Don't read "1,170x oversubscribed" as "a thousand people per token, so I probably got nothing." In a pro-rata sale it means something duller: the money committed during the subscription window added up to roughly 1,170 times what the sale was offering. Nobody is drawn or eliminated. Everyone gets a slice in proportion to what they put in, and the unused part is returned.
Rule of thumb (my own arithmetic, not an official formula): under pure pro-rata, the amount that actually fills ≈ your commitment ÷ the multiple. The rest comes back. Example: commit 1,000 USDT into a 200x sale and roughly 5 USDT buys tokens; the other 995 goes back to your account.
The word comes from stock and bond issuance. In crypto it gets used for token sales run by exchanges and on-chain launch platforms. If Launchpad and Launchpool still blur together, sort that out first; this page is only about subscription-style sales.
01Testing "commitment ÷ multiple" on a real settlement notice
The quickest way to trust a shortcut is to divide some real numbers. Binance's settlement notice for the Space ID Launchpad, published 2023-03-22, is a long-finished sale, so it is useful as a worked check and says nothing about any current sale. It reports that 103,598 participants committed 8,677,923.9471 BNB, "representing an oversubscription of 1,170x," and lists three sample tiers (50, 500 and 5,000 BNB). The table takes the first and last; I added the last column myself:
| Committed | ID received | BNB deducted | Committed ÷ deducted (my calc) |
|---|---|---|---|
| 50 BNB | 578.4411 | 0.04287406 | ≈ 1,166 |
| 5,000 BNB | 57,844.1149 | 4.28740580 | ≈ 1,166 |
1,166 against a headline of 1,170 is a gap of under half a percent. Someone who committed 5,000 BNB spent about 4.29 of it. The notice says that once the deduction is made, the tokens and the remaining BNB both land in the spot wallet.
One more thing in those two rows: the 5,000 row is exactly 100 times the 50 row, in tokens and in BNB spent. Pro-rata is that blunt. Big account, small account, same fill rate.
Which way the missing half percent went, the notice does say: 5 participants hit the 600,000 ID hard cap, “resulting in the participants’ allocation being passed down to the remaining users.” So everyone else filled slightly better than 1/1,170, and a back-calculated 1,166 sitting below 1,170 is exactly that direction. Whether the cap explains the whole gap, the notice doesn't break out; it links a separate guide to the calculation, which I haven't worked through.
021,270%, 12x, 11.7x: three ways to write it
Multiples don't always show up as "x". A Gate blog post on the second Pre-IPOs USDT pool (updated 2026-07-16) carries "1,270% Oversubscription" in its headline. The body says the pool "attracted a cumulative subscription amount of $178 million, reaching a subscription progress of 1,270%," and a few lines on, that "the nominal subscription multiple for the USDT pool exceeds 12 times."
All three describe one fact: demand was about 12.7 times the target. Working back from the two figures in the post, the target was somewhere around $14 million. Strictly, "over" should count only the part beyond the target, which is 11.7x or 1,170%. The headline treats progress as if it were the excess.
| How it's written | Example | As "demand is N times the offer" |
|---|---|---|
| Subscription progress | 1,270% | Divide by 100 → 12.7 |
| Nominal multiple | Exceeds 12 times | Use as is |
| Strict oversubscription | Over by 1,170% | Divide by 100, add 1 → 12.7 |
At 1,000x nobody cares about that extra 1. At small numbers it matters: 200% progress is 2x demand, 200% oversubscribed is 3x. When a sale gives you a percentage, I check which kind it is before dividing anything by it.
03Why the same multiple fills orders differently
The shortcut holds only if allocation follows commitment size from start to finish. Gate's English Launchpad FAQ describes two methods. In one, "tokens are allocated based on the proportion of your total commitment during the subscription period." In the other, "tokens are allocated based on the proportion of your average hourly locked amount." Which one a given sale uses is set in that sale's own notice.
Under the hourly-average method, timing counts as money. The FAQ's worked figure for the project side is a four-hour window with an average total of (100 + 200 + 300 + 400) ÷ 4 = 250 USDT. Now an illustration of my own using the same formula: lock 10 USDT only during the last hour and your hourly average is (0 + 0 + 0 + 10) ÷ 4 = 2.5. Lock the same 10 USDT from the first hour and it averages 10. Same money in the end, a quarter of the weight. The Gate blog post above puts it plainly: "The earlier a user participates and the longer the lock-up period, the higher the allocation weight."
