Next auction: estimated cut-off range

How we compute this

MAS publishes a daily benchmark yield for the 6-month T-bill — it is the market yield of the most recently issued bill, so it largely reflects where the last auction cleared, adjusted by secondary-market moves since. This page tracks several simple estimation methods built on that series, scores each one against every past auction using only data available before that auction, and shows the method with the lowest recent error.

Currently active method: latest 6-month benchmark yield (lowest (or within 0.5bp of lowest) average error over the last 24 auctions). The displayed range spans the 10th–90th percentile of this page's own estimate errors over the last 50 auctions — by construction, roughly 1 in 5 past auctions closed outside it.

Data lands on a delay: results publish early afternoon on auction day and flow in at this site's next 4-hourly refresh. Yields shown "as of" their MAS publication date.

How accurate is this?

Backtested since 11 Jul 2019 across 185 auctions, walk-forward (no hindsight). Mean absolute error: 7.4bp over the full period and 5.9bp over the last 24 auctions. Within 10bp: 77% full period, 83% recently.

Two caveats: the worst miss was +58bp (08 Dec 2022, during the 2022 rate-hike cycle) — in fast-moving rate regimes errors run far above the averages shown here. And because the benchmark partly reflects the previous auction's result, short-horizon accuracy is structurally flattered.

Last 8 auctions — estimate vs actual

Method leaderboard (trailing 24 auctions)

The active method is re-elected automatically as each new auction result arrives (7 changes since 11 Jul 2019). With ~26 auctions a year, the selection adapts over months, not days.

How this differs from our SSB projection

The SSB projection reproduces MAS's own published formula — a calculation, with a single near-certain answer. T-bill cut-offs have no formula: they are set by auction demand. This page therefore shows a market-implied level and the historical spread around it, not a forecast model. Our methodology page explains why we don't attempt yield forecasting — that position is unchanged.

Looking for a T-bill prediction?

There isn't one — not an honest one. Auction demand (and therefore the cut-off) is not knowable in advance, and anyone offering an exact "prediction" or "forecast" of the next cut-off is presenting a guess with the uncertainty removed. What can be measured is where the market currently prices 6-month bills and how far auctions have historically strayed from that level — which is exactly what this page shows, with its full error record published above.

FAQ

How is this range calculated?

Several simple estimation methods are scored against every past auction using only data available before each auction. The method with the lowest recent error supplies the central value; the range around it spans the 10th to 90th percentile of this page's estimate errors over the last 50 auctions. All inputs are MAS-published data.

Why is there no exact prediction?

Cut-off yields are set by auction demand, which is not knowable in advance. Historically, about 1 in 5 auctions closed outside the displayed range. The estimate describes where the market currently prices 6-month bills, not what the auction will do.

Where does the data come from?

Daily benchmark yields and auction results are published by the Monetary Authority of Singapore. This page recomputes whenever new MAS data lands.