IEX vs SIP Market Data: Why Stock Prices and Volume Differ

IEX and SIP are different sources of stock-market observations. IEX data describes activity on one exchange; consolidated SIP data combines reported activity across US exchanges. Different coverage can produce different trades, volume, candles, and indicator values for the same symbol and minute. When two charts disagree, first compare their feed, timestamps, session, and aggregation rules rather than assuming one strategy calculation is broken. Alpaca’s market data FAQ explains its feed selection and access rules.

This guide provides a diagnostic workflow for research and paper testing. It does not establish that either feed produces profitable signals. Trading involves risk; this is educational information, not financial advice.

What IEX vs SIP means for a stock chart

A feed is part of the meaning of an observation. A last trade from one exchange answers a narrower question than a consolidated last trade. The same distinction applies to volume: a venue-specific volume total cannot be interpreted as a total for the wider market. Two observations can both be valid while describing different sets of transactions.

Think of each chart value as a labeled record rather than a universal number. Its label should identify the symbol, source, event time, session, adjustment policy, and aggregation interval. If those labels are missing, the displayed precision can be misleading. A candle showing two decimal places still leaves unanswered which trades contributed to it. Before comparing values, write down the actual data request or chart settings that produced them. This establishes what each value means and makes later disagreements reproducible instead of dependent on screenshots alone.

Why candles and volume can disagree

Consider a fictional one-minute interval. Feed A observes trades at 20.00, 20.02, and 20.01. Feed B includes additional trades at 19.98 and 20.05. Their highs, lows, and volume totals can differ without an arithmetic error. The extra observations change the input, so matching calculations do not necessarily produce matching outputs.

This example is illustrative, not measured exchange data. It shows why comparing the final candle is insufficient. Where your data license permits it, compare the underlying event records and the rules used to group them. Check whether the minute is labeled by its beginning or end, whether late observations can revise a candle, and whether the platform excludes particular transaction conditions. A timestamp mismatch can look like a feed mismatch. A feed mismatch can look like an indicator bug. Diagnose each layer separately and preserve the original payload before changing code or settings.

A comparison table for research decisions

Choose the source according to the question being tested and the rights available to your account. A development exercise that checks parsing and reconnect behavior has different requirements from an intraday strategy that depends on a specific definition of market volume. Avoid buying data simply because a chart looks more active. Define the missing observation first, then determine whether a different feed supplies it.

How feed differences affect VWAP research

VWAP weights observed prices by observed volume. Changing the observations changes the inputs to that weighted average. If one dataset contains a narrower set of trades, its VWAP need not match a chart built from a broader set. This follows from the calculation; it is not evidence that one displayed line is manipulated.

Before studying a VWAP rule, specify the source, reset time, session, and price convention. A strategy using regular-session VWAP should not be assessed against another chart’s extended-session value without accounting for the difference. Record whether your reference is calculated from individual trades or bars, because those are separate implementations. The VWAP trading guide provides broader strategy context. For this diagnostic task, concentrate on reproducing the indicator from the exact input records. If you cannot reproduce it, keep the discrepancy open instead of choosing the value that makes the historical trade look better.

Explicit feed selection makes comparisons reproducible

Alpaca documents a feed parameter for historical requests and explains that default feed selection can depend on subscription access. Its FAQ also distinguishes access to recent SIP observations from older historical requests. Verify the current rules for the endpoint and account you actually use rather than treating a successful request as proof of every data entitlement. Alpaca market data FAQ.

For a reproducible comparison, retain a sanitized request record with the selected feed, start and end time, timeframe, symbol, and pagination behavior. Save the response metadata and the time you retrieved it. Never save API secrets in an article, screenshot, or shared research notebook. A research result that says only “Alpaca data” is incomplete: readers cannot determine which observation set generated it. Make the selection explicit wherever the provider supports that control, and record the resolved feed when a platform chooses it automatically.

Use a five-step discrepancy review

1. Select one symbol and a short interval with a clearly stated timezone.
2. Match session boundaries, interval labels, adjustment settings, and symbol identity.
3. Retrieve each authorized feed explicitly and retain sanitized metadata.
4. Compare event coverage, then candle construction, then indicators.
5. Document the remaining discrepancy and the evidence needed to resolve it.

