Full opinion text
OPINION & ORDER DENISE COTE, District Judge: Plaintiff Federal Housing Finance Agency (“FHFA”), as conservator for the Federal National Mortgage Association (“Fannie Mae”) and the Federal Home Loan Mortgage Corporation (“Freddie Mac”) (together, the GovernmenNSponsored Enterprises or “GSEs”), brings this action against financial institutions involved in the packaging, marketing, and sale of residential mortgage-backed securities (“RMBS”) purchased by the GSEs between 2005 and 2007, alleging among other things that defendants made materially false statements in offering documents for the RMBS (the “Offering Documents”). This is one of sixteen related actions brought by FHFA that have been litigated before this Court. All but this action have settled. This remaining action concerns seven RMBS created by Nomura (the “Securitizations”). In each of these Securitizations, one of the GSEs purchased a certificate backed by a pool of loans known as a supporting loan group (“SLG”). Those certificates (the “Certificates”) were sold by underwriters, including — for four of the Securiti-zations — RBS. The GSEs purchased the seven Certificates for more than $2 billion. Nomura and RBS are strictly liable for any material misrepresentations in the Offering Documents for those Certificates, unless they can avail themselves of one of a limited number of statutory defenses. On November 10, 2014, FHFA moved for partial summary judgment on the Defendants’ due diligence and reasonable care defenses under Section 11 and Section 12(a)(2) of the Securities Act of 1933 (the “Securities Act”), 15 U.S.C. §§ 77k(b)(3)(A), 77i(a)(2), and similar provisions of the D.C. and Virginia Blue Sky Acts, D.C.Code § 31-5606.05(a)(l)(B); Va. Code § 13.1-522(A)(ii) (the “Blue Sky Laws”). This motion was fully submitted on December 12. The reasonableness of a defendant’s due diligence investigation will, in most cases, be a question for the jury. It is a mixed question of law and fact that will often hinge on disputed factual issues. Even when it does not, reasonable minds could often disagree about whether a given investigation would have satisfied a prudent man in the management of his own property. In exceptional cases, where no reasonable, properly instructed jury could find for a defendant, summary judgment is appropriate. For the reasons explained at length below, this is such a case. Here, the loans within each of the groups that supported the seven Certificates were loans that Nomura itself had previously purchased. There were over 15,000 such loans within the seven SLGs. Nomura never conducted a due diligence program to confirm the accuracy of the representations in the Offering Documents about the 15,000 loans in the SLGs at or near the time of the securitization. Nor did it undertake such a review of the loans at or about the time Nomura selected the loans for and placed them within the SLGs. Instead, to support its reliance on the affirmative defenses of due diligence and reasonable care, Nomura points to its program for reviewing loans before purchasing them. The 15,000 loans in the SLGs at issue here were drawn from close to 200 pools of loans that Nomura had purchased (the “Trade Pools”) from loan originators. While it is conceivable that a review of loans at purchase — which may occur months before Nomura selects the loans to be placed in a particular SLG — might have been sufficient for a jury to find in Nomu-ra’s favor on these affirmative defenses, the pre-purchase review that Nomura conducted here was not adequate for that purpose as a matter of law. Nomura’s post hoc attempts in its briefing to piece together the fagade of a due diligence program from reviews Nomura undertook before purchasing more than fifty thousand loans in hundreds of trade pools, portions of which would later contribute to the seven relevant SLGs in. the Securitizations, simply underscore the lack of evidence that Nomura undertook any reasonable investigation of the accuracy of its representations about the SLGs before issuing the Certificates. Nomura’s pre-acquisition review was not designed to ensure the accuracy of the descriptions in the Offering Documents of the SLGs that backed the Certificates. Nomura tested samples of loans as it purchased them, but then weeks or months later pulled certain kinds of loans (reviewed and unreviewed) to form the SLGs without taking any care to ensure that the findings from its pre-purchase review program could be reliably applied to the SLGs. This broke the link between the results of Nomura’s pre-acquisition sampling and the characteristics of particular SLGs as they were described in the Offering Documents. Nomura has offered no evidence that it considered how its selection of particular loans for the SLGs impacted its reliance on the sampling of the trade pools. Indeed, there is no evidence Nomura took any care to structure its processes — its pre-acquisition sampling, its construction of SLGs, or its pre-securitization review of the sampling results — to ensure the accuracy of its representations about the SLGs in the Offering Documents. Even putting aside this fundamental error, Nomura’s pre-acquisition review was poorly designed and not implemented in a way that could give reasonable assurance that the kinds of representations that No-mura included in its Offering Documents were accurate. But, even if one assumed that that review were adequate, Nomura’s pre-acquisition review raised red flags No-mura ignored. Despite high “kick-out” rates in the Trade Pools that populated the seven Securitizations, never once did No-mura upsize its sample to test whether it had sufficiently culled loans. Indeed, in at least some cases, Nomura’s bid to buy loans from their originators included a stipulation that Nomura would limit its pre-purchase sampling. Most importantly, there is simply no evidence Nomura ever considered the implications of these kick-out rates for the quality of the loans it would later place in the SLGs and describe in the Offering Documents. No reasonable jury could find that Nomura conducted reasonable investigations or exercised reasonable care with respect to these seven Securitizations. ■ And RBS, although it agreed to act as sole lead underwriter for three of the Sec-uritizations and co-lead for a fourth, made little real effort to test the accuracy of the representations about the SLGs. Unlike Nomura, to the extent RBS conducted a review of loans, it did so after the composition of the SLGs had been determined. But, for two of the Securitizations, RBS undertook no independent review of the loan files. For one of these two, RBS knew nothing of the results of Nomura’s pre-acquisition review of the loan pools from which Nomura selected loans to populate the Securitization beyond a one-page summary that warned the summary might be neither “complete” nor' “accurate.” For the other, RBS was the sole lead underwriter, yet it relied entirely on Nomura’s pre-acquisition reviews, of the Trade Pools. RBS did review loans in the other two Securitizations for which it served as sole lead underwriter, but these reviews were manifestly inadequate as a matter of law. For one of these two, RBS’s Credit Group called the loans “crap” and asked to review one-quarter of them, but was told that RBS did not “own” these loans — RBS ultimately decided to sample less than one-fourth that number. But, even then, when RBS failed to collect all of the files for those