Introduction
A symmetrical distribution is a probability or frequency distribution in which values occur at regular frequencies around a central point so that the left and right sides of a graph mirror one another. In many familiar cases (for example, the normal or “bell” curve), the mean, median and mode coincide at that center. Traders and analysts often use symmetry as an organizing assumption—e.g., to define a “value area” and to expect reversion to the mean—while being mindful that real-world data can and often do deviate from symmetry.
Key takeaways
– Symmetry means the two sides of a distribution are mirror images; mean = median = mode in many symmetric distributions (but not always).
– The normal distribution is the best-known symmetric distribution; the central limit theorem explains why sample means tend toward a normal shape as sample size grows.
– Traders use symmetrical assumptions to establish value areas (mean ± standard deviation) and to infer possible mean reversion.
– Symmetry is a simplifying assumption; many real datasets have skewness, fat tails, or changing centers over time, so always verify and manage risk.
What is a symmetrical distribution?
– Definition: A distribution is symmetric if it looks the same to the left and right of its center. In a perfectly symmetric distribution, for any distance x from the center, the value at center − x equals the value at center + x.
– Common examples: Normal distribution, uniform distribution, some binomial cases (when parameters produce symmetry).
– Note: A symmetric distribution may sometimes be bimodal (two equal peaks) and still be symmetric even though mode(s) differ from the mean/median.
What does a symmetrical distribution tell you?
– Balance of outcomes: Positive and negative deviations of the same magnitude from the center are equally likely.
– Central tendency alignment: For many symmetric distributions, mean, median and mode coincide, simplifying descriptive analysis.
– Mean-reversion expectation: If observations follow (or approximately follow) a symmetric curve, extreme values away from the center are expected to be less frequent and there is a statistical tendency (but not a guarantee) to revert toward the mean.
Tip
Use symmetry as a guiding assumption, not an absolute law. Check visually and statistically for symmetry before relying on normal-based rules, and combine symmetry-based rules with other indicators and risk controls.
Example — How traders use symmetrical distributions (value-area approach)
Traders commonly assume price action, over a chosen timeframe, approximates a symmetric distribution to define the asset’s “value area.” A simple workflow:
1. Select a timeframe (intraday: 30-minute bars; or multi-day/week/month).
2. Collect the price samples (ticks, close prices, mid-prices).
3. Compute the mean (average price) and standard deviation (σ).
4. Define the value area as mean ± 1σ (about 68% coverage for truly normal data).
5. Interpret deviations:
• If price moves above mean + 1σ (value area upper bound), the asset may be overvalued relative to that timeframe.
• If price moves below mean − 1σ, the asset may be undervalued.
6. Trade idea: consider mean-reversion trades when price violates the value area, but confirm with volume, trend indicators, fundamental context, and clearly defined risk management.
Caveat: The 68% rule depends on normality. Markets often display skewness and fat tails, so use this approach with confirmation and position limits.
Symmetrical distributions vs. asymmetrical distributions
– Symmetric: No skewness (skew = 0); left and right tails balanced. Examples: normal, uniform (flat), some binomial parameterizations.
– Asymmetric: Distribution is skewed to one side. Left (negative) skew has a long left tail; right (positive) skew has a long right tail. Example: log-normal (right-skew).
– Practical implication: Skew affects risk assessment—positive skew implies larger rare upside outcomes, negative skew implies larger rare downside outcomes.
Limitations of using symmetrical distributions
– Real-world deviations: Financial returns, incomes, and many other variables are often skewed or have fat tails.
– Time-varying centers: Markets can shift to a new mean, creating sustained asymmetry on the original timeframe.
– Small samples: Small datasets may appear symmetric by chance or hide true skewness.
– Over-reliance risk: Trading solely on value-area breaches without confirmation can be costly if the distribution is non-normal or the center is shifting.
– Extreme events: Symmetry-based models underestimate the probability of extreme events if tails are heavier than normal.
Relationship between mean, median, and mode in a symmetrical distribution
– In many symmetric distributions (including a perfect normal), mean = median = mode; they all fall at the center.
