Connections

The catalog reads like a list. It is closer to a few facts of convex analysis wearing different institutional clothes. Four of them tie the repository together. The map shows the edges; this page states the theorems.

A word on how strong each claim is, because "connection" covers a wide range and the differences are the interesting part. Five kinds appear below and each is named where it is used. An exact equivalence is an identity between two objects with no side conditions. An equivalence under hypotheses holds once stated assumptions are met and fails without them; the hypotheses are given. A composition identity says an operation on one side corresponds to an operation on the other. An equilibrium result is a statement about what strategic agents do, and carries its solution concept — dominant strategy, Bayes-Nash, or a price-taking approximation — because those are not interchangeable. An analogy is a structural resemblance worth noticing that is not a theorem, and is labelled as such rather than dressed up.

1. Convex duality — exact equivalence, then equivalence under hypotheses

Fix outcomes $\{1,\dots,n\}$ and a convex function $G$ on the probability simplex. Savage's characterisation says a strictly proper scoring rule is exactly

$S(p,i) = G(p) + \langle\, \nabla G(p),\; e_i - p\,\rangle,$

and the edge a truthful forecaster $q$ holds over any lie $p$ is the Bregman divergence $D_G(q,p)\ge 0$ of that same $G$ (Savage 1971; Banerjee et al. 2005; Gneiting & Raftery 2007). Take the convex conjugate: $C = G^\ast$ is a cost-function market maker with prices $\nabla C$, and worst-case loss equal to the range of $G$ over the payoff hull. It is arbitrage-free because that conjugate construction confines prices to the hull, not merely because $C$ is convex: convexity alone permits a sure profit, as the quadratic potential on the maker page shows. Hanson's LMSR is the entropic case $C(q)=b\log\sum_i e^{q_i/b}$ with loss $b\log n$ (Hanson 2007; Abernethy, Chen & Wortman Vaughan 2013).

Convex duality also carries the maker to the CFMM: Uniswap and Balancer are the same potential read from the reserve side rather than a different mechanism (Frongillo, Papireddygari & Waggoner 2024). The map is a level-set or perspective construction under monotonicity and reserve-domain assumptions, though, not the bare involution $C \leftrightarrow C^\ast$. The same duality reaches the order book: a call or batch auction's uniform clearing price is the dual variable of “maximise matched volume,” and a parimutuel is a convex program whose dual yields the implied odds (Agrawal et al. 2011) — that one is for the parimutuel call auction with limit orders over state claims, not for a plain pool, whose odds are a bookkeeping ratio and need no program. So scoring rules, market makers, CFMMs, batch auctions, and parimutuels are conjugate views of one convex potential, with each arrow carrying the hypotheses named above.

2. Infimal convolution is composition — composition identity

The operation that combines two of these objects is the infimal convolution

$(f \mathbin{\square} g)(x) = \inf_{y}\ \{\, f(y) + g(x - y) \,\},$

whose one essential property is that it is dual to addition: $(f \mathbin{\square} g)^\ast = f^\ast + g^\ast$ (Rockafellar 1970; Moreau 1965). Summing in one space is inf-convolving in the other. That single identity shows up as the composition law in two literatures that never cite each other:

Same operation, three readings: pooling liquidity, sharing risk, combining agents. The hybrid CLOB+AMM and the combinatorial market are both instances of merging makers this way. Inf-convolution with a quadratic is the Moreau envelope, whose non-Euclidean form is the mirror map of the next section.

3. Exponential weights is the engine — exact equivalence of updates

Three mechanisms in three different clusters are one update in disguise, mirror descent with the negative-entropy mirror:

The softmax that prices LMSR is $\nabla$ of the log-sum-exp, the gradient of the conjugate of entropy. The market maker, the aggregator, and the sizing rule are reading the same object off different faces.

4. The scoring substrate — exact equivalence, with one equilibrium result

Underneath all of it is the dictionary proper scoring rule ↔ convex generator ↔ Bregman divergence (Savage 1971; Banerjee et al. 2005). The three classics are three choices of $G$:

$\text{log} \to \mathrm{KL},\qquad \text{Brier} \to \lVert q-p\rVert_2^2,\qquad \text{spherical} \to \text{angular separation.}$

Each then has a market-maker twin. The log score is LMSR; the Brier (quadratic) score is a quadratic-potential maker, the Euclidean sibling of LMSR's entropic one (the family cmm.py's quadratic_potential belongs to). Further links:

Companion: the relationship map (the same structure as a graph). Worked pipeline and proofs: An Algebra of Prediction-Rewarding Mechanisms. Sources: bibliography.