The screener's relative strength score, replayed over 783 Bybit perpetuals and 130 weeks. The ranking persists, but mostly by construction — and the top contracts then lag the universe in the typical case. What it really detects is volatility.
What has risen more than the rest keeps rising more than the rest. That is the idea of momentum, one of the most studied effects in finance, and it is the idea behind the screener's relative strength ranking. On equities it has been documented dozens of times. On crypto perpetuals it can be checked directly — by replaying the screener's own score, week after week.
The previous articles showed that the Stochastic RSI shifts the size of a move without changing the odds of being right, and that stretch is a unit of measurement rather than a signal. Relative strength raises the most direct question of all: do the contracts at the top of the ranking do better afterwards?
The measurement reproduces the screener's calculation exactly, without adjusting it:
One consequence often goes unnoticed. The score is an average of three ranks: to exceed 80, a contract has to lead on all three horizons at once. These are therefore not the strongest 20 %, but roughly 11 % of the universe.
The score was recomputed every week across all Bybit USDT perpetuals currently listed, 783 contracts, from their closed daily candles: 130 weekly rankings, from 11 April 2024 to 1 October 2026, each covering 201 to 738 contracts depending on the period.
Each ranking is set against what follows it and does not overlap it: the return over the next 7 and 28 days. The control is the median of the whole universe over the same period. Also counted are winning weeks, those in which the median of a score band beats the median of the universe. It is the measure closest to the experience of a trader following the ranking week after week.
The first figure seems to vindicate momentum. Among contracts scoring 80 or more in a given week, 42.6 % are still there the following week, although they make up only 11.1 % of the universe: nearly four times chance.
But that figure largely produces itself. From one week to the next, the 90-day window shares 83 days with the previous one, and the 30-day window shares 23. The score cannot change much in seven days: it is built to persist.
To isolate the genuine part, the measure must not overlap. On 7-day performance alone, which shares no day with the following week, the top quintile stays on top in 26.1 % of cases, against 20 % by chance. There is some short-term persistence, but far less than the ranking's stability suggests.
The ranking is stable. But a stable ranking does not guarantee that the contracts at the top will keep doing better. Here are the 28 days after each ranking, across the whole universe:
| Score | Median | Mean | Winning weeks |
|---|---|---|---|
| 80 to 100 | −6.86 % | +1.14 % | 34.1 % |
| 60 to 80 | −5.03 % | −1.40 % | 51.6 % |
| 40 to 60 | −5.14 % | −2.01 % | 66.7 % |
| 20 to 40 | −5.30 % | −0.74 % | 53.2 % |
| 0 to 20 | −6.70 % | +0.23 % | 42.1 % |
| Universe (control) | −5.45 % | −0.96 % | — |
The contracts at the top of the ranking do worse than the universe in the typical case. Their median is lower, and they beat the universe only one week in three. This is not down to a particular period: the proportion is 29.2 % in the first half of the sample and 39.3 % in the second. Nor is it down to a few contracts: with equal weight per contract, the top band's median falls to −9.18 %.
Over 7 days the picture is the same, only weaker: a median of −1.74 % against −1.08 % for the universe, and 40.3 % winning weeks. In every band, the share of contracts up after 28 days stays between 38 and 40 %.
The mean column tells another story, and it is a trap met before. The top band shows a mean of +1.14 % against a median of −6.86 %. A few contracts that keep flying drag the mean upward while the typical case falls back. Anyone reading only the mean would conclude that momentum works.
Both ends of the ranking, top and bottom, share the same signature: a poor median and a flattering mean. That suggests the score first detects movement, whatever its direction. Volatility over the 30 days before each ranking confirms it:
| Score | Daily volatility | Worst 10 % at 28 d | Best 10 % at 28 d |
|---|---|---|---|
| 80 to 100 | 6.28 % | −39.7 % | +38.0 % |
| 60 to 80 | 4.88 % | −32.6 % | +26.2 % |
| 40 to 60 | 4.47 % | −31.0 % | +26.3 % |
| 20 to 40 | 4.70 % | −33.3 % | +29.7 % |
| 0 to 20 | 5.62 % | −37.6 % | +37.6 % |
Volatility forms a U. The top-ranked contracts are 40 % more volatile than those in the middle, and so are the bottom-ranked ones. That makes sense: to break away from the pack, in either direction, a contract has to move a lot. A score of 90 therefore selects contracts with more dispersed outcomes. The most favourable tenth gains 38 % in 28 days; the least favourable tenth loses 40 %.
The full universe includes hundreds of small, thinly traded contracts, and that is where rallies that have already run their course cost the most. The same measurement, restricted to the 100 most traded contracts, gives this at 28 days:
| Score | Median | Mean | Winning weeks |
|---|---|---|---|
| 80 to 100 | −1.83 % | +8.39 % | 50.8 % |
| 60 to 80 | −2.42 % | +1.81 % | 54.8 % |
| 40 to 60 | −2.75 % | +0.76 % | 46.0 % |
| 20 to 40 | −3.18 % | +2.14 % | 46.8 % |
| 0 to 20 | −3.21 % | +7.03 % | 48.4 % |
| Universe (control) | −2.70 % | +2.92 % | — |
Here the top band no longer does worse than the others. But it does not reliably do better either: 50.8 % winning weeks, a coin toss. The pooled median puts it slightly ahead. With equal weight per contract the order reverses: −0.87 % for the top, +1.74 % for the bottom. When a result flips sign depending on how it is counted, there is no effect to exploit.
The volatility U shows up again, and with it high means at both ends: +8.39 % at the top, +7.03 % at the bottom.
The relative strength score does not say what will rise. It says what is moving. Read that way, it stays useful:
What it does not provide is direction. For that, the article on timeframe alignment measured a much clearer gap: an ordinary candle taken while the daily and the weekly are favourable returns +2.26 % over 7 days, against +0.09 % when neither is. Trend context does the work people expect from the ranking.
It covers only contracts still listed today. Those delisted after collapsing are missing, so the universe looks healthier than it was. It spans two and a half years, a single market cycle. And it measures the ranking as it stands, without combining it with anything else.
It also leaves open a question that extends it: how much of a contract's movement truly belongs to it, and how much is simply Bitcoin dragging it along? If everything followed Bitcoin, ranking assets against each other would make no sense. That is the subject of the article on correlation.
What it does establish is enough to correct a common reading. Being at the top of the ranking is a fact about the past, not a promise about the future — and across the full universe, it is rather a sign that the best is behind.
This is not investment advice. This article sets out technical-analysis concepts for educational purposes. Past performance is no guide to future performance, and trading crypto-assets carries a risk of losing the entire capital committed. Legal information.