SHEPHERD, Circuit Judge.
This case is before us after our limited remand in United States v. Engelmann, 701 F.3d 874 (8th Cir.2012). Marc Robert Engelmann was convicted of conspiracy to commit bank and wire fraud under 18 U.S.C. § 371, bank fraud under 18 U.S.C. § 1344, and wire fraud under 18 U.S.C. § 1343. The district court1 sentenced him to 36 months imprisonment and ordered him to pay a total of $392,937.73 in restitution to three different financial institutions. Engelmann appealed his conviction and sentence, and we ordered a limited remand for the district court to conduct an eviden-tiary hearing concerning one of Engel-mann’s arguments and to reconsider En-gelmann’s motion for a new trial after that hearing. See Engelmann, 701 F.3d at 879. We retained jurisdiction to address all of Engelmann’s points on appeal after these further district court proceedings.2 Id. The district court held the evidentiary hearing and issued an opinion. We now affirm Engelmann’s conviction and sentence.
I.
Engelmann was a real estate attorney and represented a seller in nine different transactions involving a “dual price” purchasing agreement. Through these agreements, the buyers and sellers provided lenders with inflated sales prices to secure higher loan amounts, and then the buyers pocketed the difference between the inflated and actual amounts. The buyers went into first-payment default on all nine mortgages, and the properties were sold at sheriffs sales or short sales.
Engelmann’s defense at trial was that he did not have the requisite intent to defraud because he thought the lenders knew of the dual pricing scheme. He requested the following jury instruction:
One of the issues in this case is whether the defendant acted in good faith. Good faith is a complete defense to the charge of conspiracy to commit bank and wire fraud (Count 1), bank fraud (Counts 2 and 3) and wire fraud (Counts 4 thru 9) if it is inconsistent with the defendant acting to conspire with one or more other persons to commit bank and wire fraud under element 1 of Count 1, or the intent to defraud under element 2 of the bank fraud counts and element 2 of the wire fraud counts.
Evidence that the defendant acted in good faith may be considered by you, together with all the other evidence, in determining whether or not he acted with intent to defraud.
Fraudulent intent is not presumed or assumed; it is personal and not imputed. One is chargeable with his own personal intent, not the intent of some other person. Bad faith is an essential element of fraudulent intent. Good faith constitutes a complete defense to one charged with an offense of which fraudulent intent is an essential element. One who acts with honest intention is not chargeable with fraudulent intent. Evidence which establishes only that a person made a mistake in judgement or an er[1009]*1009ror in management, or was careless, does not establish fraudulent intent. In order to establish fraudulent intent on the part of a person, it must be established that such person knowingly and intentionally attempted to deceive another. One who knowingly and intentionally deceives another is chargeable with fraudulent intent notwithstanding the manner and form in which the deception was attempted.
Appellant’s Addendum 27-28.
The language in the first two paragraphs tracks the Eighth Circuit’s model good-faith jury instruction for fraud cases. See Eighth Circuit Manual of Model Jury Instructions: Criminal 9.08 (“Model Instruction 9.08”). The language in the final paragraph is from a jury instruction that we upheld in United States v. Ammons, 464 F.2d 414, 417 (8th Cir.1972), and that the model instructions reference as potential language to include “if appropriate.” See Model Instruction 9.08; id. n. 2.
The district court essentially gave the first two paragraphs of Engelmann’s requested good-faith instruction but omitted the third paragraph. R. at 77. The other instructions on the elements of the underlying conspiracy and fraud offenses explained Engelmann could be convicted only if he “voluntarily and intentionally joined in the agreement or understanding” while knowing “the purpose of the agreement or understanding,” R. at 59; that “[a] person who has no knowledge of a conspiracy but who happens to act in a way which advances some purpose of one, does not thereby become a member,” R. at 60; and that Engelmann must have acted “knowingly” and with “intent to defraud,” R. at 66-67.
After the jury began deliberating, it asked the court to define good faith. The court denied the jury’s request over En-gelmann’s objection and directed them to review the jury instructions they already had available. The jury ultimately found Engelmann guilty on all counts.