small samples, RBS simply proceeded with a review of incomplete sets. To make matters worse, RBS appeared to ignore entirely the results of its valuation reviews in both of these Securitizations, taking no action when a substantial portion of its sampled loans in both groups appeared to have faulty appraisals. Those loans were securitized, and the Offering Documents calculated loan-to-value (“LTV”) ratios based on the potentially faulty appraisals. For these and other reasons, no reasonable jury could find that RBS undertook a reasonable investigation or exercised reasonable care as underwriter for any of these four Securitizations. BACKGROUND The adequacy of any due diligence program is a fact-intensive inquiry, and therefore, ordinarily a matter addressed at trial. But, based on the record here, summary judgment is appropriate. Below are the principal facts cited in Defendants’ major arguments, as well as needed context for those facts. All factual disputes are resolved in Defendants’ favor, and all reasonable inferences are drawn in Defendants’ favor, as non-movants. Where Defendants offered a litany of similar examples, the Court has attempted to select the strongest or most illustrative. I. RMBS, in Brief RMBS are securities entitling the holder to income payments from pools of residential mortgage loans held by a trust; these pools are called Supporting Loan Groups or SLGs. Each of the mortgage loans underlying the Securities at issue (the “Mortgage Loans”) began as a loan application approved by a financial institution, known as the loan’s originator (the “Originator”). Nomura acted as an “aggregator” here, purchasing Alt-A and subprime mortgage loans and then pooling them together, on the basis of credit or other characteristics. The loans selected for a given Securitization were transferred to a trust created specifically for that private-label securiti-zation, or “PLS” (a special purpose vehicle or “SFV”). Within a given Securitization, the loans were placed into one or more Supporting Loan Groups. For example, Nomura’s NHELI 2006-FM1 Securitization, offered through a Prospectus Supplement of October 31, 2006, was composed of fourteen classes of certificates, or “tranches,” and two supporting loan groups with an aggregated stated principal balance of over $1.1 billion. Nomura represented that the original principal balances of the loans in Group I “conform[ed] to Freddie Mac loan limits,” and made no such guarantee about the loans in Group II. The trust then issued certificates, and underwriters sold the certificates to investors like Freddie Mac. These certificates entitled the holder to a stream of income from borrowers’ payments on the loans in a particular SLG. Thus, a certificate’s value largely depended on the ability of mortgagors to repay the loan principal and interest and the adequacy of the collateral in the event of default. The process that generated the Certificates is described in greater detail below. II. Origination: A Loan Is Approved First, a homeowner or prospective homeowner applied for a mortgage loan to a bank or other financial institution (the “originator” of the loan). If the loan was needed to purchase the home, it was called a “Purchase Money Loan.” Alternatively, the applicant might be seeking to refinance an existing loan (a “Refinance Loan”), or to liquidate a portion of the applicant’s equity in an already-mortgaged home (a “Second Mortgage”). The relevant documents for each loan application were collected into “loan files.” These documents would include those submitted by the applicant, as well as certain documents created by the originator in the course of reviewing the loan application. Each loan file was reviewed in a process called “underwriting” at one or more times before the loan was placed in a securitization. Underwriting was done in connection with a set of guidelines limiting the sorts of loans an originator would make, called “underwriting guidelines.” In the first instance, the originator was expected to underwrite each loan it approved, confirming that it met applicable underwriting guidelines, was valued reasonably and accurately, and was not fraudulent. Originators could make case-by-case exceptions to their underwriting guidelines when a loan application failed to meet a certain guideline but appeared to nonetheless qualify for a mortgage program based on compensating factors indicating that the applicant was a sufficiently good credit risk. As will be seen below, originators did not always faithfully underwrite their loans. III. Enter Nomura: Purcháse of the Loans Nomura Credit & Capital Inc. (“NCCI”), acting as an “aggregator” of the mortgage loans, acquired loans from originators in order to pool them into supporting loan groups that would be tied to securities sold to investors. The originator might be selling a group of loans in bulk— a “bulk trade pool” or a “mini-bulk trade pool,” depending on whether the aggregate principal balance of the loans was greater or less than $25 million' — or, if the.loans had been underwritten to Nomura’s “conduit” guidelines, it might be selling them loan-by-loan. The Supporting Loan Groups for the Certificates here were composed of 15,806 loans, drawn from 194 Trade Pools and 122 individual loans purchased through Nomura’s conduit channel. Of those 194 Trade Pools, 140 were mini-bulk pools (the “Mini-Bulk Pools”) — contributing 1,561 loans to the SLGs (approximately 10% of the SLGs)— and the remaining 54 were bulk pools (the “Bulk Pools”), which contributed 14,123 loans to the SLGs (approximately 89%). A trader at Nomura’s Trading Desk, learning of a sale of loans by an originator Nomura had approved following a counter-party review, would bid to purchase them. Before bidding on a pool of loans, a Nomu-ra collateral analyst would receive and review a “loan tape” — a spreadsheet listing selected characteristics of each loan in the pool — and recalculate certain values, like the LTV and debt-to-income (“DTI”) ratios, based on data on the loan tape (in the case of the LTV ratio, the loan amount and the appraisal or sale price of the property) to test the internal consistency of the data. If the Nomura trader won the bid, Nomu-ra was permitted to review some or all of the loan files before final settlement of the trade. The review of loans was undertaken by Nomura’s Diligence Group (also called the “Credit Group” or “Residential Credit Group”) and by vendors chosen by No-mura. During the relevant period, the Diligence Group consisted of between three and five employees, including the head of the group: from 2005 through mid-2006, Joseph Kohout (“Kohout”), and afterward Neil Spagna (“Spagna”). According to a fellow employee, “[o]ne of [Kohout’s] favorite lines was[, ‘W]e are not staffed for this, we are not staffed for this....