How to read Stratos

Each higher timeframe is distilled to one number from −100 to +100:

• Positive — structure currently leans bullish (buyers in control).
• Near zero — balanced or unclear; no decisive structure (often a range).
• Negative — structure leans bearish (sellers in control).

The sign is the direction (LONG / WATCH / SHORT) and the size is the strength (WEAK / MEDIUM / STRONG). +80 describes a strongly bullish structure;−15 a weak, barely-bearish one. Read the number as how clearly one side is in control right now — not as a target or a countdown.

Direction and strength come straight from the number — the same cut-offs for everyone, every posture layer:

So the +72 above reads as a strong long — well past the 67 mark. A score of +20 is still WATCH: leaning up, but not enough to name a side.

A reading's size matters as much as its sign. Weak readings near zero mean the market has no clear structure — the layers disagree, or price is ranging.

In historical backtests on BTC across rising, falling, and sideways phases, the spans that overlapped with the larger part of a move were the ones where a strong posture formed and then held until it drifted back toward neutral. Readings that flip-flopped near zero were mostly noise. This is a description of past behaviour, observed in one dataset — not a rule to act on, and not a promise about the next move.

The name comes from strata, layers. The whole idea is to read them together:

• Posture layers (1W · 1D · 4H · 1H) describe the structure — the direction.
• Bollinger layers (15m · 5m) show where price sits in its recent range — the timing.

Read the example above as one moment in time: the weekly +64, daily +58 and 4-hour +45 all sit firmly long — the higher layers agree on an up direction — while the 1-hour is −38, leaning the other way. On the shorter timeframes you then read the Bollinger position — where price sits between the lower and upper band — taken only within that higher-timeframe direction: a cleaner up-structure, with band-based timing underneath.

A common way trend-followers read a multi-timeframe indicator is to take the direction from the higher layers and read the finer detail only within that direction. In our backtests this pattern showed up clearly: a posture layer pointing against the layers above it was the least reliable kind, and the short-timeframe Bollinger timing lined up far better when it agreed with the higher-timeframe direction than when it fought it.

Adding more layers tightened this further. When the weekly, the 4-hour, and the 1-hour all described the same direction, the alignment was strongest. When the layers described different directions — say a bearish week inside a bullish 4-hour — the historical reading is simply: less agreement, less to lean on.

The posture layers describe structure and direction — deliberately not a timing or entry tool. That is why the shorter timeframes (15m · 5m) show Bollinger Bands (20, 2σ), a standard volatility band, read only within the higher-timeframe direction.

In our backtests, reading a Bollinger-band position (price near the lower band, drifting back toward the upper) only when the higher-timeframe posture pointed the same way behaved more steadily than the band used on its own — the structure acts as a direction filter around the band. It was most consistent in ranging and falling phases, and weakest in sharp, one-way rallies, where band mean-reversion tends to get run over. Trend-following tools (moving-average crosses, MACD) improved similarly once filtered to the structure's direction; counter-trend tools were the exception — filtering to the trend removed the very edge they rely on.

That band-to-band read has a character worth being honest about. Price stretched to one band and drifting toward the other is a mean-reversion pattern: across our backtest windows it was right more often than not, but the moves it caught were small while the occasional move that kept running against it was larger. An edge shaped that way leans on its hit-rate — it holds together while price keeps returning to its range, and thins out, even turns negative, in a strong one-way trend where the return never comes. On the fastest timeframes, trading costs press on that thin margin too. So it describes a condition that tends to recur in balanced markets, not a dependable outcome — steady through a range, and liable to bleed through a sustained run.

For the same reason the posture reads better as a direction reference than as a fine exit tool. A fast layer like the 4-hour can slip toward neutral and back inside one move; in our backtests, treating that brief wobble as a reason to close tended to exit right as price was turning back toward the band it was drifting to. The higher-timeframe posture describes which way the structure leans; the band shows where price sits within it — two readings, most informative together and pointed the same way.

The takeaway is descriptive, not a recipe: the band shows where price sits in its range, the posture shows which way the structure leans — most useful read together, pointed the same way, not as a standalone entry trigger. Thresholds, entries, and risk are entirely your own decision.

A description is only worth reading if it is honest about the past. Two design choices keep the record trustworthy:

• Closed candles only, never repainted. Each reading is produced on a closed candle, timestamped, and written once. We never go back and change a past reading to look better in hindsight.

Each layer updates on its own clock — shortly after that timeframe's candle closes: the 5-minute band many times an hour, the weekly posture once a week. Nothing moves mid-candle, so a number you see has already closed and won't be quietly rewritten.

• A public, timestamped track record. Every reading is stored and shown on the track record page, so you can see how the posture actually read the market over time — not a cherry-picked screenshot.

Backtests replay the same method over historical candles — same closed-candle inputs and the same frozen prompt — across rising, falling, and sideways phases, with trading fees included. They are how we sanity-check the behaviour described above. They are also, by nature, a study of the past: a single set of windows on a few assets, not proof about the future.

• It is an indicator, not advice or a signal service.
• It reads the current structure; it does not forecast price.
• A clear reading is a snapshot, not a promise — structure can weaken or reverse on the very next candle. Treat every posture as the current state, not a verdict on where price is headed.
• In a straight, one-way run, simply holding can outperform reacting to every layer.
• The shortest timeframes are the noisiest. On their own they are hard to act on; fees alone erode frequent trading. They are finer detail inside a higher-timeframe direction, not standalone triggers.
• The short-timeframe band read is mean-reversion in character — a thin edge: historically right more often than not, but small gains set against larger occasional losses, leaning on the hit-rate and fading in strong trends. It fits balanced ranges more than one-way moves.
• Our backtests cover limited windows and assets — they can overfit. Broader, out-of-sample validation accumulates as the live record grows.
• Always do your own research and manage your own risk.