At Engelmann’s sentencing hearing, En-gelmann argued that the court could not enhance his base sentencing level due to the amount of loss involved in the crimes because the complexity of the sub-prime mortgage market precluded any accurate loss calculation. The district court rejected this argument and increased Engel-mann’s base offense level by 12 for the amount of loss.
Engelmann also argued at sentencing that the court could not award restitution under the Mandatory Victims Restitution Act because the lending institutions were not real “victims” under the statute and because the government could not prove restitution amounts. The district court rejected each of these arguments and ordered Engelmann to pay restitution to three companies in the following amounts: New Century Liquidating Trust ($226,-587.34), JP Morgan Chase ($108,560.48), and Lehman REO-ALS ($57,839.91).
Meanwhile, after the jury’s verdict, a trial observer contacted the district court to report that he had seen two of the government’s witnesses, FBI Special Agents Jeff Huber and Jim McMillan (collectively “the Agents”), speaking outside the courtroom about testimony that Agent Huber had given at trial.3 A witness sequestration order was in place throughout the trial. Agent Huber was the government’s designated case agent and consequently remained in the courtroom throughout the trial. Agent McMillan was called as a rebuttal witness following En-gelmann’s trial testimony. Both Agents [1010]*1010testified that Engelmann essentially confessed to the fraud when they jointly interviewed him during their investigation. During closing arguments, the government emphasized that Agent McMillan’s testimony about Engelmann’s confession was especially credible because he had not heard Agent Huber’s testimony before giving his own.
Engelmann moved for an evidentiary hearing concerning the Agents’ conversation and for a new trial, arguing that the conversation violated the court’s witness sequestration order and prejudiced him. Without holding an evidentiary hearing, the district court denied the motion for a new trial. We vacated and remanded for the court to hold an evidentiary hearing on this issue, make supplemental findings of fact, and then reconsider Engelmann’s motion for a new trial. See Engelmann, 701 F.3d at 879.
At the evidentiary hearing following remand, three trial observers testified that they saw the Agents speaking outside the courtroom during a trial recess. Two testified that they could not hear anything that was said. The third witness, the individual who originally contacted the district court, testified that he could only remember overhearing someone in the Agents’ vicinity say, “We had Engelmann at his office.”
The Agents and one of the government’s trial attorneys, who was with the Agents during the conversation in question, also testified at the evidentiary hearing. All three testified that Agent Huber did not disclose details of his trial testimony to Agent McMillan during this conversation. Agent McMillan testified that Agent Huber did tell him generally that “Engel-mann denied making certain statements during our interview,” but Agent McMillan said that Agent Huber never disclosed details of any witness’s testimony or tried to influence Agent McMillan’s upcoming testimony. Agent Huber and the government’s trial attorney further testified that they never disclosed details of any witness’s trial testimony to Agent McMillan at any other point during the trial.
Additionally, Agent McMillan testified that no one initially informed him that he should not observe other witnesses’ trial testimony. As a result, he entered the courtroom on several occasions before his own testimony on rebuttal and observed brief portions of the testimony of one of Engelmann’s co-conspirators and of an expert witness called by the defense. He also observed a minute or less of Engel-mann’s trial testimony before Agent Huber approached him and told him to leave the courtroom because he might be called as a rebuttal witness. According to his testimony, this was the first time that anyone informed him he should not be in the courtroom during other witnesses’ testimony.
The district court found that Agent McMillan violated the sequestration order by being in the courtroom during portions of three witnesses’ testimony. District Ct. Order 11, Mar. 6, 2013, ECF No. 149. However, the court found that the violation did not prejudice Engelmann because the testimony that McMillan overheard “bore no direct relationship to” and was not “in any way pivotal to” to the testimony that he ultimately gave. Id. The district court further held that because there was no evidence that the out-of-court conversation between the Agents involved disclosure of specific trial testimony, and because Agent McMillan’s testimony was on the discrete topic of his interview of Engelmann and was fully consistent with his contemporaneous notes concerning that interview, the conversation neither violated the sequestration order nor prejudiced Engelmann. Id. at 12-13. The district court conse[1011]*1011quently denied Engelmann’s motion for a new trial. Id. at 14.
With the evidentiary hearing completed, this case returns to us following our limited remand.
II.
