Anomalies | KlickAnalytics

Anomaly Detection

262.06 3.43 (1.33%) 08/24/2026

Amazon.com, Inc. (AMZN)

Watch

0204570100

Composite anomaly score

AMZN scores 40/100 — Watch. Nothing has fired in the last 5 sessions — the most recent of the 13 events below is 16 sessions back, and is decaying out of the score. Everything below is recomputed from raw adjusted prices on this request; nothing is read from stored statistics.

262.06Last close

+0.04σvs 20d envelope

13Events in window

0Last 5 sessions

XLYBenchmark

286Bars analysed

24-Aug-2026Through

All detectors 13 Price Gap 4 Volume Surge 5 Volatility Spike 2 Options Flow n/aCorrelation Break 1 Liquidity Drop 0 Model-based 1

Shaded band is the rolling 20-session mean ±2σ of the adjusted close, computed ex-ante — each session is judged against the 20 sessions before it, never a window that already contains it. Markers show the sessions a detector fired.

Event log — Volume Surge (clear filter)

Date	Sessions ago	Detector	Event	z	Severity	Confidence	Score contrib.
30-Jul-2026	17	Volume Surge	Volume surge	+3.85	High	100%	7.8
Traded 78.66M shares against a 20-session baseline of 37.25M — 2.1x normal participation (3.9 sigma on log volume). Price that session: +3.90%. 37.25MExpected 78.66MActual +3.85σZ-score 20Baseline obs. +3.91%Return that day 235.50Close 14-session context (volume)
26-Jun-2026	40	Volume Surge	Volume surge	+6.21	Critical	100%	2.3
Traded 241.16M shares against a 20-session baseline of 49.18M — 4.9x normal participation (6.2 sigma on log volume). Price that session: +2.50%. 49.18MExpected 241.16MActual +6.21σZ-score 20Baseline obs. +2.50%Return that day 232.69Close 14-session context (volume)
18-Jun-2026	45	Volume Surge	Volume surge	+3.26	High	100%	1.1
Traded 75.62M shares against a 20-session baseline of 41.41M — 1.8x normal participation (3.3 sigma on log volume). Price that session: +2.90%. 41.41MExpected 75.62MActual +3.26σZ-score 20Baseline obs. +2.90%Return that day 244.39Close 14-session context (volume)
05-Jun-2026	54	Volume Surge	Volume surge	+2.14	Low	100%	0.2
Traded 55.61M shares against a 20-session baseline of 38.77M — 1.4x normal participation (2.1 sigma on log volume). Price that session: -3.06%. 38.77MExpected 55.61MActual +2.14σZ-score 20Baseline obs. -3.06%Return that day 246.03Close 14-session context (volume)
30-Apr-2026	79	Volume Surge	Volume surge	+2.90	Medium	100%	–
Traded 100.97M shares against a 20-session baseline of 44.68M — 2.3x normal participation (2.9 sigma on log volume). Price that session: +0.77%. 44.68MExpected 100.97MActual +2.90σZ-score 20Baseline obs. +0.77%Return that day 265.06Close 14-session context (volume)

How this is scored

Each detector z-scores one series against the symbol's own recent history and fires on the session the threshold is first crossed — a transition, not a state, with a 3-session cooldown, so one three-week volatility regime is one event rather than fifteen identical rows. Every rolling statistic is ex-ante.

Detector	Series scored	Baseline	Fires at	Weight	Status for AMZN
Price Gap	1-day % return	20 sessions	\|z\| ≥ 2.5	1.00	active
Volume Surge	log volume (adjusted)	20 sessions	z ≥ 2.0	0.90	active
Volatility Spike	log 5-session realized vol, annualized	60 sessions	z ≥ 2.0	1.00	active
Options Flow	put/call volume, IV30, IV30/IV90	60 sessions	\|z\| ≥ 2.0	0.90	unavailable — Options history for this symbol is 6 sessions; the detector needs 20 to form a baseline.
Correlation Break	20-session correlation to benchmark	120 sessions	z ≤ -2.0	0.80	active
Liquidity Drop	log traded value (price × volume)	60 sessions	z ≤ -2.5	0.70	active
Model-based	8-feature joint vector	up to 500 sessions	both models agree	0.60	active — Nightly batch, last run 23-Aug-2026.

Severity comes from |z| alone, so the same number means the same thing in every detector: Critical ≥ 4.0σ (base 40), High ≥ 3.0σ (28), Medium ≥ 2.5σ (18), Low below that (10). Confidence scales with how deep the baseline actually was, which is why options events on a thin history are discounted rather than trusted at face value.

