iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
A sudden plunge in a price chart can be a bad record, a change in symbol or units, or a corporate action—but it can also be a real market move. Treat an outlier as a reason to investigate, not as proof that the data is wrong. Before changing a value, check the record, the instrument, the event and independent observations, then document what you did.
How do I clean stock price data without deleting a real crash?
Use a review-first workflow. Keep the original series intact, flag questionable observations, and make a correction or exclusion only when you have a defensible reason. The right checks depend on what the data represents—trades, quotes, daily bars or a reference rate—and on the instrument, venue and sampling interval.
- Preserve the original. Save the raw file or feed and record its source, retrieval date and version. Do not overwrite the only copy.
- Validate each record. Check for a present, numeric price; confirm it is positive when positivity is appropriate for that instrument; and inspect the timestamp for missing, malformed or implausible values. These are practical examples, not universal validity rules for every financial instrument.
- Confirm what the record identifies. Verify the symbol or instrument identifier, currency, scale, venue and timestamp convention. Keep the source and identifier with the data so a flagged point can be traced.
- Check the series convention and event calendar. Determine whether prices are raw, split-adjusted or part of a return or index series. Look for corporate actions or other relevant events around the apparent jump.
- Compare context and independent observations. Inspect nearby records and, where available, compare with an independent source or venue. First establish that the feeds refer to the same instrument, period, currency and price definition.
- Record the decision. Keep the original value, the action taken, the reason, the rule or evidence used, and the data and code versions. If evidence is insufficient, leave the value in place and mark it for review.
For influential information, the U.S. Securities and Exchange Commission’s Final Data Quality Assurance Guidelines call for identifying supporting sources where appropriate and disclosing data sources, quantitative methods and assumptions so others can reproduce the analysis.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsWhat should you check when a price looks wrong?
Structural checks: is the record interpretable?
- Is the price present, numeric and plausible for the instrument’s data convention?
- Is the timestamp valid, in the expected time zone or convention, and paired with the intended observation interval?
- Does the symbol identify the same instrument throughout the series, including before and after any symbol change?
- Are currency, unit scale, venue and price definition consistent?
A price recorded in a different currency or scale can look like a dramatic market move even when the underlying value did not change comparably. Retain enough source and identifier information to establish exactly what each observation represents.
#1 Best Overall
- Wiley
- Language: english
- Book - storytelling with data: a data visualization guide for business professionals
Continuity and context: did something happen?
- Compare the observation with nearby prices and the size and direction of the surrounding move.
- Look for an event that could explain a discontinuity, including a corporate action.
- Check whether the series is raw, split-adjusted, or based on returns or an index; do not compare unlike series without accounting for that distinction.
NYSE Regulation lists reverse splits and reorganizations among corporate actions affecting listed securities and provides corporate-action and market-watch information. An event-related discontinuity is not automatically a corrupt observation. Whether you need a price-return or total-return series depends on the question you are analyzing. The source material cited here does not establish a universal adjustment formula across asset types, so do not apply one by assumption.
Cross-source checks: does another observation support it?
- Compare with a genuinely independent feed or venue when one is available.
- Confirm that sources cover the same instrument, time, venue scope and price measure.
- Consider whether thin trading, differing market hours or stale observations could explain a disagreement.
A second source is useful only if it measures a comparable thing. Agreement can support an observation; disagreement is a prompt to examine source definitions and timing, not automatic proof of which source is correct.
Action log: can someone reproduce the treatment?
- Log the flagged timestamp and instrument identifier.
- Record the raw value, any corrected value or exclusion, and the reason.
- Document filters, assumptions, comparison sources and the relevant data or code version.
- Preserve raw inputs and the transformation history so another analyst can reproduce the result.
Why can a generic outlier threshold mislead?
A statistical rule can identify an unusual value, but unusual does not mean erroneous. A threshold suitable for one instrument, sampling frequency or calculation may discard legitimate jumps—or fail to catch a bad record—in another. Illiquid markets and differences between trades, quotes, daily bars and reference rates also change what counts as a meaningful comparison.
Choose a validation method by asking what it checks and what happens when it fires:
- Record validity: Does the rule catch missing, malformed or otherwise invalid fields?
- Statistical extremeness: Is it calibrated for this asset, interval and analytical objective, and does it flag or automatically alter data?
- Cross-source disagreement: Are the comparison observations independent and sufficiently comparable?
- Reproducibility: Can you explain the rule and recreate each change from retained inputs?
For general cleaning, use a threshold to prioritize review unless you have established that automatic treatment is appropriate for the defined data and purpose. No universal bad-tick threshold or prevalence rate is established by the sources cited here.
What does one documented price-validation method do?
A methodology filed with the U.S. Commodity Futures Trading Commission (CFTC) for a specific bitcoin and ether reference-rate calculation illustrates how several controls can work together. It checks transaction validity, compares venue volume-weighted average prices (VWAPs) within a partition, and uses a separate benchmark comparison with a fallback. These controls describe that calculation; they are not general stock-price cleaning standards.
- The methodology excludes a venue VWAP when it deviates by more than 10% from the median of the venue VWAP set.
- For the final reference rate, a deviation greater than 5% from a separate Lukka Global VWAP triggers calculation failure and fallback.
Those percentages have meaning only within the methodology’s stated calculation and comparison sets. They should not be transplanted as automatic deletion rules for ordinary stock histories or other financial series. The CFTC-hosted filed methodology also describes checks for malformed messages and absent, non-numeric or non-positive prices, as well as future-dated execution times. Such checks are examples from this method, not universal rules for all instruments.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →What do formal data-quality rules require?
Regulatory requirements illustrate the value of explicit detection and reconciliation, but their scope matters. In the UK, the Financial Conduct Authority’s rules for consolidated tape providers (CTPs) address identifying incomplete or likely erroneous trade reports, price and volume alerts for equities, user flagging mechanisms, and periodic reconciliation between reports received and published. The cited provisions show effective dates of 31 July 2026; they are obligations for that regulatory context, not requirements on every analyst or data user.
The FCA provision states: “A CTP must set up and maintain appropriate arrangements to identify on receipt trade reports that are incomplete or contain information that is likely to be erroneous”. Read that requirement in its context: FCA Handbook MAR 9.2B.
The SEC’s guidance is aimed at data quality in its own context. It states: “The Commission is committed to disseminating information that is accurate, clear, complete and unbiased both in its content and in its presentation.” The SEC guidelines also describe source, method and assumption disclosure that supports reproducibility.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