Engelmann argues the district court erred by (1) refusing to give his requested good-faith instruction; (2) denying his motion for a new trial based on violations of the witness sequestration order; (3) increasing his base offense level by 12 for the amount of loss involved in his convictions; and (4) ordering him to pay $392,937.73 in restitution.
A.
Engelmann argues that the district court’s good-faith instruction did not sufficiently define good faith or explain that fraudulent intent must be personal to En-gelmann and cannot be imputed from co-conspirators. He asserts that the jury’s request for the court to define good faith illustrates the instruction’s inadequacy.
“We review the district court’s jury instructions for abuse of discretion and will affirm if the instructions, taken as a whole, fairly and adequately submitted the issues to the jury.” United States v. Whitehill, 532 F.3d 746, 751 (8th Cir.2008) (internal quotation and alteration marks omitted). “Defendants are entitled to a theory of defense instruction if it is timely requested, is supported by the evidence, and is a correct statement of the law, but they are not entitled to a particularly worded instruction.” Id. at 752. District courts have “considerable discretion in framing the instructions,” and an instruction is sufficient if it “adequately and correctly covers the substance of the requested instruction.” Id. (internal quotation marks omitted). Good-faith instructions are not evaluated in a vacuum, but rather must be evaluated “by looking at [the jury instructions] as a whole and in the context of the trial.” Id.
Wq conclude that the district court did not abuse its discretion in giving an abbreviated form of Engelmann’s requested good-faith instruction. Although En-gelmann claims the jury’s request for the court to further define good faith shows the instruction given was inadequate, the request can also be construed as demonstrating that the instructions did, in fact, “direct[] the jury’s attention to the defense of good faith.” See United States v. Casperson, 773 F.2d 216, 223 (8th Cir.1985). Moreover, “in responding to a jury’s request for supplemental instruction, it may be proper at times to simply refer the jury back to the original instructions.” United States v. Beckman, 222 F.3d 512, 521 (8th Cir.2000).
Taken as a whole, the jury instructions specified that Engelmann could only be convicted of the fraud and conspiracy offenses if he “voluntarily and intentionally joined in the agreement or understanding” while knowing “the purpose of the agreement or understanding,” R. at 59; that “[a] person who has no knowledge of a conspiracy but who happens to act in a way which advances some purpose of one, does not thereby become a member,” R. at 60; and that Engelmann must have acted “knowingly” and with “intent to defraud,” R. at 66-67. The theory of defense instruction further explained that good faith was “a complete defense ... if it is inconsistent with [Engelmann] acting to conspire with one or more other persons to commit bank and wire fraud ... or the intent to defraud,” and that “good faith may be considered by you, together with all the other evidence, in determining whether or not [Engelmann] acted with intent to defraud.” R. 77. While Engelmann’s requested good-faith instruction was more detailed than the instruction the district court gave, [1012]*1012“[t]he jury need not be instructed on every inference that it might draw bearing on the issue of good faith.” United States v. Ammons, 464 F.2d 414, 417 (8th Cir.1972). Here, the jury instructions “fairly and adequately submitted the issues to the jury.” See Whitehill, 532 F.3d at 751 (internal quotation marks omitted).
B.
Engelmann argues the district court erred in denying his motion for a new trial due to witness sequestration violations. The district court found on remand that (1) the Agents’ out-of-court conversation during trial did not violate the sequestration order or prejudice Engel-mann and (2) although Agent McMillan violated the sequestration order by observing portions of trial testimony, these violations did not prejudice Engelmann. “We review a district court’s rulings regarding sequestration orders for abuse of discretion, granting wide latitude to the court and requiring the moving party to show prejudice.” United States v. Camacho, 555 F.3d 695, 702 (8th Cir.2009). Likewise, “[w]e review the denial of a motion for a new trial for abuse of discretion and give great deference to the district court’s ruling.” Holm v. BNSF Ry. Co., 707 F.3d 995, 1002 (8th Cir.2013) (internal quotation marks omitted).