– Exceptions: A symmetric distribution can be multimodal (e.g., two equal peaks on either side of the center), in which case the mean and median lie at the vertical symmetry axis but the modes do not coincide with them.
Is the median symmetric?
– The median is the 50% midpoint: half the observations are below and half above. In a symmetric distribution, the median is at the center and produces mirror-image frequencies on both sides. In asymmetric distributions, the median is not the symmetry center and will differ from the mean and mode.
What is the shape of a frequency distribution?
– The “shape” is the visual pattern of frequency counts plotted across values (histogram or density plot). Shapes commonly observed: bell-shaped (normal), skewed left or right, uniform (flat), bimodal, or heavy-tailed. Shape is the first place to check for symmetry.
What is symmetric vs. asymmetric data?
– Symmetric data: Frequency/occurrence of values is balanced around a central point.
– Asymmetric data: Frequency of values is uneven; one tail extends more than the other (skew) or frequencies are irregular due to heterogeneity or noise.
Practical steps to check and use symmetry (for analysts and traders)
1. Visual checks
• Plot a histogram/density and look for mirror-image shape around a center.
• Use a boxplot to see skew (long whisker on one side suggests skewness).
• Q–Q plot against a normal distribution: deviations from the straight line indicate departures from normality/symmetry.
2. Compute simple summary statistics
• Mean, median, and mode: compare them—large differences suggest skew.
• Calculate sample skewness (Pearson’s or Fisher’s): skew ≈ 0 suggests approximate symmetry; positive means right skew, negative means left skew.
• Compute kurtosis to evaluate tail-weight (fat tails reduce normal-based confidence).
3. Perform formal tests (if needed)
• Normality/symmetry tests: Shapiro–Wilk, D’Agostino’s K-squared, Jarque–Bera, or other normality/symmetry assessments. These provide p-values but are sample-size sensitive.
• Use robust tests or bootstrap when sample size is small or data are heavy-tailed.
4. Robust alternatives if symmetry fails
• Use median and median absolute deviation (MAD) instead of mean and standard deviation.
• Apply transformations (log, Box–Cox) to reduce skewness; interpret results carefully.
• Model tails separately (e.g., generalized Pareto for extremes) if you need accurate tail risk.
5. Sample size and central limit theorem (CLT)
• The CLT: sample means tend to a normal (symmetric) distribution as sample size increases, even if the underlying population is skewed. This helps justify normal assumptions for sample means but not always for raw observations or small samples.
6. Trading and decision-making checklist
• Confirm symmetry visually and statistically.
• Define the timeframe and ensure enough observations for reliable estimates.
• Compute mean and standard deviation; define value areas but treat the ±1σ rule as approximate.
• Confirm breaches with volume, trend, volatility regime, and fundamental context.
• Use stop-losses and sizing rules—symmetry-based mean-reversion strategies can fail during regime shifts.
Quick reference: interpreting skewness
– Skewness ≈ 0: distribution approximately symmetric.
– Skewness > 0: right-skew (long right tail).
– Skewness < 0: left-skew (long left tail).
(Thresholds for “≈ 0” depend on sample size and variability.)
Practical example (simple numerical illustration)
– Suppose you collect 100 price samples across a trading session.
– Compute mean = 100, standard deviation = 2.
– Value area ≈ 98 to 102 (mean ± 1σ). If price trades at 103.5 (well above 1σ), consider it outside the 68% range; evaluate reversion-to-mean possibilities, but verify with volume and recent trend—price could be trending upward to a new mean.
Conclusion
Symmetrical distribution is a useful and intuitive concept that simplifies analysis by aligning central tendency measures and enabling familiar rules (e.g., the 68–95–99.7% rules for normal distributions). It is widely used in trading and statistics, but should be applied cautiously: always check for skewness, changing means, fat tails, and limited sample size. Use robust measures, visual diagnostics and confirmatory indicators before acting on symmetry-based inferences.
