feat(orderscaler): introduce order scaling with config, API & audits; integrate engine/emitter #42

terwey · 2025-10-31T10:57:34Z

feat(orderscaler): introduce order scaling with config, API & audits; integrate engine/emitter

Add an order-scaler subsystem that applies configurable multipliers to mirrored orders, preserves take-profit stack ratios under venue constraints, and records auditable changes. Integrates across engine, storage, API, docs, and tests.

Area	Key changes
Config	Add `OrderScalerMaxMultiplier` (flag `--order-scaler-max-multiplier`, env `RECOMMA_ORDER_SCALER_MAX_MULTIPLIER`). Validate >0. Wire into API handler and engine.
Engine	Insert scaling step before submission; skip orders that fall below venue minimums (with audit); preserve TP stack ratios; feed scaled size/price into fill tracker.
Emitter	Return/propagate Hyperliquid order status; log `requested_size`, `executed_size`, and IOC retry metadata.
HL constraints	Add metadata cache for coin rounding (`SizeStep`, significant figs, min notional helper).
Storage	Persist default + per-bot scaler configs; compute effective scaler; store `scaled_orders` audit with rounding delta, actor, skip reason; expose list/query helpers.
API (OpenAPI)	New endpoints: `GET/PUT /api/v1/order-scaler`, `GET/PUT/DELETE /api/v1/bots/{botId}/order-scaler`, `GET /api/orders/scalers`. Enforce multiplier bounds and vault session.
Streaming	Emit `order_scaler_config` and `scaled_order_audit` entries on `/sse/orders` with actor and effective config.
Docs	Architecture + operations runbook for the scaler; navigation updates; README pointer.
Tests	Unit tests for scaler service, engine integration, emitter logging, fill tracker updates, and API handler.

Notes:

Multiplier clamped by configured max; below-minimum-notional legs are skipped and audited with reason.
Effective config resolves per-bot overrides against the global default and is included in stream/audit payloads.

Refs: #43, #44, #45, #46, #47

https://chatgpt.com/codex/tasks/task_e_690493b50a808326a027433a459e035a

terwey · 2025-10-31T11:22:44Z

@codex review and check if it matches specs/order_scaler.adoc

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2025-10-31T11:24:42Z

storage/order_scalers.go

+func (s *Storage) UpsertOrderScaler(ctx context.Context, multiplier float64, updatedBy string, notes *string) (OrderScalerState, error) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+
+	params := sqlcgen.UpsertOrderScalerParams{


Publish SSE events when order scaler settings change

The spec requires multiplier configuration to be “hot‑reloadable” and broadcast via the storage stream, but UpsertOrderScaler (and similarly UpsertBotOrderScaler/DeleteBotOrderScaler) only persist to SQLite and immediately return without calling publishStreamEventLocked. Components that cache the effective multiplier will never see updates unless they poll or restart, so UI/API consumers cannot react to configuration changes as described in specs/order_scaler.adoc.

Useful? React with 👍 / 👎.

…ion-in-operations docs: document order scaler operations and architecture

…port feat: add order scaler configuration endpoints

…vice feat: add Hyperliquid order scaling service

…d-recordscaledorder feat(storage): add order scaler read models and events

terwey · 2025-10-31T14:06:14Z

@codex review and confirm if specs/order_scaler.adoc is fully implemented

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2025-10-31T14:10:02Z

engine/engine.go

+func (e *Engine) applyScaling(
+	ctx context.Context,
+	md metadata.Metadata,
+	latest *recomma.BotEvent,
+	action recomma.Action,
+	logger *slog.Logger,
+) (recomma.Action, *orderscaler.Result, bool, error) {
+	if e.scaler == nil || latest == nil {
+		return action, nil, true, nil
+	}
+
+	switch action.Type {
+	case recomma.ActionCreate:
+		if action.Create == nil {
+			return action, nil, true, nil
+		}
+		req := orderscaler.BuildRequest(md, latest.BotEvent, *action.Create)
+		result, err := e.scaler.Scale(ctx, req, action.Create)
+		if err != nil {
+			if errors.Is(err, orderscaler.ErrBelowMinimum) {
+				logger.Warn("skipping scaled order below venue minimum", slog.Float64("price", req.Price), slog.Float64("size", req.OriginalSize))
+				return recomma.Action{Type: recomma.ActionNone, Reason: "scaled order below minimum"}, nil, false, nil
+			}


Emit an audit when a scaled order is skipped

The spec requires deals that fall below Hyperliquid minimums to be “flag[ged] for operator review and skip submission rather than silently discarding legs”. In applyScaling a below‑minimum leg is handled by logging and returning ActionNone, but no storage audit or SSE event is emitted. As a result operators who rely on the order scaler records or the streamed logs have no visibility that an order was skipped, so the requirement to flag the deal is not met.