The district court did not abuse its discretion in finding that the Agents’ out-of-court conversation did not violate the sequestration order. As the district court concluded, the evidence presented at the remand hearing showed that the conversation did not involve disclosure of any details regarding trial testimony.4 Rather, Agent McMillan was told in general terms in preparation for his upcoming rebuttal testimony that “Engelmann denied making certain statements during our interview.” It was within the district court’s discretion to conclude that this conversation did not violate the sequestration order, but rather was permissible contact between a government attorney, the government’s case agent, and a government witness in preparation for the witness’s trial testimony. See United States v. Stewart, 878 F.2d 256, 259 (8th Cir.1989) (“Federal Rule of Evidence 615 ... does not authorize trial courts to prevent executive branch officials from conferring with their witnesses.”).
Furthermore, the district court did not abuse its discretion in finding that the sequestration violations which occurred when Agent McMillan observed trial testimony did not prejudice Engelmann. Agent McMillan only testified as a rebuttal witness, and his testimony was limited to Engelmann’s statement to the Agents during their investigatory interview. The district court found that the testimony Agent McMillan observed “bore no direct relationship to” and was not “in any way pivotal to” Agent McMillan’s own later testimony. Engelmann does not contest these findings on appeal. Consequently, the district court did not abuse its discretion in finding that Agent McMillan’s sequestration violations did not prejudice Engelmann. See United States v. Collins, 340 F.3d 672, 681 (8th Cir.2003) (finding no prejudice when witnesses “offered testimony on two completely different issues that did not overlap and did not involve any of the same facts”).
Absent any prejudicial sequestration violations, the district court did not abuse its discretion in denying Engelmann’s motion for a new trial. We note, however, that the result of the proceedings on remand illustrate the importance of holding an evi-dentiary hearing in this case. The hearing [1013]*1013revealed that the government did not timely advise Agent McMillan that he was subject to a sequestration order since he was a potential witness. Agent McMillan consequently violated the sequestration order on several occasions by entering the courtroom to view portions of trial testimony before Agent Huber finally advised him that he should not be in the courtroom. Moreover, before the evidentiary hearing, a courtroom observer alleged that he overheard the Agents discussing details of trial testimony. At the hearing, however, this observer only testified that he overheard someone in the Agents’ vicinity make a brief reference to Engelmann. The hearing gave Engelmann the opportunity to present evidence concerning the Agents’ conduct, which in turn allowed the district court to meaningfully evaluate Engel-mann’s claims of prejudice. It is important for courts not only to ensure that justice is done, but also to preserve the appearance of justice so that litigants and the public maintain confidence in our legal system. Holding an evidentiary hearing in this case furthered these twin goals.
C.
Engelmann argues the district court erred in increasing his base offense level by 12 due to the amount of loss involved in his convictions. See United States Sentencing Commission, Guidelines Manual, § 2B1.1(b)(1). He asserts that the district court erred in calculating actual loss by comparing the unpaid balances on the mortgages to the prices paid for the properties at sheriffs sales or short sales. According to Engelmann, this method did not take into account the realities of the sub-prime mortgage market, where mortgages were regularly repackaged and sold as securities.5 “The government bears the burden of proving the amount of loss by a preponderance of the evidence.” United States v. Sample, 213 F.3d 1029, 1034 (8th Cir.2000). “We review the district court’s factual findings regarding the amount of loss for clear error.” Id.
The guidelines define actual loss as “the reasonably foreseeable pecuniary harm that resulted from the offense.”6 USSG § 2B1.1 comment. (n.3(A)(i)). Reasonably foreseeable pecuniary harm, in turn, “means pecuniary harm that the defendant knew or, under the circumstances, reasonably should have known, was a potential result of the offense.” USSG § 2B1.1 comment. (n.3(A)(iv)). “Because the damage wrought by fraud is sometimes difficult to calculate^] a district court is charged only with reasonably estimating the loss using a preponderance of evidence standard.” United States v. McKanry, 628 F.3d 1010, 1019 (8th Cir.2011) (internal alterations and quotation marks omitted).