References
– “Symmetrical Distribution.” Investopedia. (source of definitions, trading application, and conceptual guidance)
(Continuing from the previous material)
Additional sections
Statistical measures and formal tests for symmetry
– Skewness
• Definition: skewness quantifies asymmetry of a distribution. A skewness near 0 suggests symmetry; positive skewness indicates a longer right tail; negative skewness indicates a longer left tail.
• Sample calculation (one common estimator): g1 = [n/(n-1)(n-2)] * Σ((xi – x̄)^3) / s^3, where x̄ is the sample mean, s is sample standard deviation, and n is sample size.
• Interpretation rules of thumb: |skewness| 1 strong skew (these are heuristic).
– Kurtosis
• Measures tail thickness and peakness. For symmetry assessment, kurtosis is secondary but helpful: two symmetric distributions can have different kurtosis (e.g., normal vs. platykurtic/ leptokurtic).
– Formal normality / symmetry tests
• Shapiro–Wilk: powerful for small to moderate sample sizes; tests whether data come from a normal distribution.
• Kolmogorov–Smirnov (one-sample) and Anderson–Darling: compare sample distribution to a specified distribution (often normal); AD is more sensitive to tails.
• D’Agostino’s K-squared: test that uses skewness and kurtosis.
• Note: a “normality” test is not exactly a test for symmetry only, but normality implies symmetry. You can also test skewness = 0 directly with a t-test on the skewness estimator.
Graphical methods for detecting symmetry
– Histogram with fitted density: quick visual check for mirrored sides.
– Kernel density estimate (smoothed density): useful when histogram binning hides structure.
– Boxplot: reveals tail lengths and outliers; a symmetric distribution tends to have a centered median and equal whisker lengths.
– Q–Q plot against a normal distribution: symmetric shape that follows the reference line indicates approximate normality (and hence symmetry); systematic curvature indicates skew.
– Mirror plot: split the data at the median and plot mirrored densities to inspect similarity.
Practical, step-by-step procedure to assess symmetry (for analysts/traders)
1. Collect data for the period and variable of interest (e.g., intraday price ticks, daily returns).
2. Visualize: draw histogram and a KDE; make a Q–Q plot.
3. Compute basic statistics: mean, median, mode(s), standard deviation, sample skewness, and kurtosis.
4. Run formal tests if needed (Shapiro–Wilk, Anderson–Darling, or test skewness = 0).
5. Check for outliers and structural breaks (plot the time series; look for regime changes).
6. If distribution is approximately symmetric, you may apply normal-based rules (e.g., ±1 SD value ranges). If asymmetric, consider transformations or nonparametric/robust methods.
7. For trading strategies, always require confirmation by other indicators and include explicit risk controls (size, stop-loss, time horizons).
8. Reassess periodically, since symmetry on one time scale or sample window may not hold in another.
Concrete example — applying symmetry to a trading value-area decision
– Scenario: You analyze 30-minute price bars for a stock during a trading session (n = 13 bars). Suppose computed statistics: mean price = $100, standard deviation = $4.8, sample skewness ≈ 0.12 (near zero).
– Value area (approximate 68% rule for normal-like data): mean ± 1 SD = $95.2 to $104.8. Under the symmetric assumption, about two-thirds of price observations should fall inside that range.
– Trading interpretation:
• Price moves above $104.8: possible short/mean-reversion opportunity if other indicators align (volume spike, resistance, RSI overbought).
• Price moves below $95.2: possible long/mean-reversion if supported by confirmations.
– Practical cautions:
• Use stop-loss beyond recent structure, because symmetry is not a guarantee—structural breaks can shift the mean.
• Confirm with volume profile, order-flow, or momentum indicators.
• Recompute mean/sd frequently (sliding window) because the value area can shift intra-session.
Non-financial example — exam scores
– Suppose 200 students’ exam scores are collected and a histogram resembles a bell curve with mean = median = 72. Skewness ≈ 0, Q–Q plot roughly linear: conclude the distribution is symmetric and use mean and SD to discuss expected ranges.
– If the scores are right-skewed (many low scores and a few high achievers), median and percentiles better summarize typical performance than the mean.
