How Strategies Are Proven

What this is

A trading strategy is just a set of rules for when to buy and when to sell. The hard part is knowing whether those rules actually work, or whether they only looked good on the particular slice of history someone happened to test them on. SentinelHub treats that question as the single most important thing it does. No strategy reaches your real account on a story, a hunch, or a flattering chart.

Instead, every candidate is put through the same evidence test, in the same order, every time:

• ·Out-of-sample testing — the rules are tuned on one stretch of market history, then judged on a differentstretch they have never seen. This is the test that matters: it's easy to invent rules that fit the past perfectly and fail on what comes next. Holding back unseen history is how the mirages get caught.
• ·Realistic costs— every test includes the real-world frictions of trading, so a strategy can't look profitable on paper and then bleed money in practice.
• ·A high bar, set in advance — the targets a strategy must clear are written down before the test is run, so the goalposts can never be quietly moved to make a weak result pass.

Why it matters to you

You are not being asked to trust a black box, a marketing claim, or a backtest that was quietly tuned until it looked good. Every strategy carries an evidence dossier— a plain record of what it has proven, where it works and where it doesn't, and what happened before and after every change ever made to it.

• ·Proven, not promised — the edge is measured on history the strategy was never trained on, so the number means something forward, not just backward.
• ·Specific, not blanket— the evidence says exactly where a strategy works: which sectors, and in which kind of market. "It works" is never the claim; "it works here, in theseconditions" is.
• ·Auditable, not opaque— every change to a strategy is logged as a before-and-after with the reasoning attached. You can trace why any strategy is in the state it's in.

The proof point

When several strategies were tested together, only one — a mean-reversion approach called RSI-2 — earned a defensible, well-evidenced edge on unseen history. It is the workhorse carrying the paper book today.

A different candidate looked genuinely promising on the history it was tuned on. But when it was tested on unseen data, with real trading costs included, the edge evaporated — it went from marginally profitable to actually losing money. So it was held back, not deployed. That is the system working exactly as intended.

How the edge is re-checked over time

Proving an edge once isn't enough — markets change, and an edge can fade. Three habits keep the picture current long after a strategy goes live.

• ·Regime-aware— the right strategy for the right kind of market. A market "regime" is simply the mood of the market — calm and rising, falling, or overheated. The evidence shows which regimes a strategy thrives in and which it should sit out, so it isn't trading into conditions where it has no edge.
• ·Re-checked on a cadence— a live strategy is regularly re-tested against the standard it first passed. If its edge starts to decay, that's caught early — and a weakening strategy can be throttled or stood down before it bleeds capital, never silently left running.
• ·Fully auditable— the dossier is the running record: what was proven, every re-check, every change and its result. Nothing about a strategy's status is a mystery.

Where this is heading: the maturity ladder (proposed)

A proposal now under review (APP-062, not yet approved or built) would make a strategy's whole journey visible as three rungs. A new idea — including one you suggest — first trades tiny thesis size, then larger validation size, and only reaches full deployment after clearing the same statistical promotion gate, independent audit, and explicit operator approval that guard every promotion today. The small rungs use a simple "6 wins out of 10" filter to kill weak ideas cheaply; the final say always belongs to the statistical gate, not the tally. The same proposal adds a Discuss Strategy tile where you can put an idea or concern to the system in plain language and watch it climb — or fail — the same ladder everything else faces.

You can see the current picture for yourself on the Strategies & the Matrix page, which shows which approaches are live and where each one has earned a real track record. And the safeguards that protect every individual trade are covered in How Your Capital Is Protected.