Here, the district court did not clearly err in basing its actual loss calculation on the difference between the unpaid loan balances and the prices obtained for the properties at sheriffs sales or short [1014]*1014sales. This is the method the guidelines recommend, see USSG § 2B1.1 comment. (n.3(E)(ii)), and we have previously upheld this method in cases involving mortgage fraud, see, e.g., McKanry, 628 F.3d at 1019; United States v. Parish, 565 F.3d 528, 535 (8th Cir.2009). Although Engel-mann argues this method does not reflect the realities of a market where mortgages were being securitized, “[t]he appropriate test is not whether market factors impacted the amount of loss, but whether the market factors and the resulting loss were reasonably foreseeable.” Parish, 565 F.3d at 535. It was reasonably foreseeable “that a scheme premised on false loan applications and inflated real estate prices would unravel, and that market conditions could exacerbate the losses.” See United States v. Spencer, 700 F.3d 317, 323 (8th Cir.2012). Although securitization may make it more difficult to allocate losses among individual banks or investors, the district court’s method produced a reasonable estimate of total loss. Cf USSG § 2B1.1 comment. (n.3(F)(iv)) (“In a case involving a fraudulent investment scheme ... loss shall not be reduced by the money or the value of the property transferred to any individual investor in the scheme in excess of that investor’s principal investment (ie., the gain to an individual investor in the scheme shall not be used to offset the loss to another individual investor in the scheme).”). Thus, the district court did not clearly err in increasing En-gelmann’s base offense level by 12 due to the amount of loss.
D.
Engelmann argues the district court erred in ordering him to pay restitution of $57,839.91 to Lehman REO-ALS, $108,560.48 to JP Morgan Chase, and $226,537.34 to New Century Liquidating Trust. He argues that (1) these parties are not “victims” under the Mandatory Victims Restitution Act, 18 U.S.C. §§ 3663A-3664 (“MVRA”), and (2) even if they are “victims,” the government did not present sufficient evidence of the amounts of loss to each victim. “We review de novo the district court’s interpretation of the [MVRA]. We review for clear error the district court’s factual determinations underlying an order for restitution, as well as the district court’s finding as to the proper amount of restitution.” United States v. Statman, 604 F.3d 529, 535 (8th Cir.2010) (internal citations and quotation marks omitted).
1.
The district court properly rejected Engelmann’s argument that Lehman REOALS, JP Morgan Chase, and New Century Liquidating Trust are not “victims” under the MVRA. The MVRA defines “victim” as “a person directly and proximately harmed as a result of the commission of an offense for which restitution may be ordered including ... any person directly harmed by the defendant’s criminal conduct in the course of the scheme, conspiracy, or pattern.” 18 U.S.C. § 3663A(a)(2). Engelmann essentially argues that these three financial institutions are not “victims” because they are bad actors whose conduct caused the sub-prime mortgage crisis. However, New Century was the original lender on seven of the nine mortgages at issue, and Lehman REO-ALS and JP Morgan Chase each directly or indirectly acquired the remaining two loans from the original lender. This puts all three entities squarely within the MVRA’s definition of “victim” as a party “directly and proximately harmed” by the mortgage fraud conspiracy. See 18 U.S.C. § 3663A(a)(2).
2.
The district court did not clearly err in determining the amount of [1015]*1015restitution owed to Lehman REO-ALS, JP Morgan Chase, and New Century Liquidating Trust. “The burden is on the government to prove the amount of restitution based on a preponderance of the evidence.” Statman, 604 F.3d at 535 (internal quotation marks omitted). Restitution awards are “limited to the victim’s provable actual loss.” United States v. Chalupnik, 514 F.3d 748, 754 (8th Cir.2008). The court cannot award restitution to a single victim based on collective losses to a market of which the victim is a member, nor can the court award a victim restitution on behalf of other unidentified victims. Id. at 755. Moreover, “general invoices ostensibly identifying the amount of loss without further explanation are insufficient.” United States v. Haileselassie, 668 F.3d 1033, 1037 (8th Cir.2012) (internal quotation and alteration marks omitted).
Engelmann makes two arguments about why the district court erred in calculating restitution. First, he challenges the district court’s method of subtracting prices paid at sheriffs sales or short sales from unpaid loan balances to calculate the restitution owed on some of the mortgages at issue. He asserts that this method ignores the reality of the subprime mortgage market and the fact that mortgages were being securitized and sold to investors as part of a package. Second, he argues generally that the district court erroneously relied on the victims’ own claims of loss, without requiring any additional corroborating evidence.