Dealing with asymmetry (practical approaches)
– Transformations:
• Log transform: often useful for positive-valued data (prices, incomes) to reduce right skew.
• Square-root or Box–Cox: other transformations to approach symmetry; Box–Cox includes a parameter to optimize.
– Use robust statistics:
• Median and interquartile range (IQR) instead of mean and SD; quantile-based risk measures (e.g., Value at Risk by percentiles).
– Nonparametric methods:
• Bootstrapping for confidence intervals without assuming symmetry.
– Model distributions that capture skew:
• Fit skew-normal or log-normal, or use mixture models if data have multiple modes.
Limitations and pitfalls when relying on symmetry
– Sample size: small samples can give misleading skewness estimates; central limit theorem helps for means but not for raw-data symmetry.
– Outliers: single extreme values can make a symmetric population appear asymmetric in a sample.
– Non-stationarity / regime shifts: means and spreads can move over time; symmetry on one window does not ensure future symmetry.
– Multimodality: data may be symmetric overall but composed of two or more subpopulations with different centers—this can mask meaningful structure.
– Over-reliance in trading: assuming symmetry (and reversion to mean) can be dangerous during trending markets or structural breaks; always combine with other analyses.
Relationship of mean, median, and mode in symmetric vs asymmetric distributions
– Symmetric distributions: mean = median = mode (for a single-peaked symmetric distribution). Example: normal distribution.
– Asymmetric distributions: mean, median, and mode differ; their order depends on direction of skew:
• Right (positive) skew: mode < median < mean.
• Left (negative) skew: mean < median < mode.
– The median is the 50th percentile; for any symmetric distribution (about point m), m is the median and splits the data into equal halves.
Practical computing commands and small recipes
– Excel: AVERAGE(range), MEDIAN(range), SKEW(range), KURT(range). Use histogram tool or chart options for visual checks.
– Python (pandas/scipy/statsmodels):
• mean = data.mean(); median = data.median(); skew = data.skew()
• from scipy import stats; stats.shapiro(data) for Shapiro–Wilk; stats.anderson(data, dist='norm')
• from statsmodels.graphics.gofplots import qqplot; qqplot(data, line='s')
– R:
• mean(x); median(x); library(moments); skewness(x); library(nortest); shapiro.test(x)
• qqnorm(x); qqline(x)
Additional worked example — skew and transformation
– You have daily returns with sample skewness = 1.2 (right-skewed). Steps:
1. Visualize histogram and Q–Q plot to confirm right tail stretch.
2. Consider log(1 + return) only if returns are positive or if modeling prices; for returns that include negatives, a signed Box–Cox or other transformation may be applied carefully.
3. If transformation doesn’t produce symmetry, adopt quantile-based risk metrics (e.g., median absolute deviation, tail percentiles) instead of mean ± SD rules.
When and why symmetry matters (summary)
– Symmetry simplifies statistical modeling and inference: many classical techniques (t-tests, confidence intervals for means) rely on approximate normality of sample means.
– In finance, symmetry assumptions underpin mean–reversion strategies and value-area rules (±1 SD bands). They can offer quick benchmarks for over/undervaluation.
– However, markets and many real-world datasets are not perfectly symmetric and can exhibit heavy tails, skew, and regime shifts; robust checks and risk controls are essential.
Concluding summary
Symmetrical distributions are a useful and familiar concept: they simplify interpretation because the central tendency measures (mean, median, mode) align, and standard deviation bands have intuitive coverage properties (e.g., ~68% within ±1 SD for a normal curve). In practice, evaluate symmetry with both visuals (histograms, Q–Q plots) and statistics (skewness, normality tests). For traders, symmetry can define a value area and inform mean-reversion ideas, but it must be paired with confirmation signals and solid risk management because symmetry can break down—especially across different time scales or during regime changes. When data are asymmetric, prefer transformations, robust statistics, or models that explicitly allow skew so decisions are based on accurate summaries of the true distribution.
Sources and further reading
– Investopedia, “Symmetrical Distribution”
– Standard texts on probability and statistics (for skewness/kurtosis definitions and central limit theorem)