[’] They could be ordering lunch, I think he was understaffed for that....” Kohout and Spagna have both testified that their staffing was “adequate” to conduct pre-acquisition loan reviews. The Diligence Group was responsible for three different reviews for each trade pool: (1) credit review, in which the loan files for some or all of the loans were examined to determine if the loan was originated in accordance with the originator’s credit guidelines; (2) compliance review, in which the loan files for some or all of the loans were examined to determine if the loan complied with federal, state, and municipal regulations; and (3) valuation review (or “collateral review”), in which the Diligence Group determined if some or all of the appraisals of the loans’ underlying properties were reasonable and accurate. The number of loans reviewed depended upon the deal and the type of review. Valuation review was conducted on all loans. Credit and compliance reviews were usually conducted on all loans in a mini-bulk trade pool, as well as any individual loans submitted through Nomura’s conduit channel, but only for a sample of loans in a bulk pool. Nomura’s sampling methods are described below, following a description of Nomura’s valuation review. A. Nomura’s Valuation Review • Nomura submitted the loan tapes reflecting all loans in a trade pool to a vendor, either Hansen or CoreLogie, for valuation review. Hansen applied its valuation product, “PREVIEW,” which contained the ‘ValueSure” automated valuation model (“AVM”) — a computer program that computed an appraisal value for a property based on a database of real estate transactions, taking into account factors like recent transactions nearby, area history, and regional economic risk. The loan tape information sent to Core-Logic was first reviewed by “HistoryPro,” a “proprietary risk assessment engine” that measured the risk of fraud or default for each loan and assigned it a corresponding “F-Score” between 0 and 25, with higher scores more likely to default. His-toryPro considered a number of factors, including “pricing and appraisal attributes” compared against property characteristics, sales history, comparable sales, and local market data. If a loan received an F-Score of 0, no further valuation testing was done. Approximately 52% of loans in the SLGs received F-Scores of 0. A loan with an F-Score between 1 and 9 was then reviewed by an AVM. If an AVM valuation by Hansen or CoreLogie was within 10% of the originator’s appraisal of a subprime loan, or within 15% of the appraisal of an Alb-A loan (the relevant “variance thresholds”), no further valuation review was conducted. If not, a broker price opinion (“BPO”)— typically offered by a realtor, who visited the property, took photographs, and considered recently sold or listed comparable properties (a “drive-by”) — was usually ordered from a BPO vendor. Among the 46,032 loans in the Trade Pools that underwent Nomura’s valuation review (via His-toryPro or AVM), BPOs were ordered for 8,003 loans (17.4%), including 2,129 loans selected for the SLGs. For loans with an F-Score above 9, the AVM was bypassed and a BPO was ordered directly, or the loan was added to the credit and compliance sample discussed below. Following the broker’s drive-by, the BPO vendor would attempt to reconcile the BPO with the originator’s appraisal. If the reconciled BPO differed from the originator’s appraisal by more than the variance threshold (10% for a subprime loan, 15% for Alt-A), the loan was identified as defective. Nomura would ordinarily refuse to purchase the loan (“kick-out” the loan from the trade pool), unless the originator presented evidence rebutting the BPO, which was “rare.” In such cases, Nomura sent the originator’s rebuttal evidence to the BPO vendor, who would consider it and issue a “final [BPO] value.” 162 loans — approximately 1% of the loans in the SLGs — had final BPOs that differed from the originator’s appraisal by more than the relevant variance threshold, yet were included in the SLGs anyway. Of these 162 loans, more than 75% came from just four Trade Pools: Fremont’s SP02 and SP03 (55 loans), and People’s Choice’s SP01 and SP02 (67 loans). No-mura has provided no documentation that reflects that it made an individualized assessment of the valuation discrepancies for these 162 loans or explains why they were not kicked out despite exceeding the variance threshold. Nomura has speculated that the Originators might have provided “additional information” here, but neither explains why this was not factored into the BPOs’ “final values” nor presents any documentary evidence that this rare event occurred for any of these 162 loans, let alone 55 times for the two Fremont Pools and 67 times for the two People’s Choice Pools. B. Nomura’s Sampling Methods for Credit and Compliance Reviews 1. Sample Size Nomura required a sample size of at least 20% of loans in each bulk trade pool for credit and compliance reviews; John Graham (“Graham”), who headed Nomu-ra’s Contract Finance and Transaction Management Groups, testified that samples were often approximately 30% of the pool. For 24 of the Bulk Pools, Nomura conducted credit and compliance review on all or nearly all of the loans. For the remaining 30 Bulk Pools — constituting approximately 80% of the loans in the 54 Bulk Pools, and 12,971 (82.1%) of the loans in the SLGs — sample sizes ranged from just over 20% to 50% (the “Sampled Bulk Pools”). Kohout explained that sample sizes might be higher in Nomura’s first trades with an originator. The Trading Desk, not the Diligence Group, ultimately determined the appropriate sample size for each pool. In some cases, the Trading Desk entered into agreements with originators that prohibited Nomura from sampling more than a fixed percentage of loans in the pool. For instance, Nomura agreed to review no more than between 24% and 30% of the following six trade pools: Own-It SP02, Gateway 17A, People’s Choice SP01 and SP02, and Silver State 62 and 66. Nomura’s actual sample sizes were within 2% of those caps, with a single exception. In other cases, Nomura agreed to limit its sampling, but reserved the right to request a larger sample (to “upsize” the sample) in certain circumstances. Together, the Trading Desk agreed to limit its sampling for 15 of the 30 Sampled Bulk Pools and one of the Mini-Bulk Pools. As discussed below, there is no evidence Nomura upsized its sample in any of the Sampled Bulk Pools here. 2. Selection of Credit and Compliance Sample Once the Trading Desk determined the sample size for credit and compliance review, it would relay that number to the Diligence Group, which would then select the sample. The Diligence Group selected an “adverse sample,” which was meant to include the “most risky” loans. Kohout has estimated that 90% of Nomura’s adverse sample was selected, by a proprietary computer program created by rating agency Standard & Poor’s (“S & P”) called “LEVELS” that purported to measure the credit risk level of each loan. Kohout objected to the use of LEVELS, stating in an email of April 21, 2005 to the Managing Director of Whole Loan Trading, Steven Katz (“Katz”), that “[tjhis is a non industry standard approach,” that “our process does not conform to what is generally deemed to be effective by industry standards,” and that “when presenting our process to both internal and external parties, it will have to be made clear that Credit’s role in both the sample selection and management of risk on bulk transactions has been diminished to the point of that of a non effective entity pursuant to our limited role in the process.” RBS traders recognized that credit risk did not necessarily correlate perfectly with the risk of fraud. In an email exchange in November 2006 concerning RBS’s adverse sampling, one RBS employee asked another, “Given how fast loans are going bad in deals and how much fraud there appears to be, do you think we need to think about further refining our diligence efforts on the front.” When the second employee replied that RBS “reunderwrite[s] about 25 to 30% of the [trade] pool selected in an adverse sample,” the first responded, “we target lots of low [FICO] type loans but the low [FICO] type loans are not where we find all the fraud.” The 10% of the sample not selected by LEVELS was chosen in an ad hoc fashion by the Diligence Group, considering risk factors including high DTI ratio, high LTV ratio, geographic “soft” markets, high loan amounts, documentation type, and concerns about the accuracy of the property appraisal. The Diligence Group’s selections were relayed to the originator, who sent the loan files for the sample loans. 