The number that moves a float is the one nobody has to write down

A sentence that moves a billion dollars

The most powerful statement in an Australian float is four words long and is almost never written down: the book is covered. It tells an institution that scarcity is real, that scaling is coming, and that hesitating means missing out. It is the reason bids get placed at the top of a range rather than the bottom. And there is no standard definition of it, no requirement to attach a price to it, and nothing that obliges anyone to keep a record of having said it.

That gap has just been put on display. Firmus, a data centre operator with a site at St Leonards in Launceston, was marketing a float billed as the biggest since Telstra in the 1990s, with a listing scheduled for 23 October. Investment banks stopped taking bids on 8 October. The ABC reported that Goldman Sachs was tapped at midday that day to rescue the raising, that analysts had raised concerns about the company’s debts and the speed at which its valuation had risen, and that Firmus did not appear on the exchange’s own “upcoming floats and listings” page. The fund manager Roger Montgomery said there had been inconsistencies in the messages given to market participants, and that investors had previously been told indicative orders were already above the size of the offer at $11 a share.

I am not asserting that anyone misled anybody. That is precisely the problem. Neither the critics nor the banks — who presumably would like to clear this up — can settle it, because the thing in dispute left no trace.

What “covered” can mean and still be true

Start with the word. An indication of interest is not an order. It can be withdrawn at any point before the book closes, for any reason or none. A book that is covered on Tuesday can be half empty on Thursday without a single person having lied.

Coverage is also a function of price, and the price is usually left out. A book can be comfortably covered at the bottom of a range and thin at the top; “orders above the size of the offer” says nothing about where those orders sit unless someone attaches a number to it. Nor does it say who is in the book. Cornerstone investors who negotiated their allocation months earlier, existing holders rolling over, and funds bidding deliberately large in anticipation of being scaled back all count the same in a raw total. Strip those out and the same book can be covered twice over and genuinely weak.

So the headline claim can be literally accurate and close to uninformative. The investor on the other end of the phone cannot tell which, because only the syndicate can see the book.

The law is watching the wrong document

Australian practice concentrates liability on the prospectus — the document, its forecasts, its risk section. The bookbuild runs beside it, aimed at professional investors on the assumption that they can look after themselves. For most things, they can. They can read accounts, model debt, and discount a management forecast. What they cannot do is verify demand. The only party with that information is the party selling.

The retail defence is thinner than it looks, too. Australians who buy on day one inherit a price struck in a room they were never in. And the funds in that room are mostly investing compulsory retirement savings, so the supposedly sophisticated end of the market and the household end are, here, the same money.

The case for leaving it alone

The serious objection is that informality is the product. A bookbuild works because indications are soft. Investors reveal price sensitivity early precisely because nothing binds them. Make every utterance a record and institutions will say less, bid later and bid lower, which hurts issuers — and ultimately the pension money that owns them.

There is a second objection: demand genuinely moves. Between a soft sounding and a book close, a closer read of a balance sheet or a global repricing of data centre infrastructure can empty a book honestly. A withdrawn indication is not a false statement, and a rescue call to another bank is evidence of a deal in trouble, not of misconduct.

Both are right, and neither argues against a record. I am not proposing to make indications binding, or to publish the book live, or to second-guess pricing. Keeping a timestamped log changes nothing about how the negotiation is conducted. It changes only what can be established afterwards. If demand was real at $11 and then left, the log exonerates the people who said so. The only party worse off is one whose claims do not match it.

The quiet tell

One detail deserves more attention than it has had: a float of this scale was not on the exchange’s public list of upcoming listings. Whatever the administrative reason, a calendar that shows deals only once they are safe is not a calendar. It is a marketing page. Market infrastructure should record attempts, not just outcomes, because the attempts are where the information is.