Background
On the morning of October 12, 2018, Edwin Castro — who was wearing a GPS ankle bracelet and subject to a home inclusion zone — allegedly entered his neighbor’s apartment in Chelsea and committed a home invasion, armed robbery attempt, and sexual assault. His home was 152 feet from the victim’s building. The GPS bracelet generated a location data point at 7:02:38 a.m. showing the monitor inside the victim’s building with an accuracy range of twenty-six feet. DNA from a torn shirtsleeve matched Castro at a ratio of one in 529.9 octillion individuals. The victim identified her attacker as a stocky, dark-haired man who spoke Spanish with a Salvadoran or Honduran accent; Castro is from El Salvador with a stocky build and black facial hair.
The ankle bracelet generated location points every minute within the home inclusion zone (green) and every fifteen seconds when Castro was outside it (yellow for fewer than ten minutes out, red for longer). The GPS data showed Castro’s monitor moving from green to yellow to red beginning shortly before 7 a.m. — placing him outside the inclusion zone during the timeframe of the offense. A jury convicted him of home invasion, armed assault with intent to rob, indecent assault and battery, assault with intent to rape, and assault and battery by means of a dangerous weapon. Castro appealed, challenging admission of (1) the fact that he was subject to a home inclusion zone, (2) the GPS location data on reliability grounds, and (3) three aspects of the prosecutor’s closing argument.
The Court’s Holding
Affirmed on all issues.
On the home inclusion zone, the court held the trial judge acted within her discretion in admitting evidence that Castro was subject to home confinement, even if this arguably qualified as prior bad act evidence under Mass. G. Evid. § 404. The judge balanced probative value against prejudice, restricted testimony to “yes or no” when the probation officer used the phrase “home confinement,” gave an immediate limiting instruction that the jury not speculate about why Castro was monitored, and repeated the caution in the final charge. The evidence was not mere background — the home inclusion zone directly explained why Castro, if he were going to commit a crime while monitored, would logically do so within walking distance of his home. See Commonwealth v. Walker, 460 Mass. 590, 613 (2011). The jury heard nothing about the underlying reason for the monitoring condition.
On GPS reliability, the court applied the standard from Commonwealth v. Thissell, 457 Mass. 191, 198 (2010): GPS technology “is widely used and acknowledged as a reliable relator of time and location data.” The account manager for the GPS monitoring provider testified about multi-satellite triangulation, that 95 percent of data points are accurate within about thirty feet, and that signal strength affects per-point accuracy. Castro argued that because the victim’s building was only 152 feet from his home and some data points had a margin of error up to 141 feet, the entire data set was too imprecise to be admitted. The court rejected this framing: the critical 7:02:38 a.m. data point had an estimated accuracy range of twenty-six feet — well within which the victim’s building fell — and the expert’s testimony gave the jury the tools to evaluate each point individually. Accuracy concerns go to weight, not admissibility. The judge also appropriately excluded a separate request to use GPS data to calculate the defendant’s speed, consistent with Commonwealth v. Davis, 487 Mass. 448, 456 (2021).
On closing argument, the court found no error on three challenged comments. First, the prosecutor’s reference to a “hypothetical world” of defense speculation was a permissible call for the jury to base its verdict on the evidence, not a wholesale disparagement of the defense theory. See Commonwealth v. Lewis, 465 Mass. 119, 130 (2013) (prosecutor may attack a specific defense argument as a sham, but not the “whole defense”). Second, references to the Salvadoran backgrounds of both the victim and Castro were grounded in the evidence — the victim testified about recognizing the attacker’s accent — and did not appeal to harmful ethnic stereotypes. See Commonwealth v. Graziano, 368 Mass. 325, 331–32 (1975) (repeated “Godfather” references to Italian-American defendant impermissible). Third, the prosecutor’s invitation to the jury to “discard” GPS data points with larger accuracy ranges was permissible: the expert had explained that data points reflect varying satellite signal strength, providing a rational basis for the jury to treat high-accuracy and low-accuracy points differently.
Key Takeaways
- GPS ankle-bracelet monitoring evidence is admissible in Massachusetts under Thissell. Reliability challenges go to weight, not admissibility, as long as the Commonwealth lays a foundation through expert testimony about how the technology works and what per-point accuracy estimates mean.
- A worst-case accuracy margin does not invalidate the entire GPS data set. Courts evaluate the specific data points the prosecution relies upon. A data point with a twenty-six-foot accuracy range can place a defendant at a location 152 feet away; a data point with a 141-foot range may be of limited value but does not contaminate higher-quality points.
- Evidence that a defendant was subject to GPS home confinement is admissible to explain geographic proximity to the crime, but the reason for confinement is not. Defense counsel should move to exclude the underlying basis for electronic monitoring; prosecutors should agree to “home inclusion zone” rather than “home confinement” language to reduce prejudice.
- A prosecutor may invite the jury to discount or “discard” low-reliability data points when the expert has explained the scientific basis for differing accuracy levels within the same data set.
- References to a party’s national origin in closing are permissible when drawn from the evidence and relevant to identification, without invoking ethnic stereotypes. Graziano sets the outer boundary; this case falls well within it.
Why It Matters
Commonwealth v. Castro extends Massachusetts’s growing body of GPS evidence law into a genuinely close-quarters scenario. The 152-foot distance between Castro’s home and the victim’s building — barely half the length of a football field — meant that some GPS data points, taken at their worst-case margin of error, could not have distinguished inside the inclusion zone from outside it. The court’s answer is that per-point accuracy matters: the time-critical data point had a twenty-six-foot accuracy estimate, which was sufficient. Practitioners in electronic monitoring cases should closely examine the accuracy metadata for each data point the prosecution relies upon; a blanket challenge to the entire data set is unlikely to succeed, but a targeted challenge to specific low-confidence points may be more productive.
The closing argument rulings provide useful boundary guidance for both sides. The “hypothetical world” language was upheld because it targeted specific defense claims rather than the entire defense; calling the entire defense a “sham” would not have been. The “discard the outliers” instruction to the jury was upheld because expert testimony provided a rational scientific basis for doing so — prosecutors should not ask juries to ignore data without expert foundation. The national-origin reference underscores that accent and language evidence tied to victim identification is proper; arguments that invoke ethnic character or cultural propensity to crime are not.