3. Vendor Review of Credit and Compliance Sample Nomura outsourced all of its credit and compliance review, in the first instance, to third-party vendors. For the Bulk Trade Pools here, Nomura employed the Clayton Group (“Clayton”) and American Mortgage Consultants (“AMC”). Clayton is a leading RMBS review vendor; AMC, too, was used by a number of other RMBS issuers during the relevant period. Clayton and AMC reviewed the loan files for the sample loans against the originator’s underwriter guidelines (“reunder-writing”), in addition to certain “overlays” imposed by Nomura, and gave each loan a grade for both credit and compliance. “Event Level 1” (“EV1”) indicated that the loan met the originator’s guidelines (and Nomura’s overlays); a grade of “EV2” indicated that a loan deviated from the guidelines (or overlays), but the deviation was immaterial or offset by compensating factors; and a grade of “EV3” indicated that the loan did not meet the guidelines (or overlays), or could not be evaluated because of documents missing from the loan file. The vendor could request further information or documentation from the originator. Nomura exercised a great deal of control over the personnel assigned to its reviews: it required Nomura-specific teams composed of employees whose qualifications Nomura had reviewed, and it selected its own project leads. Vendors provided daily reports to the Diligence Group. At least once, in September 2005, an employee of the Diligence Group visited Clayton “to help review [a] .trade.” Employees of Nomura’s vendors testified that Nomura took its reviews “seriously”; one called Nomura’s Diligence Group “knowledgeable” and “professional.” As evidence of Nomura’s “active engagement with its due diligence vendors,” No-mura cites discussions between James Burt (“Burt”), Clayton’s project lead for trade pool Fremont SP 02, and Kohout on October 5 and 6, 2005. Burt informed Kohout of an issue concerning certain forms in the sample; Kohout ordered Clayton to “[p]ull 20 files at random” (within the sample) to investigate the issue. Burt and Kohout also discussed a Massachusetts regulatory issue concerning the borrower’s benefit from a loan. Clayton explained that “[i]t usually is left up to the client [aggregator] to decide if they feel like the [borrower’s] benefit is adequate” to satisfy Massachusetts law; in response, Kohout instructed Clayton to “[c]lear [all of] the MA loans,” instead of requesting individualized inquiry into borrower benefit. 4. Nomura’s Review of Vendor Reports on Credit and Compliance Samples The Diligence Group reviewed the vendors’ exception reports or “Individual Asset Summaries” concerning all sampled loans graded EV2 or EV3, which identified the ways in which a loan deviated from the guidelines or overlays. Some of these reports also identified compensating factors relied upon in assigning a loan a grade of EV2. The Diligence Group did not review the loan files. It also reviewed, at random, vendor reports concerning some portion of loans graded EV1. Kohout estimated this sample could be “anywhere from 25 to 50 percent of the Is”; another Diligence Group employee testified he reviewed EVls “[i]f time allowed.” Following its review of EV3 grades and a vendor’s report, the Diligence Group frequently directed its vendor to regrade a loan as EV2; this was called a client override” or “waiver.” By FHFA’s count, in the 54 Bulk Pools which contributed loans to the SLGs, Nomura’s vendors graded 501 loans as EV3. The Diligence Group issued client overrides for 203 of these loans (approximately 40%), instructing the vendor to regrade them as EV2— i.e., acceptable for purchase and securitization. There is no evidence that the Diligence Group ever directed a vendor to regrade as EV3 a loan a vendor had graded as EV1 or EV2. • Nomura has produced a single post-closing quality control audit of Nomura’s vendors’ pre-acquisition reviews, performed by IngletBlair, LLC (“IngletBlair”). In July and August 2006, IngletBlair reviewed 189 loans securitized by Nomura from the fifteen originators Nomura had purchased the most loans from; 39 of these loans are in the SLGs for the Certificates. Of these 189, IngletBlair reviewed 109 loans that had been previously reviewed by a Nomura vendor, each receiving a final grade of EV1 or EV2. Upon its own review of those loans, IngletBlair graded 7 loans EY3 and another 29 loans EV4, indicating that “[t]he loan is missing critical documentation to determine loan eligibility.” Accordingly, more than 30% of the securitized loans that had been graded EV1 or EV2 were determined either to warrant an EV3 or to lack critical information in the loan file that would permit an EV1 or EV2 grade. IngletBlair delivered these results to Nomura on August 24, 2006. Nomura has identified no evidence that it took any steps in response to this audit, including any change in its use or supervision of its vendors. Sales of four of the Certificates settled after this date. 5. Credit and Compliance Kick-Outs Generally, any loan graded EV3 that was not regraded to EV2 following Nomu-ra’s review of the vendor reports was “kicked-out” of the trade pool: the Diligence Group would inform the originator that Nomura would not purchase those loans, and the originator would remove those loans from the trade pool before the trade settled. As Kohout testified, “there really isn’t a recommended kickout,” since “anything that remains in event level 3 is, in fact, kicked out.” In practice, in the Bulk Pools, Nomura purchased and then included in the SLGs 418 loans (2.6% of the SLGs’ loans) that received a “final grade of [EV]3” — 235 of those received a final EV3 for credit and 197 received a final EV3 for compliance (14 received an EV3 for both) (the “Securi-tized EV3 Loans”). For the NHELI 2007-1 Securitization, 8% of the SLGs’ loans had received a final grade of EV3. Nomura has identified no evidence explaining the Diligence Group’s or Trading Group’s decision to purchase the Securi-tized EV3 Loans, although one of Nomu-ra’s experts has reunderwritten 17 of these loans and offered post hoc justifications that he contends would have supported regrading some of them as EV2s. The typical kick-out rate in Nomura’s subprime or Alt-A trade pools is disputed. In an email of November 20, 2006, Spagna wrote, concerning review of certain Fremont trades, that “our kickout rate on some of these deals are much higher than our typical 7-8% for most subprime deals.” The kick-out rates for those three pools were 6.48%, 11.22%, and 12.12%. The parties’ expert reunderwriting witnesses have testified that the meaning of a kick-out rate depends upon the reason the loans were kicked out of the trade pool. 6. Upsizing a Credit and Compliance Sample When the Diligence Group sent the results of its review to the Trading Desk, it could recommend expanding the sample to include additional loan files. Graham recognized the “industry standard that you could increase the sample size ... if you found a trend that could reveal some particular issue in the origination,” and testified that Nomura would upsize where it saw a negative “trend” in order to “determine if indeed that was something that was systematic and [Nomura would] further increas[e] the ... size of the sample until [it] w[as] satisfied.” When asked if Nomura ever upsized a sample, Kohout “[could] not point to a specific trade,” but confirmed “it did, in fact, happen.” There is no evidence Nomura upsized a sample in any of Sampled Bulk Pools at issue here.