Methodology

A complete explanation of how Squared calculates its numbers and how to read a brief.

§1

What this brief is

Each brief is a statistical snapshot of a specific asset at a specific moment. The numbers shown in a brief — MR24, MR48, Edge48, action levels — come from deterministic calculations run against historical price data. The same inputs always produce the same outputs.

The narrative text — summary, cases, context — is written by a large language model (LLM) that receives the same underlying numbers as input. The LLM does not generate new information. It translates numbers into human-readable language. Phrasing may vary between similar setups, but the underlying numbers are always the same.

Trust the numbers (percentages and price levels) in a brief. The narrative sentences are a restatement of those numbers — they add no information beyond what the numbers already contain.

A brief does not recommend any specific buy or sell action. It describes how well historical patterns match the current setup. All trading decisions remain entirely your own responsibility.

§2

Analysis windows

All indicators are calculated from three fixed time windows.

1H window · 2,160 bars90 days of hourly data. Primary unit for EMA50, EMA200, ADX(14), and ATR(14) calculations. Historical similar-case matching is also performed across this window.

4H window · 540 bars540 complete 4H bars resampled from 1H data. Used for trend direction confirmation and a smoother EMA200. Partially-complete 4H bars are excluded.

Chart window · 168 bars (7 days)The most recent 168 1H bars. Used for the brief's chart visualization and for calculating the 24H high/low that feeds into action level definitions.

§3

Indicators

EMA50Exponential moving average, period 50, over the 1H window. Serves as the mean-reversion anchor and the source of the reversion target price.

EMA200Exponential moving average, period 200, over the 1H window. Used to confirm long-term trend direction. Price above EMA200 = bullish regime; below = bearish.

Distance%(close − EMA50) / EMA50 × 100. Positive = above EMA50, negative = below. This value drives setup bucket classification.

ADX · DI+ · DI−Wilder smoothing, period 14. ADX measures trend strength regardless of direction. DI+ (positive directional index) above DI− signals dominant bullish pressure. ADX ≥ 20 is treated as trend-confirmed.

ATR%ATR(14) / close × 100. Current volatility normalized to price. Used as a reference when sizing stops. Not displayed directly in the brief but used in internal calculations.

RegimeClassified into one of five regimes based on price position relative to EMA200 and ADX strength: strong uptrend / weak uptrend / neutral / weak downtrend / strong downtrend.

§4

Similar setup definition

To calculate probabilities, we need to define what counts as a "similar historical case." Squared requires all three of the following criteria to match.

1. SignThe sign of Distance% must match. A setup currently below EMA50 only matches historical cases that were also below EMA50.

2. Distance bucketThe absolute value of Distance% must fall in the same range. Five buckets:0–1% · 1–2% · 2–3% · 3–5% · 5%+

3. ADX bucketThe ADX value must fall in the same range. Three buckets:ADX < 20 · ADX 20–25 · ADX 25+

The combination of these three criteria forms the Setup Bucket. There are up to 30 buckets (2 signs × 5 distance × 3 ADX). n is the count of historical cases in that bucket. Only rows with 48H forward data available are included, so n is identical for both MR24 and MR48.

§5

Probability definitions

For the n historical cases in the same bucket as the current snapshot, we aggregate what happened at the specified horizon.

MR24The percentage of cases where the absolute Distance% at the 24H horizon was smaller than at entry. Example: currently −2.5%, 24H later −1.1% → mean reversion occurred. Unit: % (0–100).

MR48Same definition, at the 48H horizon. Generally higher than MR24 because more time is available for reversion.

Continuation100 − MR48. The percentage of cases where price moved further from EMA50. Not displayed directly, but a low MR48 implies higher continuation probability.

Edge48MR48 − 50. The excess prediction accuracy above random coin-flip (50%). A positive Edge means this bucket shows statistically meaningful mean-reversion bias. Example: Edge48 = +24 implies MR48 = 74%.

These probabilities describe "what happened historically in cases that looked like this." They are not predictions and do not guarantee future outcomes.

§6

Action level definitions

Action levels are derived directly from indicator values at the time the brief is generated. They are not trading recommendations — they map notable price levels.

Reversion TargetThe current 1H EMA50 value. The primary destination level in a mean-reversion scenario.

Recovery ConfirmationThe recent 24H high. A breakout above this level suggests recovery beyond simple mean reversion — potentially trend resumption.

Lower BoundaryThe recent 24H low. A breakdown below this level invalidates the mean-reversion scenario and signals potential for further downside.

Trend ConfirmationADX ≥ 20 = trend confirmed. Under this condition continuation scenarios carry more weight. ADX < 20 = weak or neutral trend, where mean-reversion scenarios are relatively more likely.

§7

Sample-limit warning

The smaller the sample size (n), the less reliable the probability estimates.

n ≥ 30Reliable. Use the probabilities with confidence.

n = 15–29Sufficient but limited. Use probabilities as a reference alongside other context.

n < 15"Sample limited" warning is shown. Probabilities are statistically unstable in this bucket. Do not rely on the numbers at face value.

Newer assets (short price history) and extreme price levels (Distance > 5%) are most likely to have low sample counts. You will see this warning most often with emerging assets or outlier setups.

§8

Glossary

Korean / English term reference used throughout the app.

한국어 (Korean)English

Analysis metrics

평균 회귀Mean Reversion

추세 지속Trend Continuation

24H 회귀율MR24 (24H Mean Reversion Rate)

48H 회귀율MR48 (48H Mean Reversion Rate)

48H 엣지Edge48 (48H Edge over random)

거리%Distance% (from EMA50)

사례 수 / nSample Size / n

셋업 버킷Setup Bucket

샘플 제한Sample Limited

Action levels

회귀 목표Reversion Target

회복 확인 레벨Recovery Confirmation

추세 지속 레벨Continuation Level

하단 경계Lower Boundary

추세 확인Trend Confirmation (ADX ≥ 20)

Indicators

지수 이동평균EMA (Exponential Moving Average)

평균 방향성 지수ADX (Average Directional Index)

상승 방향성 지수DI+ (Positive Directional Index)

하락 방향성 지수DI− (Negative Directional Index)

평균 진폭ATR (Average True Range)

Market regimes

강한 상승 추세Strong Uptrend

약한 상승 추세Weak Uptrend

중립Neutral

약한 하락 추세Weak Downtrend

강한 하락 추세Strong Downtrend

Trade plans

진입가Entry Price

무효화 레벨Invalidation Level

목표가Target Level

손익비 (R:R)Risk-Reward Ratio (R:R)

매수 / 롱Long

매도 / 숏Short

진행 중Active

완료Completed

취소됨Canceled

§9

Disclaimer

All information provided by Squared is for educational and statistical analysis purposes only. Nothing on this platform constitutes investment advice, a recommendation to buy or sell, or financial counsel of any kind.

Past statistical patterns do not guarantee future results. Financial markets are inherently uncertain and all investments carry the risk of loss. Always make investment decisions based on your own judgment and appropriate risk management.

By using this service you agree that all investment decisions are entirely your own responsibility. Squared bears no liability for your trading outcomes.

← Example brief (BTC)Dashboard