The composite is contribution = base × weight × confidence × 0.5^(sessions_ago / 10), summed over events in the last 60 sessions, then squashed with score = 100 × (1 − e^(−sum/60)) so it saturates instead of running away. Bands: Nominal < 20, Watch 20–44, Elevated 45–69, Critical ≥ 70. This symbol's raw sum is 30.35. Engine v1.0.

Known limits — read before trading off this

Liquidity is a tape proxy. KA has no order-book depth, so the Liquidity Drop detector scores traded value (price × volume). Read it as "harder to get size done", not as a measured spread or book thinning.

Options history is shallow. options_iv_history is a forward-only nightly snapshot; for most of the universe the baseline is weeks rather than years, so options events carry low confidence and will strengthen on their own as the snapshot accumulates.

Correlation is US-session only. A same-dated close on a non-US venue is a different trading day, so benchmarking it against SPY or a sector ETF would measure the time zone rather than the decoupling. Non-US symbols simply do not run that detector.

The model-based detector asks a different question. The six univariate detectors are rolling — each session is judged against the 20 to 120 sessions immediately before it, so they fire throughout a window. The model is fitted once over the symbol’s whole stored history and marks the most extreme sessions in it, so a symbol whose violent period was eighteen months ago can legitimately show zero model events in a six-month window. A count of zero here means “nothing in this window ranks among the symbol’s most extreme sessions”, not that the detector failed.

The model-based detector cannot explain itself. The other six each score one series, so every flag decomposes into expected vs actual. IsolationForest and LocalOutlierFactor return a score over the joint vector and nothing decomposable; the drill-down shows the standardised features they saw, which is the nearest honest substitute. It is weighted below the six for that reason, it runs from a nightly batch rather than live, and a flag landing on a day another detector already explained contributes nothing to the score.

Detection is not prediction. A high score says the symbol is behaving unlike itself, not which way it goes next. The detectors have not been validated against a hand-labelled event set, and the score has not been tested for correlation with subsequent realized volatility.

HelpAMZN - Anomalies

Six statistical detectors on one symbol, scored 0-100, with every flag explained in plain English.

Why it helps

Triage in one number

A single 0-100 composite tells you whether a symbol is behaving unlike itself, before you open a single chart.

Every flag is explainable

Each event carries its z-score, expected vs actual, baseline depth and a plain-English sentence - never an unexplained red dot.

Nothing goes stale

Detectors recompute from raw adjusted prices on every request, so you are never reading a cached statistic from last month.

Know what the score cannot see

Thresholds, weights and the decay formula are printed on the page, along with any detector that has no data for this symbol.

How to use it

Pick a symbol and a window

3 months through 2 years. The engine always reads 160 extra sessions behind your window so the longest baseline is full before the first reported day.
Read the score first

Nominal under 20, Watch 20-44, Elevated 45-69, Critical 70 and above. It decays with a 10-session half-life, so it reflects what is happening now rather than six months ago.
Filter by detector

The chips carry live counts for the whole window; clicking one filters the event log without changing those counts.
Open an event

Click any row for expected vs actual, the z-score, how deep the baseline actually was, and a 14-session sparkline of the exact series that tripped.
Check it against the envelope

The shaded band is the rolling 20-session mean plus/minus 2 sigma of the close, computed ex-ante; markers show the sessions a detector fired.

What's in it

Six detector classes

Price gap, volume surge, volatility spike, options flow, correlation break and liquidity drop.

Composite score

Severity-weighted and recency-decayed across every detector, with the full formula disclosed on the page.

Sigma-envelope chart

Rolling 20-session mean plus/minus 2 sigma with breach markers and a volume subplot.

Transitions, not states

A detector fires when a threshold is first crossed, with a cooldown, so one long regime is one event rather than fifteen identical rows.

Stated limits

The tape-based liquidity proxy, the shallow options history and the US-session-only correlation are all called out rather than hidden.

FAQ

Does a high score mean the stock is about to move?

No. It says the symbol is behaving unlike itself, not which way it goes next. The detectors have not been validated against a hand-labelled event set.

Why is a detector showing n/a?

It has no usable data for this symbol. Correlation break needs a US-session benchmark, and options flow needs at least 20 sessions of stored implied-vol history.

Why does the shaded band lag the price?

It is ex-ante by design - each session is judged against the 20 sessions before it, never a window that already contains it, so a big day cannot inflate its own baseline and hide itself.

Is the liquidity detector reading the order book?

No. KA has no order-book depth. It scores traded value (price x volume) as a tape proxy, which reads as 'harder to get size done', not as a measured spread.

Video library Contact support

Global Symbol Search

Anomaly Detection

Ask the market a question. Get a calculated answer.

86,000+ instruments

A real analytics engine

It shows its working

Your own documents

Agents and workflows

MCP, CLI and API