- Although the Diligence Group could recommend to the Trading Desk that a sample be upsized, the decision to request an upsize from an originator was ultimately the Trading Desk’s. As Kohout explained, “[w]e would present the results and make recommendations, but whether sample sizes were ultimately increased or not was a function of the relationship between the trading desk and the counterparty.” Up-sizing required “buy in from the seller” because “[t]he seller ha[d] the loan files.” As noted above, in some cases the Trading Desk entered into agreements with originators that prohibited Nomura from sampling more than a fixed percentage of loans in the pool (in the examples cited above, between 25% and 30%); in other eases, Nomura agreed to limit its sampling, but reserved the right to request an upsize in the sample in certain circumstances. In all cases, Nomura could refuse to purchase the trade pool if an originator refused to permit upsizing. Again, there is no evidence Nomura upsized its sample of the Trade Pools at issue here. An email exchange between Kohout and Katz on April 6, 2006 illustrates some of the concerns at play when considering an upsize. Katz emailed Kohout and others in the Diligence Group to “discuss the fallout on [a] trade” with originator People’s Choice. In response, Kohout noted that 90 loans had been kicked out due to faulty appraisals and that 80 of those were accepted by the originator’s in-house appraiser. Kohout stated, “[w]here a seller’s in-house appraiser agrees with ±90% of the loans with value issues pursuant to [our] BPO’s, there is obviously an inherent flaw in their origination process.” Later that day, Kohout wrote that he “took a closer look” and “property valuation declines are off the charts” and reiterated that “the simple fact that only ±10% of the declines in this category were even disputed is further evidence of a systemic issue in this area on the origination side.” Katz asked, “should we test more values? ? even if they passed muster on the initial screen? ?” Kohout replied: Would not be a bad idea. Especially, the higher LTV/CLTV loans. However, playing devils advocate, doing so, would likely place Nomura in a position where we will not be given consideration on future trades. Do we care? Katz responded: “we care.... We can always run them if you think we are at risk ... if there are large differences; we can hold onto them and present [them to the originator for repurchase] when they go down.... or if they go down.” IV. Securitization: Nomura Bundles the Loans to Create Securities 1. Holding the Loans When the Trading Desk’s purchase of a trade pool settled, NCCI took title to the loans and received the loan flies for the loans that had not been part of the pre-acquisition sample. There is no evidence Nomura reviewed any of these flies, or conducted any further review of those loans, prior to the commencement of this litigation. These loans were then held on NCCI’s books until they were securitized. If a loan suffered an early payment default while Nomura was holding the loan, No-mura would not securitize it. More than two-thirds of the loans in the SLGs for the Certificates were held on Nomura’s books for at least two months; approximately 12% were held for five months or longer. 2. Selecting Loans for a Securitization from Trade Pools As noted above, all but 122 of the 15,806 loans that comprise the SLGs were drawn from 194 Trade Pools. The Trading Desk would instruct collateral analysts, who then selected loans from the Trade Pools to populate the SLGs in a given securitization. As one trader explained the “art of selecting the loans,” he would “tell [his] Collateral Analyst what I want, how I want [the securitization pool to] look, what I think will suit the market, what’s in demand.” He considered factors including geographic concentrations, weighted average FICO scores, owner-occupancy status, and weighted average LTV ratios. 3. Representations in the Offering Documents After the securitization was structured, Nomura’s Transaction Management Group, with the assistance of outside accounting firms and outside counsel, would draft the offering documents to be sent to potential investors. In the Offering Documents for each Securitization, Defendants made representations to purchasers, like the GSEs, concerning the Mortgage Loans’ adherence to applicable guidelines and the loans’ characteristics. The Offering Documents included a Shelf Registration Statement filed with the Securities and Exchange Commission (“SEC”), as well as the relevant Prospectus and Prospectus Supplements. For instance, with respect to Supporting Loan Group I in Nomura’s Securitization 2006-FM2 (“2006-FM2”), an SLG that backed a senior Certifícate purchased by Freddie Mac, Nomura represented that: (1) “[a]ll of the mortgage loans were originated or acquired by [originator] Fremont, generally in accordance with the underwriting guidelines described in this section”; (2) 57.5% of the loans (or 68.4% of the pool by principal balance) had an LTV ratio of 80% or lower, and 26.5% of the loans (or 31.1% of the pool by principal balance) had a CLTV ratio of 80% or lower; (3) 93.2% of the underlying properties were owner occupied; and (4) the most senior class, I-A-l would be given the highest credit rating by Standard & Poor’s, Moody’s, Fitch, and DBRS. With respect to the second and third representations, Nomura stated that “[t]he Group I Mortgage Loans are expected to have [those] characteristics as of the Cutoff date,” thirty days before the Securitization’s closing date. Nomura also stated: Prior to the Closing Date, we may remove Mortgage Loans from the mortgage pool and we may substitute other mortgage loans for the mortgage loans we remove. The depositor believes that the information set forth in this prospectus supplement will be representative of the characteristics of the mortgage pool as it will be constituted at the time the certificates are issued, although the range of mortgage rates and maturities and other characteristics of the mortgage loans may vary. The characteristics of the mortgage loans as described in this prospectus supplement may differ from the final pool as of the closing date due, among other things, to the possibility that certain mortgage loans may become delinquent or default or may be removed or substituted and that similar or different mortgage loans may be added to the pool prior to the closing date. The actual mortgage loans included in the trust fund as of the Closing Date may vary from the mortgage loans as described in this prospectus supplement by up to plus or minus 5% as to any of the material characteristics described in this prospectus supplement. The’ Prospectus Supplement for 2006-FM2 also disclosed the following regarding compensating factors: On a case by case basis, Fremont may determine