AMID the legal battles between investors who lost money in mortgage securities and the investment banks that sold the stuff, one thing seems clear: the investment banks appear to be winning a good many of the early skirmishes.

But some cases are faring better for individual plaintiffs, with judges allowing them to proceed even as banks ask that they be dismissed. Still, these matters are hard to litigate because investors must persuade the judges overseeing them that their losses were not simply a result of a market crash. Investors must argue, convincingly, that the banks misrepresented the quality of the loans in the pools and made material misstatements about them in prospectuses provided to buyers.

Recent filings by two Federal Home Loan Banks — in San Francisco and Seattle — offer an intriguing way to clear this high hurdle. Lawyers representing the banks, which bought mortgage securities, combed through the loan pools looking for discrepancies between actual loan characteristics and how they were pitched to investors.

You may not be shocked to learn that the analysis found significant differences between what the Home Loan Banks were told about these securities and what they were sold.

The rate of discrepancies in these pools is surprising. The lawsuits contend that half the loans were inaccurately described in disclosure materials filed with the Securities and Exchange Commission.

These findings are compelling because they involve some 525,000 mortgage loans in 156 pools sold by 10 investment banks from 2005 through 2007. And because the research was conducted using a valuation model devised by CoreLogic, an information analytics company that is a trusted source for mortgage loan data, the conclusions are even more credible.

The analysis used CoreLogic’s valuation model, called VP4, which is used by many in the mortgage industry to verify accuracy of property appraisals. It homed in on loan-to-value ratios, a crucial measure in predicting defaults.

An overwhelming majority of the loan-to-value ratios stated in the securities’ prospectuses used appraisals, court documents say. Investors rely on the ratios because it is well known that the higher the loan relative to an underlying property’s appraised value, the more likely the borrower will walk away when financial troubles arise.

By back-testing the loans using the CoreLogic model from the time the mortgage securities were originated, the analysis compared those values with the loans’ appraised values as stated in prospectuses. Then the analysts reassessed the weighted average loan-to-value ratios of the pools’ mortgages.

The model concluded that roughly one-third of the loans were for amounts that were 105 percent or more of the underlying property’s value. Roughly 5.5 percent of the loans in the pools had appraisals that were lower than they should have been.

That means inflated appraisals were involved in six times as many loans as were understated appraisals.

David J. Grais, a lawyer at Grais & Ellsworth in New York, represents the Home Loan Banks in the lawsuits. “The information in these complaints shows that the disclosure documents for these securities did not describe the collateral accurately,” Mr. Grais said last week. “Courts have shown great interest in loan-by-loan and trust-by-trust information in cases like these. We think these complaints will satisfy that interest.”

The banks are requesting that the firms that sold the securities repurchase them. The San Francisco Home Loan Bank paid $19 billion for the mortgage securities covered by the lawsuit, and the Seattle Home Loan Bank paid $4 billion. It is unclear how much the banks would get if they won their suits.

Among the 10 defendants in the cases are Deutsche Bank, Credit Suisse, Merrill Lynch, Countrywide and UBS. None of these banks would comment.

As outlined in the San Francisco Bank’s amended complaint, it did not receive detailed data about the loans in the securities it purchased. Instead, the complaint says, the banks used the loan data to compile statistics about the loans, which were then presented to potential investors. These disclosures were misleading, the San Francisco Bank contends.

In one pool with 3,543 loans, for example, the CoreLogic model had enough information to evaluate 2,097 loans. Of those, it determined that 1,114 mortgages — or more than half — had loan-to-value ratios of 105 percent or more. The valuations on those properties exceeded their true market value by $65 million, the complaint contends.

The selling document for that pool said that all of the mortgages had loan-to-value ratios of 100 percent or less, the complaint said. But the CoreLogic analysis identified 169 loans with ratios over 100 percent. The pool prospectus also stated that the weighted average loan-to-value ratio of mortgages in the portion of the security purchased by Home Loan Bank was 69.5 percent. But the loans the CoreLogic model valued had an average ratio of almost 77 percent.

IT is unclear, of course, how these court cases will turn out. But it certainly is true that the more investors dig, the more they learn how freewheeling the Wall Street mortgage machine was back in the day. Each bit of evidence clearly points to the same lesson: investors must have access to loan details, and the time to analyze them, before they are likely to want to invest in these kinds of securities again.