Reference
The ByteTrend score
The exact rules, windows and regime conditions behind every ByteTrend score. This is the reference; ByteTrend 101 walks through the same ground at a gentler pace.
Every score is a count: five yes/no tests, one point for each that passes. No weightings, no tuning, no discretion. Every asset is scored by the same rules every day, so a score means the same thing for a US mega-cap, a copper future or bitcoin.
Reading a score
| Score | Reading | What it means |
|---|---|---|
| 5 | Strong Bullish | All five conditions met — a full uptrend |
| 4 | Bullish | Four conditions met — the trend is well established |
| 3 | Neutral-Bullish | Three met — a trend forming, or starting to break |
| 2 | Neutral-Bearish | Two met — a trend broken, or starting to form |
| 1 | Bearish | A single condition holding |
| 0 | Strong Bearish | No condition met — a full downtrend |
The five rules
Numbered in the order the scoring engine evaluates them. The order makes no difference to the score: it is a count of five booleans.
| # | Rule | What it tells you |
|---|---|---|
| 1 | Price is above its long moving average | The big-picture trend is up |
| 2 | Price is above its short moving average | Momentum is building here and now |
| 3 | The long moving average is rising | The underlying trend agrees |
| 4 | The short moving average is rising | The recent trend agrees |
| 5 | The most recent extreme touched was the high, not the low | The series has made a new high more recently than a new low |
Rules 1 and 2 compare today's value with the averages; rules 3 and 4 ask whether the averages themselves are higher than yesterday. Rule 5 looks back over a fixed window and asks which came last: the window's high or its low.
Everything is computed on adjusted closes, so a stock split or a dividend doesn't read as a crash. An asset needs enough history to fill a window before it carries the score built on it — the Weekly long average needs 1,000 days of prices (200 weeks for a stock).
Windows
The rules run on two sets of windows: Daily and Weekly. Both scores are recomputed every day — the names describe the length of the averages, not how often they update. For stocks, ETFs, futures and FX the windows count trading days, and a week is five of them; digital assets count calendar days.
Digital assets trade seven days a week, so their windows count calendar days and are scaled by 7/5 to cover the same span of time. The Daily last-touch window is the one exception: it stays at 20 days.
FX is scored on Monday-to-Friday rates only. Weekend rates are kept for currency conversion and repeat Friday's scores. FX counts every weekday, holidays included, so its Daily windows are weekdays and its Weekly windows are exact in weeks (1,000 weekdays is 200 weeks). One consequence: FX's "12-month" return looks back 252 weekdays, about 50 weeks, rather than a calendar year.
| Stocks, ETFs, futures, FX | Digital assets | |
|---|---|---|
| Daily | ||
| Long moving average | 200 trading days | 280 days |
| Short moving average | 30 trading days | 42 days |
| Last-touch window (rule 5) | 20 trading days | 20 days |
| Regime high/low window | 30 trading days | 42 days |
| Weekly | ||
| Long moving average | 200 weeks (1,000 trading days) | 1,400 days |
| Short moving average | 30 weeks (150 trading days) | 210 days |
| Last-touch window (rule 5) | 20 weeks (100 trading days) | 140 days |
| Regime high/low window | 30 weeks (150 trading days) | 210 days |
The Daily score suits month-scale momentum and early turns; the Weekly score, multi-year trends and regime calls.
Local and CAPR
The same rules run on two series for every asset:
- Local — the asset's adjusted close in its own currency. Is it trending up in absolute terms?
- CAPR (Currency-Adjusted Price Relative) — the adjusted close converted into US dollars, divided by the MSCI World index. Is it beating the global equity market, on a constant-currency basis?
Four scores per asset
Two timescales times two series:
| CAPR | Local | |
|---|---|---|
| Weekly | Weekly CAPR | Weekly Local |
| Daily | Daily CAPR | Daily Local |
Trend regimes
Each of the four scores also carries a trend regime. A regime combines the score with where the series sits in its recent range: the named regimes need the series to be at a high or a low of the regime window — today's value equal to the window's highest (or lowest) value, 0% away from it.
| Regime | Condition |
|---|---|
| Leading Trend | Score 5, at the high |
| Emerging Trend | Score 1–4, at the high |
| Weakening Trend | Score 1–3, at the low |
| Bear Trend | Score 0, at the low |
| Neutral | Anything else |
The high and low belong to the same series and timescale as the score. The CAPR regimes use the CAPR series' high and low; the Local regimes use the asset's own adjusted close. The Weekly regimes look back over the Weekly regime window (30 weeks), the Daily regimes over the Daily one (30 days; 42 for digital assets).
Those are the scores that occur. The classifier itself accepts any score above 0 at the low, but the window geometry narrows it: the regime window is the same length as the short moving average and contains the last-touch window, so at a low rules 2 and 5 always fail — a Weakening Trend never scores 4 or 5. At a high, rules 2 and 5 usually pass too, so an Emerging Trend almost always scores 2–4; it scores 1 only when the price hasn't moved for the whole last-touch window (a stale or pegged price, say). A 0 at the low is a Bear Trend, never Weakening.
If the series hasn't moved at all across the regime window, it is at its high and its low at once. The high wins: a score above 0 reads Emerging (rules 2 and 5 both fail on a flat window), and a 0 reads Bear.
Most assets, most days, are Neutral: being at an extreme is a high bar, and that rarity is what makes the named regimes worth watching. A Leading Trend asset can fall back to Neutral with no change in its score, simply by moving off its high.
The Daily regimes
The Daily regimes follow the same five rules on the Daily score and the Daily window, so they fire more often and turn over faster than the Weekly ones. An asset's Daily and Weekly regimes can legitimately disagree on the same day — the Daily one usually turns first.
What the site shows
- Weekly CAPR is the default everywhere. The asset tables lead with the CAPR (W) score and its regime, and the dashboards, Macro and the Trends pages open on Weekly CAPR.
- Switch the score with the Weekly / Daily and CAPR / Local switches at the top of the dashboards, the Macro pages and the Trends pages. The tables carry all four scores as columns.
- On a security page, the badge in the header is the Weekly CAPR regime beside the Weekly CAPR score. The chart opens on Weekly CAPR over ten years, and its selector switches between the four scores. Trend Analysis lists all four, each with its own regime.