that, based upon compensating factors, a prospective mortgagor not strictly qualifying under the underwriting risk category guidelines described below is nonetheless qualified to receive a loan, i.e., an underwriting exception. Compensating factors may include, but are not limited to, low loan-to-value ratio, low debt to income ratio, substantial liquid assets, good credit history, stable employment and time in residence at the applicant’s current address. It is expected that a substantial portion of the mortgage loans may represent such underwriting exceptions. And the Prospectus Supplements for five of the Securitizations defined the LTV ratio for loans other than Refinance Loans as “generally the lesser of (a) the appraised value determined in an appraisal obtained by the originator at origination of that loan and (b) the sales price for that property.” These numbers were taken from the loan tapes created by the Originator that listed these characteristics, and others, for each loan. Nomura has offered no evidence to suggest that these representations were altered in any way to reflect the results of its pre-acquisition reviews. For example, for the 162 loans in the SLGs with BPOs that differed from the Originator’s appraisal by more than the relevant variance threshold, Nomura does not dispute that its LTV and CLTV representations in the Offering Documents were based on the possibly faulty origination appraisals, not the out-of-threshold BPOs. Nomura has offered the following evidence of steps taken, after the pre-acquisition- review, to confirm the accuracy of these representations. The Transaction Management Group sometimes received results from the Diligence Group’s pre-acquisition review — they received results for 89 of the 194 Trade Pools — and, at times, participated in telephone calls with underwriters of the securities. For four of the seven Securitizations, a single-page chart titled “Due Diligence Summary” was circulated to the Transaction Management Group that listed the percentage of loans to be securitized that had been reviewed and the kick-out rates for credit, compliance, and valuation reasons for the contributing Trade Pools; three of these summaries broke out these rates for the top two Originators. All four include the following disclaimer: “The material contained herein is preliminary and based on sources which we believe to be reliable, but it is not complete, and we do not represent that it is accurate.” Nomura has identified no evidence concerning its use or discussion of these summaries. Graham, who headed Nomura’s Transaction Management Group, has testified that he or someone else “would have at some point vetted th[e] language [concerning compliance with the originator’s underwriting guidelines] with someone in the Due Diligence Group to verify that it generally reflected the underwriting guidelines that were used to originate the loans.” There is no record that the Diligence Group took any steps, after the loans were acquired, to verify the accuracy of this Offering Document representation. Graham explained he relied on “indirect verification,” as he “had confidence in the processes and systems that were involved in the acquisition of mortgage loans,” which “would include due diligence” at that stage. Nomura did hire an outside accountant, Deloitte & Touche LLP. (“Deloitte”), to confirm that the Offering Documents accurately calculated the number of loans with certain characteristics based on the data listed on the loan tape (e.g., an LTV between 75% and 80%), but Deloitte undertook no examination of the accuracy of the data on the loan tapes concerning the loans’ characteristics, and it made “no representations as to ... the accuracy of the information” in the Offering Documents. Nomura also hired Wells Fargo as a “collateral custodian” to ensure that certain required documents concerning the mortgages (e.g., any assignments or title policies) were in its possession. Wells Fargo did nothing to verify the accuracy of the information on the loan tapes concerning LTV ratios or owner-occupancy. 4. Underwriting As noted above, an SPV that held the loans in the supporting loan groups would issue certificates tied to different classes or tranches of the security. Those certificates were sold to underwriters, who in turn sold them to investors, including the GSEs. In each of the Securitizations, Nomura kept some of the most junior certificates (the “Residual Certificates”). These Residual Certificates were the first to take losses should borrowers default. Nomu-ra’s Residual Certificates had recorded market values, at the time of the Securiti-zations’ respective closings, of between approximately $11 million and $40 million; together, they totaled approximately $190 million. Nomura sold some of these interests within one month of the Residual Certificates’ issuance — including nearly two-thirds of its residual interests in NHELI 2006-FM2 and three-quarters of its residual interests in NHELI 2007-2 — and it sold all of its remaining interests within approximately one year of each Securitization. According to a Nomura presentation entitled “RMBS Residual Analysis,” “[ljosses [on residuals] are realized in years 2-4 and much of the cash flow has already been received in year 1.” V. Enter RBS:' Underwriter Nomura Securities acted as sole lead underwriter for two of the Securitizations (NAAC 2005-AR6 and NHELI 2006-FM1); RBS was the sole lead underwriter for three (NHELI 2006-FM2, NHELI 20071, and NHELI 2007-2), and was identified as a co-lead underwriter for a fourth (NHELI 2006-HE3). Non-party Lehman Brothers, Inc. acted as the sole lead underwriter for the final Securitization (NHELI 2007-3). Although RBS’s Underwriting Committee was charged with approving sponsors before RBS could underwrite their RMBS, “approval was mistakenly not obtained from the RBS[ ] Underwriting Committee” until after RBS had underwritten Nomura’s NHELI 2006-HE3 and NHELI 2006-FM2. A. NHELI 2006-HE3 RBS is identified as a co-lead underwriter in the Offering Documents for NHELI 2006-HE3 (“2006-HE3”). RBS’s expert, Charles Grice, has explained that during the relevant period, “typically only one underwriter serves as the true lead underwriter,” and RBS’s “role can be best described as that of a non-lead underwriter” in connection with 2006-HE3. RBS had previously underwritten a Nomura securi-tization that closed on July 28, 2006, but RBS has identified no specific knowledge of Nomura’s processes that RBS gained from that experience. On August 4, 2006, Timothy Crowley, a Vice President at Nomura Securities and member of the Transaction Management Group, emailed a group including RBS employee Adam Smith (“Smith”) to circulate “the initial draft of the term sheet” for 2006-HE3 and request “comments ... by 2:00 Monday [August 7].” In response, Smith emailed Crowley to ask, “Can you send me a summary of the due diligence done on the he3 collateral?” On August 7, another Nomura employee, Michael Orfe (“Orfe”), emailed Smith the single-page “Due Diligence Summary” created for 2006-HE3. This summary included the following disclaimer: “The material contained herein is preliminary and based on sources which we believe to be reliable, but it is not complete, and we do not represent that it is accurate.” Smith responded that the listed balance for the trade, $4 billion, “looks incorrect”; Orfe explained that it represents the total balance of any trade that a loan in this pool