Skip to content
Market structure · May 14, 2026 · 6m read

Two signals that matter. Fifty that do not.

After thousands of closed trades and years of log data, the inputs that actually predicted outcomes were a small fraction of the ones we tracked. Here is what the simplification looked like.

The discovery

Most signal sets are built through accumulation. You add an indicator because it helped on one memorable trade. You add another because a respected trader uses it. You add a third because the backtest looked clean. Over time you have a system that requires fifteen things to align.

When we isolated each input against actual trade outcomes, most of them contributed less than their presence implied. They were confirming information already present in the stronger signals, or adding noise that justified bad entries.

What the useful ones had in common

The inputs that survived isolation were measuring the same underlying thing from different angles: where liquidity was sitting relative to current price, and whether the order flow behind the current move was genuine or manufactured.

Everything else — most oscillators, most momentum indicators, most pattern-based signals — was downstream of price, which is downstream of those two things. Watching the effect when you can watch the cause is a waste of a filter.

The practical implication for your setup

Take your current signal set and ask: does this tell me something about where orders are sitting or whether this move has real flow behind it? If the answer is no, it is probably redundant.

The goal is to reach a state where a clear signal genuinely is clear — not hedged by four partially-contradicting indicators. That clarity is what allows conviction, and conviction is what allows correct sizing. Read how the desk weighs signals for the specifics of what survived the simplification process.

More from the desk.

Like this?

Read the method.

See the full system view - how the desk reads markets, places trades, and explains every decision.