LZLZL/Prediction markets/Engineering
FREEDO IT G · EngineeringBasis

pp vs %
the error that flips signs

2026-08-21 · one symbol apart, and a loss becomes a profit

"Hit rate from 50% to 52%" is +2pp, not +2% (that would be 51%). It sounds like pedantry. On this venue it has a concrete consequence: if cost and revenue are stated on different bases, subtracting them can produce the wrong sign.

1The definitions

Percentage point (pp)Per cent (%)
Measuresan absolute differencea relative change
50% → 52%+2pp+4%
50% → 51%+1pp+2%
2% → 4%+2pp+100%

The last row makes the point: the same "+2pp" is a 4% relative change or a 100% one, depending on where you started. And "doubled" means very different absolute things at different bases.

2Why this venue is especially prone

Because it has two quantities that both appear with a percent sign and mean different things:

QuantityBasisAt p≈0.5
Taker fee as a share of stakeper cent (%)3.50%
Taker fee per sharepercentage points (pp)1.75pp

The same fee, twice the number. Because the denominators differ:

share of stake = fee ÷ (price × shares)  per share = fee ÷ shares at a price of 0.5, the first denominator is half the second

3★ A concrete way to get it wrong

⚠ "My edge is 2%, the fee is only 1.75%, so I profit"

Probably wrong. That "2%" is most likely percentage points per share (which is how strategy excess is normally computed), while "1.75%" may be a share of stake. They cannot be subtracted.

Convert both to one basis:

edge 2pp (per share) vs fee 1.75pp (per share) → net +0.25pp

Correct this time. But subtract the 3.5% share-of-stake figure instead and you get −1.5 — the opposite sign. Same numbers, wrong basis, reversed conclusion.

4The house rules

RuleWhy
All edges in ppper share, because settlement pays $1 a share so margin is per share
Convert fees to pp before comparing1.75pp at p≈0.5
"Share of stake" answers only "how much was taken"it never participates in an edge calculation
Every number carries its unit"+2" is not a number; "+2pp" or "+2%" is

The last is the cheapest defence: if every figure carries its unit, mixing them is caught immediately. A unitless number passed through three documents loses its basis and nobody can recover it.

5A real case of mine

An internal note recorded a line's expectation as "about −$110/day", with a parenthetical "about −$65/day after reaching a certain rebate tier".

Re-checking later: that conversion was computed as a share of stake. Recomputed on this site's usual per-share pp basis, the same strategy comes out at about −$220/day — a factor of two.

Neither figure is wrong; they answer different questions. The problem was that the basis was not stated, which turned the number into a time bomb the moment it was quoted.

⚠ And another layer: that "−$65/day" never happened

It is a counterfactual — "net after reaching a tier that would return about $44/day". Its purpose was to show that even with the rebate it still loses.

Printed beside a real figure it reads as "it lost $65 a day" — while that line in fact received zero rebates.

Rulea counterfactual must be labelled as one, with its assumption stated. Otherwise it gets passed on as a measurement.

6Four questions for any number

AskBecause
What is its denominator?that is the entire difference between pp and %
Does the thing you are subtracting share it?if not, you cannot subtract
Is it measured or counterfactual?a counterfactual must not travel as a measurement
Absolute difference or relative change?"doubled" means little on a small base

7One line to keep

pp is an absolute difference, % is a relative change. On this venue the same fee differs by a factor of two between the two bases — so before subtracting, confirm both sides share a denominator. Cheapest defence: every number carries its unit.

EvidenceCheck it yourself

Two bases share of stake = 0.07(1−p); per share = 0.07·p(1−p). At p=0.5 those are 3.50% and 1.75pp. Verify by hand.
My own mixing an internal note recorded "−$110/day (−$65/day after a tier)" computed as a share of stake; on a per-share pp basis the same strategy is about −$220/day. Neither is wrong — the fault was not stating the basis.
The counterfactual that parenthetical assumed reaching a tier returning about $44/day; the line actually received zero rebates.
Fee formula see the taker fee.
Checked 2026-08.

NextWhere to go

A · MECHANICS
The taker fee: where both bases come from
B · REBATES
Rebates are a multiplier
G · ENGINEERING
Eight API traps
A · MECHANICS
Glossary
This is an educational and research record. It is not investment advice, promises no returns, and offers no personalised trading recommendations. Rules and API behaviour are per the official documentation; this page states when it was checked and both can change without notice. Prediction markets are restricted or unavailable in some jurisdictions — confirm your own before taking part.