Path Integral

Feynman’s answer: count every path.

In 1948 Richard Feynman published a new way to describe how a particle travels from one point to another. Classical physics assumed it takes one path. Feynman’s answer was that it takes all of them. To know where the particle will arrive, you add up every possible route, each with its own weight. Most of the routes cancel one another out. The path we actually observe is the one that survives the sum.

Mike McKenzie met that idea as an undergraduate studying physics, and it never left him. A forecast is not a path. It is what remains after every path has been counted.

A loan works the same way. Between today and resolution it has many possible futures: a refinancing, an amendment, a sponsor’s equity cure, a turn to PIK, a debt-for-equity exchange, a credit bid, a liquidation. Most lenders price the one path in the base case. We price the sum.

A grade is not a probability.

In 1958 the National Bureau of Economic Research published W. Braddock Hickman’s Corporate Bond Quality and Investor Experience, a study of American corporate bonds from 1900 to 1943. Hickman found that lower-grade bonds, taken as a group, rewarded their holders more than their reputation suggested. Investors had consistently overpriced the risk. They had read the rating as the outcome.

A Berkeley undergraduate named Michael Milken found the study in the 1960s. He later said he was “struck by the disparity between theory and reality.” He studied capital structure at Wharton, joined the firm that became Drexel Burnham Lambert in 1969, and in 1978 moved its high-yield desk to Beverly Hills. A single observation about mispriced fear became a market.

In private credit the letter grade has been replaced by the model: the mark, the nonaccrual flag, the mandate of the vehicle that holds the loan. When a credit falls outside that model, it is priced as if its worst path were its only path. Path Integral Funds is built on that disparity.

When the board said fast.

Before electronic trading, Mike McKenzie worked as a runner on the floor of the Chicago Mercantile Exchange, carrying orders by hand into the pit. When the market moved hard, the mechanical price board above the floor could not keep pace. Instead of prices, it posted a single word against the last trades:

The traders knew what that meant. The market was moving faster than the machine could record it.

Mike has spent his career on the moment when the machine catches up. At The Receivables Exchange, which was backed by NYSE Euronext and Bain Capital Ventures, he helped build an electronic marketplace where companies, including NYSE-listed issuers, sold corporate receivables to institutional buyers. At Earn.re, he worked on a blockchain platform, co-developed with Walker & Dunlop, for trading tokenized commercial real estate interests.

The board no longer says fast. It keeps up now, and so can we.

AI makes it possible. Markets make it essential.

Summing every path for one loan used to be a thought experiment. Doing it for a portfolio was out of reach. Each credit has its own agreement, intercreditor, amendment history, monthly reporting, and cash forecast. No team could read them all, weigh every path, and update the weights as conditions changed.

Frontier AI models now can. They read the documents, build the family of possible paths for each credit, and re-weigh those paths as new information arrives. That is the path integral made practical: every path explored at once, across a whole book of loans.

The market has made this a necessity. A single base case fails fastest exactly when markets move fastest, and private credit is moving fast.

These are Level 3 assets with no observable market. The same loan can be marked at sixty in one vehicle, reserved against at a bank, and still accruing in an evergreen fund. Each holder has priced one path. None has priced them all.

How Path Integral Funds works.

Every credit we consider goes through the same discipline. We map its paths, weight them, and price the sum rather than the base case. When a loan’s full set of paths is worth more than its holder’s mark, we have a reason to act, through one of three channels.

The machine is built to be inexpensive. John Bogle built Vanguard on a simple argument: investors as a group earn the market’s return minus what they pay to get it, so cost is the variable a manager can actually control. Distressed credit has always been costly to run, because the reading and analysis were done by hand. AI turns that work into a small expense, which leaves experienced judgment to concentrate on the price, the plan, and the negotiation.

The design also draws on the Federal Reserve. Mike McKenzie interviewed with the Federal Reserve Bank of New York for Managing Director of Markets in Chicago, leading the fifty-person shadow operation for the New York Fed’s portfolio, then $8.3 trillion. A shadow desk exists for the moment the primary operation cannot keep going. It runs on rules, models, and redundancy, with a small team carrying very large scale. Path Integral Funds brings that structure to the loans the primary model has set aside.

Less loss, for the whole sector.

Most losses in a credit downturn come less from the credits themselves than from the way they are sold. Forced sales at the bid, extensions that only defer the problem, and second restructurings of the same names all turn a recoverable loan into a realized loss.

Path Integral Funds is built to give lenders a third choice. A bid that reflects every path narrows the gap between the fire-sale price and the recovery value. Fewer fire sales keep the marks across the sector closer to what loans are actually worth. A workout bench raises recoveries on the credits that need one most.