was part of. So it is the case that there may be one loan in this pool that came from a trade of $100mm [million], and that $100mm is included in the $4 billion. The idea is to give an overall picture of our DD [due diligence] process. This summary for 2006-HE3 listed the percentage of loans (by unpaid principal balance) that had been kicked out of the contributing Trade Pools for reasons of “Credit,” “Compliance,” and “Property” (together, 7.5%), and the same statistics for the two Originators with loans comprising 10% or more of the Securitization’s loans, People’s Choice and First NLC. The summary did not identify the percentage of loans that had been sampled for credit and compliance review in the underlying Trade Pools, or identify the percentage of loans to be securitized that had been reviewed; it only identified the percentage of People’s Choice’s and First NLC’s loans that had been reviewed (accounting for 60% of the loans, by unpaid principal balance). Smith emailed this summary to Brian Farrell (“Farrell”) and James Whittemore (“Whittemore”) in RBS’s Credit Group, asking them to “review No-mura due diligence on the HE3 transaction that we are a co-manager. Seems to be in-line with subprime loans, please confirm that you are ok with the results.” Farrell asked to see “LTV, FICO, DTI, PPP [prepayment penalty], Property Types” for the collateral; he was sent a summary of the collateral and ten minutes later replied, “Overall snapshot of this looks ok.” Nomura also provided RBS with a list of the six Originators who contributed loans comprising more than 5% of the Securiti-zations’ loans. One RBS employee asked another for the “complete list,” writing, “[n]ot to be a pain in the ass but that still leaves [unaccounted for the originators of] over 20% of the pool.” He was told, “No-mura will only disclose those originators that comprise over 5% of the pool.” On August 10, Smith requested confirmation from Katz of Nomura’s “General Due Diligence Procedures,” outlining his understanding in eleven sentences; Katz provided a few additional details and attached a June 2006 presentation made to S & P entitled “Nomura Securities International Residential Whole Loan Securitization Platform.” The presentation includes three slides under the heading “Due Diligence Process” that represent, among other things, that Nomura’s sampling of bulk trade pools was one-third “[r]andom” and two-thirds “fa]dverse.” RBS also tested the “data integrity” of the loan tape to identify any data input errors; received a “negative assurance letter” from Nomura’s counsel, Thacher Prof-fitt & Wood LLP (“Thacher”), stating that Thacher was not aware of any facts that would render the Offering Documents for 2006-HE3 misleading; and received confirmation from Deloitte that the Offering Documents accurately calculated the number of loans with certain characteristics, based on the loan tapes. RBS participated in a “post-securitization due diligence conference call,” although RBS has identified no details concerning that call. On the basis of this work, RBS underwrote 2006-HE3. At no point did RBS review any of the loan files for the loans underlying the Securitization. B. NHELI 2006-FM2 RBS served as the sole lead underwriter for NHELI 2006-FM2 (“2006-FM2”), which securitized loans from two Bulk Pools, both purchased from Fremont: Fremont SP03 (“SP03”) and Fremont SP04 (“SP04”). In September 2006, RBS received three spreadsheets including information from loan-level reviews conducted for Nomura by the vendor AMC. The first, entitled “Seller Trade Breakout,” listed the number of loans in SP03 and SP04 that underwent credit and compliance review, an AYM review, and a BPO review, as well as the number of loans kicked out of each of these pools “for Credit,” “for Compliance,” “for Valuation,” and “for Collateral.” The Seller Trade Breakout showed that Nomura’s credit and compliance samples were 24.6% and 24.8% of the loans in SP03 and SP04, respectively; that 30.0% and 34.8% of those samples, respectively, had been kicked out for compliance issues; another 5.3% and 8.3%, respectively, of these samples had been kicked out due to credit issues. The other two spreadsheets included the results of AMC’s and Nomu-ra’s credit, compliance, and valuation reviews, for loans reviewed in SP03 and SP04. RBS received all of AMC’s results. Before these three spreadsheets were sent to RBS, Spagna, the head of Nomu-ra’s Diligence Group, instructed AMC to retroactively regrade 19 EV3 loans that had been purchased by Nomura. In an email with the subject line “Huge Favor— Fremont ASAP,” Spagna noted that “the last set of exception reports” from AMC “shows that there are 12 loans in Fremont 3 and 7 loans in Fremont 4 that AMC had marked as [EV]3s but, for what[ ]ever reason, we decided to buy from Fremont.” Spagna then instructed AMC to “[p]lease mark these loans as client overrides Credit Event 2s for all 19 loans in question. Then please forward to me the updated set of reports for these two deals.” Nomura has provided no other evidence to explain the change in the classification of these 19 loans. These revised reports were sent to RBS later that day, and Smith then sent these reports to others at RBS, asking them to “review the results and sampling methods so that we can discuss the extent of our required due diligence as an underwriter.” Farrell emailed Spagna and another member of Nomura’s Diligence Group, Mendy Sabo (“Sabo”), asking them to “elaborate more on high risk characteristics” used to select the adverse sample. Sabo said No-mura would be unable to send a “formal” response in time — Farrell asked for a response the same day — and so instead provided “a quick ad hoc” description. There is no evidence RBS was provided with a fuller description before the Securitization closed. Farrell wrote to Smith that “[t]he only concern is the high number of payment stream exceptions,” and that “[i]f the payment stream issues are isolated, the rest of [the] pool looks good.” Farrell noted RBS had “r[u]n into similar issues in April/May of this year” and that “Fremont stated that they intended to fix their process, which we believe is now true as a result of our [$]1.1 [billion] Fremont review this month.” Earlier in that email chain, a member of Nomura’s Diligence Group wrote that “#’s are skewed (because of the Compliance drops) because we found an issue with the payment stream on [certain] loans ... which we performed 100% [due diligence] on.” Whittemore wrote that “[i]t appears the due diligence sample was sufficient for the size of the pool,” “[t]heir sample methodology and AVM/BPO process appear to be sound,” and “[t]he exception ratios excluding the payment stream issue appears to be what we see when we do our due diligence at Fremont for whole loan trades.” Later, Farrell wrote to Smith that “Credit was ok with results and sampling methodol[og]y.” Months later, in February 2007 — after 2006-FM2 had closed — Farrell was asked by another RBS employee, Grace-Ann Didato (“Didato”), about RBS’s diligence on this deal. Farrell wrote, “We did not perform actual diligence on this. Diligence was performed by another company for Nomura. We signed off on their results.” When Didato asked, “How frequently is this done?,” Farrell replied, “Since being employed, this is the only review type I was involved in w[h]ere [due diligence] results were reviewed and a new diligence was not ordered.” Before 2006-FM2 was issued, RBS