Floating Only Wins Now If SORA Falls. Here Is the Breakeven.

The cheapest floating package quoted in Singapore this week is 3-month SORA plus 0.20%, which works out to 1.43% a year. The cheapest two-year fixed rate is 1.45%. Two basis points apart, floating in front, and the comparison tables draw the obvious conclusion.

The conclusion does not survive the arithmetic. The 1.43% is an average of interest rates that have already happened. The 1.45% is a price for the next two years. Setting them side by side compares a photograph of the last quarter against a quote for the next eight — and once you correct for that, floating needs SORA to fall before it wins.

The two rates are not measuring the same thing

Compounded SORA is a backward-looking instrument by construction. MAS publishes the SORA index and the 1-month, 3-month and 6-month compounded averages by 9am on the next business day, each one the realised average of overnight prints across its own window. The methodology is published and none of this is in dispute.

So the 1.4337% on a 3-month-pegged package is the price of money from early July to the end of September. The price of money on 1 October was the overnight print: 1.2965%. That is 6.3 basis points above the figure sitting in the loan document.

Which gives the mechanism. If overnight SORA never moves again, the 3-month average climbs to 1.2965% on its own, as the older and cheaper days roll out of the window. The package becomes 1.4965%. Nothing has to happen in the world for that to occur; it is already in the data.

The breakeven, done properly

Floating beats fixed over a given term if the average 3-month SORA across that term stays below the fixed rate minus the spread. That is the entire test, and it is two lines of arithmetic.

Against the lowest fixed rate quoted this week, floating has one and a half basis points of room. Against a 1.40% two-year fixed, it has already lost: the breakeven sits below where 3-month SORA actually printed on 1 October.

What it costs on $800,000

A $800,000 loan over 25 years, monthly rest, held for 24 months:

The middle row is the one the comparison tables are implicitly quoting, and it is the single state of the world that cannot persist. Running the floating loan properly, with quarterly resets and the instalment recomputed at each one:

Floating wins in one of those three states, and that state requires rates to come down. Doing nothing is enough for fixed to win.

The cheapest-looking peg carries the most locked-in increase

The same mechanism ranks the three tenors — and it ranks them inversely to their headline rates.

The 6-month peg is 7.1 basis points cheaper than the 3-month today, and carries 8.7 basis points more guaranteed increase. The discount is the lag. You are not buying a cheaper rate; you are buying a later one.

Worth noting on its own: the term structure is inverted — 1-month above 3-month above 6-month on every date in our series — which is the signature of a rising overnight path rather than a falling one.

The Fed is not the variable here

The FOMC raised its target range by a quarter point to 3.75–4% on 16 September, on a 12–0 vote, with the statement reading "Inflation remains elevated" and "The Committee will deliver price stability." The next decision is 27–28 October.

But the fed funds midpoint of 3.875% sits 258 basis points above overnight SORA at 1.2965%. Singapore's rates are not a function of the Fed; they are a function of the exchange-rate path MAS runs, and MAS publishes its monetary policy statements in January, April, July and October.

This matters for the direction of the conclusion. An MAS tightening pushes SORA down, because a stronger currency path pulls SGD interest rates below the US dollar's. That is path C — the one state in which floating wins. So nothing here is an argument that the Fed is hiking and you should therefore fix. It is an argument that the breakeven has already been crossed, and that exactly one thing has to happen for floating to come back in front. We have written separately on why fixed rates move before the Fed does.

The HDB loan remains a different conversation

The HDB concessionary rate is 2.6% — the CPF Ordinary Account rate of 2.5% plus a tenth of a point. For the October-to-December 2026 quarter, CPF discloses the formula rate behind that 2.5%: 0.32%, being the three-month average of major local banks' rates for May to July 2026. The 2.5% is a legislated floor, not a market rate, and it cannot fall.

On $500,000 over 25 years, that is $2,268.35 a month against $1,987.96 on a 1.45% bank fixed — $280.39 a month, and $11,246 of interest across two years. Leaving the HDB loan is a one-way door, which is why a gap that size still does not settle the decision on its own. We have taken that question separately.

What to do with this

If you are repricing with your existing bank, the comparison that matters is the fixed rate on offer against the breakeven in the table above, not against the floating rate's current print. At 1.45% fixed you have 1.6 basis points of headroom, which is not a margin anyone should plan around.

If you are choosing a peg rather than a product, do not read the 6-month discount as a saving. It is 13.4 basis points of increase that has already happened and has not reached you yet. Pick the reset frequency you want to live with, then compare like with like.

If you are on floating now with no lock-in, that optionality is worth something real and this arithmetic does not price it. The question is not whether floating is dear today but whether you would rather hold the option or the certainty for the next two years.

If you are weighing an HDB loan against a bank loan, the 116-basis-point gap is genuine and the one-way door is also genuine. Neither fact cancels the other, and the decision turns on how likely you are to want back in.

Method

SORA levels are MAS's published series as read on 1 October 2026 from cashew.sg — overnight 1.2965%, 1-month compounded 1.2492%, 3-month 1.2337%, 6-month 1.1625% — cross-checked for the 1-month and 3-month tenors against nexusmortgage.sg's daily series and against propertynet.sg as at 4 September. The 6-month print is single-sourced. The MAS statistics portal was serving a scheduled-maintenance page throughout our checks, so every figure here is a third-party read of MAS's series rather than a reading taken at MAS; MAS remains the authority and should be checked before acting. Package rates are from Mortgage Master's rate board as updated 3 October 2026 (fixed from 1.45%, floating from 1.43% on completed properties and refinancing) and from cashew.sg's listing of four floating packages at 3-month SORA plus 0.20%, being Maybank on HDB property with a two-year lock-in and HSBC, Standard Chartered and one undisclosed lender on private property without one. The 1.40% two-year fixed attributed to HSBC and Citibank is propertynet.sg's per-bank table dated 4 September 2026 and is five weeks old; no 1-year or 3-year fixed package was quoted on any source we could retrieve. Spread step-up schedules were not published on any page we retrieved, and their absence is material: if the +0.20% is a year-one-and-two rate, floating loses by more than shown, so the dollar figures here are floors. All instalments and interest totals were computed as standard monthly-rest amortisation over a 300-month tenure; the floating paths recompute the instalment at each quarterly reset on the remaining tenure. Path B applies a single 25 basis point policy move at 50% pass-through to SORA. Convergence to spot is an identity about the composition of the current averaging window, not a forecast: if overnight SORA falls, all three tenors fall with it and the tenor ranking holds while the magnitudes shrink. CPF Ordinary Account, Special, MediSave and Retirement Account rates and both formula rates are from CPF's own rate page for 1 October to 31 December 2026; the 4% floor on the Special, MediSave and Retirement Accounts was extended to 31 December 2027 on 22 September 2026. The HDB concessionary rate of 2.6% is stated in that CPF material; HDB's own loan pages returned errors on every path we tried. Fixed-rate and SORA levels change weekly and the figures here are dated throughout. This is general information, not advice on your own loan; confirm rates with MAS, CPF, HDB and the lender before deciding anything.