Start with a small interval because it is easier to inspect than a full year of data. Once you understand one mismatch, apply the same checks to a broader sample. Do not assume that a finding for one symbol or session explains every disagreement. Keep records of both resolved and unresolved cases. Resolved cases teach you which configuration rules matter; unresolved cases identify where you need provider clarification, additional permissions, or a better observation method. This produces a diagnostic report rather than a preferred-feed advertisement.

Compare decisions as well as indicator values

An indicator difference matters operationally when it changes the decision rule being examined. Suppose a hypothetical rule flags a symbol when volume exceeds its own historical baseline. Compare whether the two feeds produce the same flag, the same flag time, and the same later invalidation under separately consistent baselines.

Do not divide a venue-specific current volume by a consolidated historical average and call the result relative volume without explaining the mixed inputs. Build the baseline from the same definition used for the current observation. Then compare complete pipelines across feeds. Record cases where decisions agree despite different values and cases where a threshold changes the outcome. This prevents small numerical differences from being exaggerated and large behavioral differences from being overlooked. Keep the rule frozen during the comparison; adjusting the threshold repeatedly to fit the sample makes the conclusion harder to evaluate.

Keep historical and forward tests consistent

A backtest and a forward paper test should document their data sources separately. If their sources differ, treat the change as an experimental variable. Preserve the historical configuration, describe the forward configuration, and examine whether input differences explain decision differences before blaming the strategy itself.

The backtesting guide discusses historical evaluation. Add a feed section to that process: coverage, timing, session, adjustments, and missing-data policy. For forward review, record gaps and stale observations when they happen, not just at the end of a favorable session. A clean historical file may conceal operational problems that are visible in real time. Conversely, a forward disagreement might reflect an input change rather than a new market regime. Neither test proves live performance, and switching feeds does not remove execution, model, or selection risk.

How this fits a Tradewink research workflow

Use a watchlist and a signal review to identify the question you want to investigate. If a displayed volume or indicator differs from another source, record the source labels and timestamps before interpreting the difference as a stronger or weaker trading idea. A private paper-track decision records your reasoning; it does not establish a simulated brokerage fill.

The paper-trading workflow explains that distinction. Public subscriptions are paper-only; separately approved private beta accounts may submit live broker orders. This article does not claim that Tradewink exposes every feed-selection setting or offers a consolidated-data entitlement to every account. Confirm the actual data source and permissions for the view you are using. If the information is unavailable, label it unknown and request clarification instead of assigning the chart an assumed source.

Limitations of a feed comparison

A feed audit can identify input differences, but it cannot determine whether a strategy has an economic edge by itself. Wider coverage does not guarantee a useful model, accurate order timing, or realistic fills. A perfectly reproduced indicator can still support an unprofitable rule.

Provider documentation can change, and access may differ by account or endpoint. This guide checked the linked documentation on October 4, 2026; it does not promise future access terms. It also does not measure IEX market share, compare latency across vendors, or report a tested performance improvement. Those questions require their own timestamped datasets and methodology. Keep the conclusion proportional to the evidence: a matched observation explains a chart value, a matched pipeline explains a decision, and neither alone establishes what would happen to a real order in a future market.

Frequently Asked Questions

Is IEX data the same as SIP data?

No. The feeds cover different observation sets. Compare source coverage before comparing volume, candles, or indicators.

Why does my Alpaca volume differ from another chart?

Feed coverage is one possible cause. Also check interval timing, session boundaries, adjustments, and aggregation rules before deciding which cause applies.

Should I use SIP for every strategy?

Define the data requirement first. Account entitlement, coverage, reproducibility, and the strategy’s sensitivity to input differences determine what needs investigation; no feed guarantees results.

Can different feeds change a VWAP signal?

Yes, different price and volume inputs can change the computed value and a threshold decision. Test the complete rule with internally consistent inputs.

Does a matching chart prove my backtest is realistic?

No. Matching inputs do not validate execution assumptions, costs, selection methods, or future outcomes.