How to read a Treasury auction result

Drafted with AI assistance from official sources; checked against the sources listed below.

There is no official grade. Treasury does not rate its auctions; it publishes the numbers. "Weak" and "strong" are words traders and reporters use when demand looks low or high for that kind of security, against its own past auctions. So the useful question is not "was it weak?" but "how does this compare with earlier auctions of the same term, over a stated window?" Every auction page on 500metrics answers that in its first sentence.

The bid-to-cover ratio

The bid-to-cover ratio is the amount bid for divided by the amount accepted. A ratio of 2.5 means bidders asked for two and a half times what Treasury sold.

• Treasury leaves the Federal Reserve's SOMA add-on out of both sides. In our test on 2,058 auctions, tendered divided by accepted, both without SOMA, matched Treasury's published ratio within 0.01 in 1,827 cases (89%).
• Treasury publishes the ratio from 28 September 1994. For older auctions 500metrics can only show a "tendered to accepted" ratio of its own, labelled as such and never ranked against the official one.
• The ratio is only comparable within a term. Bill auctions and short notes usually draw higher ratios than 20-year and 30-year bonds, so the same number can be high for one and low for another.

The bidder shares

Treasury's results split the competitive awards three ways. Shares are of the competitive amount accepted. The bidder amounts start on 7 April 2008 for bills and 10 April 2008 for notes and bonds.

• Indirect bidders bid through a primary dealer or another direct submitter. They include foreign central banks bidding through the Federal Reserve Bank of New York, and also domestic and foreign investment funds. The data does not say who the indirect bidders were.
• Direct bidders bid for their own account straight to Treasury without being a primary dealer, such as large banks, pension funds and asset managers.
• Primary dealers are the banks and broker-dealers that trade with the New York Fed. They are expected to bid in every auction, so they take whatever the others do not. Their share is sometimes called the dealer takedown.

The tail

The tail is the high yield minus the yield at which the same security was trading in the when-issued market just before bidding closed. A positive tail means Treasury had to accept a higher yield than the market expected; a negative one is said to "stop through". When-issued quotes are licensed market data, not part of Treasury's free results, so 500metrics does not publish tails. Tails published elsewhere come from each publisher's own market feed and can differ.

Three real results, read the 500metrics way

Source: U.S. Treasury auction results via Fiscal Data, checked 2 October 2026; the comparisons are our calculation.

None of these is a verdict. The 5-year result was the lowest ratio in about a year for that term; the 30-year result left dealers with the smallest share in about a year. Each auction page states the windows it uses (the last 12 auctions of the term, and all auctions since April 2008), so the reader can judge.

What the high yield does and does not show

The high yield mostly reflects where the market already was on the day. A 30-year bond sold at 5.3% and a 10-year note sold at 4.8% are different securities, so 500metrics does not compare their yields. The measure that sets the yield against the market just before the deadline is the tail, which we cannot publish. 500metrics therefore compares a yield only with the previous auction of the same term.

What these numbers cannot tell you

• Why demand moved. A low ratio can come after a large offering size, a holiday week or a market move that morning; the results do not say.
• Who the indirect bidders were, or how much of the auction went to foreign investors. Treasury's separate investor class data comes later and in less detail.
• Anything about the future. A comparison with past auctions is a description, not a forecast, and not advice about any security.

How each measure is calculated, the fields used and the method changelog are in the methodology.