebuild in progress" or "temporary pause" — if the edge has gone, the responsible thing is to wind it down, refund the month, and start over.

The clearest signal would be a drawdown that exceeds 15% and refuses to recover within four months. We've never hit either threshold; the policy is documented for the day we might.

---

Want the full methodology updates?

We publish a quarterly note describing what changed in the algorithm, what didn't, and how the regime is shifting. Available on the research page. Subscribers receive a more detailed version with attribution by setup.