So the multiple and the split may not be measured with the same ruler. That USDT pool's 1,270% came from cumulative subscriptions; its allocation runs on hourly averages. Come in during the final hour, divide your commitment by a multiple like that, and you'll overstate your fill. The idea is close to a whole-period average airdrop snapshot, which the snapshot article walks through.
Caps push their share onto everyone else. Whichever method applies, the English FAQ says "your allocation cannot exceed the specified Max Cap per Subscriber." The English page stops there; the Chinese-language version of the same FAQ adds that the portion above a user's cap is passed to subscribers who haven't reached theirs. So when large accounts get capped, smaller ones fill a little better than the shortcut suggests. That last step is my reading of the rule; the FAQ gives no figure for how much better.
Too small counts as nothing. If your calculated allocation "is less than 0.00000001 (8 decimal places), it will be considered an unsuccessful subscription." That's the zone where a very high multiple meets a very small commitment. ChainGPT Pad's documentation for on-chain public sales is just as direct: "Your allocation may be reduced under the campaign's allocation rules, or you may receive no allocation."
Some sales aren't pro-rata at all. The same ChainGPT Pad page says allocation "is based on your Boosted Commitment (W) relative to all participants." Its example: a No Tier user commits 1,000 USDC, the accepted amount comes to 600 USDC and 400 USDC is returned as an excess refund. With a weighting layer in between, the multiple doesn't translate straight into a fill, and the shortcut is off.
This page explains how a multiple turns into the number of tokens you receive. It doesn't forecast any sale's multiple or say which projects are worth joining. Tokens bought at the subscription price can trade below it after listing, and committed funds are locked for the subscription window. Whether you can live with both is your call.
04What gets deducted and what comes back
Price per token has nothing to do with the multiple. The FAQ gives two formulas:
- Token Price = Fundraising Target / Total Number of Tokens Issued
- Deducted Amount = Number of Tokens Received × Token Price
Both inputs are set before the window opens, so the price is too. Crowd the sale all you like; what shrinks is your token count, and the deduction shrinks with it. The settlement notice from section 01 agrees: the 50 BNB and 5,000 BNB tiers paid the same BNB per ID.
On timing, the FAQ says the amount for your allocation "will be deducted from your account within 4 hours after the subscription period ends" (page dated 2025-04-02, checked November 2026; the live page wins if it changes). For how the unfilled part comes back and where to look for it, see refunds when a subscription doesn't fill. The older allocation-by-VIP-level rule is a separate formula; it lives in how VIP allocation worked.
These days when I read a settlement notice I skip past the multiple and go straight to the deduction line. It tells you what pro-rata actually left you with, which the multiple only hints at. Next time a sale posts a four-digit number, four questions are enough:
- Is that figure progress, a nominal multiple, or the excess?
- Does this sale split by total commitment, by hourly average, or with a weighting?
- What's the per-user cap, and could I hit it?
- Does the shortcut leave me below eight decimal places?
Compiled from official material: Gate's English Launchpad FAQ (page dated 2025-04-02) and, for the cap-overflow sentence, the Chinese version of the same FAQ; a Gate blog post on the second Pre-IPOs USDT pool (updated 2026-07-16); Binance's Space ID Launchpad settlement notice (2023-03-22); and ChainGPT Pad's public sale documentation. All checked 2026-11-05. We have no record or screenshots of taking part in these sales. The "committed ÷ deducted" column, the implied target and the 2.5 vs 10 illustration are our own calculations, not platform figures.
Allocation methods, the per-user cap, price and deduction formulas, the eight-decimal threshold and deduction timing: Gate Launchpad FAQ. The three ways of writing a multiple: Gate blog, second Pre-IPOs USDT pool. Settlement figures used for the check: Binance Space ID Launchpad settlement notice. Weighted allocation and the "no allocation" wording come from ChainGPT Pad's Public Sale Subscription documentation. All are plain reference links with no invite code attached. Checked 2026-11-05.