With respect to Engelmann’s first argument, the district court’s method of calculating loss was proper. We previously have stated that “for goods held by a merchant for sale, lost profits ... are the proper measure of ‘actual loss.’ ” Chalupnik, 514 F.3d at 755. This reasoning is equally applicable to Engelmann’s situation involving residential real estate. To the extent that the government proved a victim was entitled to collect the payments on a particular mortgage, subtracting the price paid at a sheriffs sale or short sale of the property (ie., the amount the victim actually collected) from the unpaid loan balance (i.e., the amount the victim was entitled to collect) would reflect the amount of profit the victim lost on that particular mortgage. This is true regardless of whether the victim was the original lender or someone who later acquired the securitized mortgage as part of a package. To the extent that Engelmann argues the district court should have credited his expert witness’s testimony that the loss amounts in this case were not provable due to market conditions, the district court’s implicit rejection of that testimony was a credibility determination that is “virtually unreviewable” on appeal. See United States v. Holthaus, 486 F.3d 451, 456 (8th Cir.2007).
Turning to each individual victim, we hold that the district court did not clearly err in determining the government proved each loss amount awarded. Agent Huber testified at sentencing that he reviewed documents and consulted with real estate agents about the sales of the properties at issue. Sent. Tr. 7. Although none of the three victims sent a representative to testify at the hearing, we previously have held that a district court can determine loss amounts by relying on the testimony of a government agent who spoke with victims about their losses. See United States v. Smiley, 553 F.3d 1137, 1146 (8th Cir.2009).
Here, with respect to Lehman REO-ALS, Agent Huber testified that Wells Fargo was the original lender on the mortgage and that Lehman later purchased the mortgage from Wells Fargo. Sent. Tr. 22. Agent Huber testified that he was unable to speak to anyone at Lehman about this property because “Lehman Brothers is no [1016]*1016longer in existence,” and that the claimed loss amount on the property was approximately $89,000 according to a “number that was provided to me by Wells Fargo.” Id. at 22. The district court found that it was “unable to determine the full amount of the claimed restitution for Lehman REO-ALS” based on the evidence offered and instead determined that “the amount of restitution payable to Lehman REO-ALS is the difference between the outstanding loan balance and the sale value, for a restitution amount of $57,839.91.” Appellant’s Addendum 25. The district court did not clearly err in determining the government proved, by a preponderance of the evidence, that Lehman had the right to collect on the mortgage loan. As noted above, the method the district court used to calculate loss was appropriate.
With respect to JP Morgan Chase, Agent Huber testified that Wells Fargo originated the mortgage and then sold it to EMC Mortgage, which in turn was bought out by JP Morgan. Sent. Tr. 24. JP Morgan reported its loss amount as the unpaid principal balance, plus interest, minus “Initial Billing Proceeds.” Appellee’s Addendum A9. Huber explained that this latter number was the price paid at the sheriffs sale. Sent. Tr. 23. Although JP Morgan originally reported the price as $44,697.92, Appellee’s Addendum A9, Agent Huber testified at sentencing that the true price was $50,560, Sent. Tr. 25. In its restitution award, the district court stated it was using the higher sales price to calculate restitution, apparently again basing the award on the difference between the unpaid loan balance and the price obtained at the sheriffs sale, and then adding accrued interest. Cf. Appellant’s Addendum 25.7 The district court did not clearly err in awarding restitution of $108,560.48 to JP Morgan Chase.
With respect to New Century, Huber testified that New Century sold six loans it originated in this case to Morgan Stanley. Sent. Tr. 17. When the buyers defaulted on the first payment of all of these mortgages, New Century had to pay penalties to Morgan Stanley per their contractual agreement. Id. 17-19. New Century submitted documentation showing the amount of these penalties. Appellee’s Addendum A3; Sent. Tr. 20. New Century also submitted documentation showing a loss on a seventh property, apparently calculated by subtracting the price obtained at a short sale from the unpaid loan balance. Cf. Appellee’s Addendum A5. The district court did not clearly err in awarding restitution of $226,537.34 to New Century Liquidating Trust.
III.
Accordingly, we affirm Engelmann’s conviction and sentence.