participated in a “due diligence teleconference” with RBS’s counsel, Nomura, Nomu-ra’s counsel, and non-lead underwriters. General corporate issues were discussed; according to Spagna, only two questions were addressed to Nomura’s Diligence Group. Spagna wrote to Sabo: ‘We had 2 questions. I took the liberty to bullshit them. I think it worked.” In addition to this call, RBS received a negative assurance letter from Nomura’s counsel, Thacher, and Deloitte verified.the accuracy of the information on the loan tape and the calculations based on that data in the Offering Documents. C. NHELI 2007-1 and NHELI 2007-2 RBS was also the sole lead underwriter for NHELI 2007-1 (“2007-1”) and NHELI 2007-2 (“2007-2”). RBS conducted its own loan reviews, through its vendor Clayton, in connection with these securitiza-tions. While RBS did receive the results of Nomura’s pre-acquisition review for the trade pools that fed into 2007-2, there is no evidence RBS ever received the results of Nomura’s review of the pools that populated the relevant SLG of 2007-1. For both Securitizations, RBS’s sampling was partly “semi-random” and partly adverse. The semi-random sample was created by stratifying the pool to be sampled by unpaid principal balance into bands of $50,000 or $100,000, and then using a random number generator to select loans within each band. The larger the aggregate balance was in a band, the more loans that band would contribute to the sample. The adverse sample was selected according to characteristics like loan balance, FICO score, LTV ratio, and region. Where only some loans with particular adverse characteristics were to be reviewed, a random number generator was used to select them. The adverse sample would be selected before the semi-random sample. In addition, RBS ordered a drive-by BPO for a sample of loans. RBS determined the appropriate sample size according to a number of factors, including the Originator, the type of product, and other risk characteristics. According to the RBS Greenwich Capital Credit Procedures Manual, “[t]he number of files selected for review and the manner of selection may vary due to a number of factors, the most important of which, is [RBS] Greenwich’s exposure to the transaction.” When asked at depositions, RBS employees stated they were not aware that RBS policy varied sample size according to RBS’s “exposure.” 1. Sample Selection a) 2007-1 Sample Selection For 2007-1, RBS selected two samples — each partly semi-randóm and partly adverse — one from a group of fixed-rate loans (“Group I”) and a second from a group of adjustable-rate loans (“Group II”). Group II, composed of 1,751 loans, was later divided into two different supporting loan groups, one of which supported the Certificate purchased by Freddie Mac. Farrell selected samples of 250 loans from Group II; that sample was reduced from 250 loans (14.3% of the group) to 102 loans (5.8%) because Nomu-ra reported it “did not have imaged files for all the loans in [Farrell’s] original samples.” 32 of these loans were selected semi-randomly; 70 were adversely selected. There is no evidence RBS followed up with Nomura, and RBS simply reviewed the requested loan files that Nomura did send. b) 2007-2 Sample Selection RBS did receive the results of Nomura’s review of the loans securitized in 2007-2. Yet, Farrell wrote to Whittemore, “[t]his one is crap. I’m looking for a suggestion.” Farrell warned Smith, “[t]his [sample] will be larger than 250.” When asked why, Farrell explained: “Because it’s crap.” Smith replied, “OK. Do what you feel comfortable with.” Farrell then told Smith, “I would like to review 25% of the total loan population,” or 1,284 of the 5,136 loans in 2007-2. Within one minute, Smith replied: “We don’t own the pool. Call me. [Extension] 2271.” When asked at his deposition “why Mr. Smith would want to take this conversation off line,” Farrell said he did not know. Ultimately, RBS selected a sample of 368 loans for 2007-2 (7.2% of the 5,136 loans). 138 were selected semi-randomly, 168 adversely. As was the case for 2007-1, Nomura reported that it did not “have imaged files” for 60 of those loans, so Farrell’s sample was reduced to the 308 loans (6.0% of 2007-2) for which Nomura transmitted imaged loan files. 2. RBS Loan Reviews There is some additional evidence that RBS used different standards when it reunderwrote loans in what its employees called “securities” — i.e., RMBS to be issued by third-parties — than when it performed pre-acquisition review of “whole loan purchases” RBS might securitize itself. In connection with another third-party securitization of Fremont loans RBS was underwriting, on January 31, 2006, RBS employee Donald Lawson (“Lawson”) gave feedback to another employee, Anne Shera (“Shera”), who had just submitted a draft report of findings concerning those .loans, writing: “As this is a security, we will not be as tough on appraisal and underwriting issues....” A few days before, Shera had asked Lawson for advice, as Clayton flagged a loan as “high cost” but Fremont disputed that. Shera asked Lawson if RBS should “kick” the loan. Lawson replied: “OK for one loan and we’re securitizing off their shelf. We would not buy this loan. Let them know that because we just agreed to buy a $1 Billion pool from them which closes in March.” a) 2007-1 Loan Reviews Of RBS’s sample of 102 loans from Group II, in its initial report to RBS, Clayton graded 28 loans (or 27.5%) “3” due to credit issues; of these 28, 16 loans were graded “3C” — which indicated “only curable material exceptions” — and three were graded “3D,” which meant “missing material documentation.” Nine loans (or 8.8%) received a grade of “3” due to a compliance issue; one of these was graded “3C,” and four “3D.” All told, 33 loans were graded “3” for either credit or compliance reasons (and four for both) according to Clayton’s report prepared January 18, 2007 at 5:35 p.m. Clayton issued a revised report approximately one hour later, at 6:41 p.m., showing that all but three of those 33 loans had been regraded “2W,” indicating a client override, for credit or compliance. RBS’s Whittemore testified that, when “review[ing] a loan file to see if there were compensating factors for exceptions,” he might “flip through the pages and review” in “20 minutes,” or spend as many as “three hours ... [i]f [he] thought it was important.” No documentation of any compensating factors identified by RBS for these 30 loans has been produced. For six of the loans, however, Farrell has recently reviewed them and reports that they appear to have “sufficient compensating factors” that he would have deemed them acceptable at the time. Farrell indicates in conclusory terms that the other loans appear also to be loans he would have “deemed acceptable in 2007 for similar reasons.” Ultimately, RBS overrode all of the “3” grades for the sampled loans. b) 2007-2 Loan Reviews In RBS’s 308-loan sample for 2007-2, Clayton initially graded 50 loans (16.2%) a “3” for credit, including 8 loans graded “3C” and 7 loans graded “3D.” RBS overrode all 50 of these initial grades. Again, there is no documentary evidence of compensating factors identified by RBS for these loans. 3. Valuation Diligence RBS ordered drive-by appraisals for 50 loans within the 2007-1 Group II credit and compliance sample. It is not clear how these 50 loans were selected. Nine of those appraisals were canceled. For 6 of the 41 loans for which a drive-by appraisal was