Useful? React with 👍 / 👎.

@codex plan a task how we resolve this properly

The current implementation in Engine.applyScaling turns a below-minimum leg into ActionNone without recording anything in storage or the SSE stream, so operators never see that the scaled order was skipped. This violates the spec’s requirement to flag sub-minimum deals for review instead of silently discarding them. engine/engine.goL500-L536 specs/order_scaler.adocL14-L18

:::task-stub{title="Record and broadcast skipped scaled orders"}
Step 1. Extend the scaler audit model to represent skipped submissions (e.g., add a skipped flag or reason) by updating storage/sqlc/schema.sql, corresponding queries, and regenerating storage/sqlcgen. Update storage/order_scalers.go conversions and publishStreamEventLocked to include the new metadata, and reflect the shape change in openapi.yaml/generated API types so SSE consumers can detect skipped legs.

Step 2. Teach engine/orderscaler.Service to invoke the new storage hook when constraints.NotionalBelowMinimum triggers, ensuring the audit row and SSE event are created before returning ErrBelowMinimum. Add focused coverage in engine/orderscaler/service_test.go for the skip path.

Step 3. Update Engine.applyScaling so that when Scale reports a skipped order it forwards the scaler’s audit metadata (if any) and logs, ensuring the engine records the skip without re-entering storage. Extend existing engine tests (or add new ones) to verify the audit emission when an order is skipped.
:::

View task →

…or-skips feat: audit skipped scaled orders

terwey · 2025-10-31T14:52:41Z

@codex review and confirm if specs/order_scaler.adoc is fully implemented. I do not care about database migrations, nobody is running this software yet in prod.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2025-10-31T14:56:37Z

engine/orderscaler/service.go

+	scaledSize := req.OriginalSize * multiplier
+
+	constraints, err := s.constraints.Resolve(ctx, req.Coin)
+	if err != nil {
+		return Result{}, fmt.Errorf("resolve constraints: %w", err)
+	}
+
+	roundedPrice := order.Price
+	if req.Price > 0 {
+		roundedPrice = req.Price
+	}
+	roundedPrice = constraints.RoundPrice(roundedPrice)
+
+	roundedSize := constraints.RoundSize(scaledSize)
+	roundedSize = clampToNonNegative(roundedSize)


Keep take‑profit ratios intact when rounding scaled legs

The order scaler currently multiplies and rounds each order independently (scaledSize := req.OriginalSize * multiplier followed by constraints.RoundSize), which can change the proportional sizes of stacked take‑profit orders when lot rounding kicks in. The spec in specs/order_scaler.adoc requires that “scaled legs preserve ratio orderings” after applying venue tick/lot rules. With the existing implementation two legs sized 1.0 and 1.05 with a 0.5× multiplier and 0.1 size step become 0.5 and 0.5 after rounding, altering the leg ratios and changing exposure. Consider adjusting the algorithm to scale the entire stack and redistribute rounding deltas so the relative sizes remain consistent while still respecting the venue’s increments.

Useful? React with 👍 / 👎.

@codex propose a better implementation

Summary

Extended the order scaler request metadata and scaling routine to resolve take-profit stack sizes from storage and apply ratio-aware rounding across each stack while preserving lot constraints. engine/orderscaler/service.goL66-L190 engine/orderscaler/service.goL204-L320

Added a storage helper that queries the latest threecommas take-profit legs for a deal to supply stack sizing data, alongside the supporting SQL definition. storage/order_scalers.goL158-L210 storage/sqlc/queries.sqlL402-L424

Introduced a unit test covering stacked take-profit legs to ensure scaled outputs maintain the expected ordering after rounding. engine/orderscaler/service_test.goL107-L166

Testing

⚠️ go test ./... (hung waiting on external module downloads; aborted)

View task →

terwey · 2025-10-31T15:35:28Z

@codex review

chatgpt-codex-connector · 2025-10-31T15:40:09Z

Codex Review: Didn't find any major issues. Breezy!

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

…or-orders feat(orderscaler): introduce order scaling with config, API & audits; integrate engine/emitter Add an order-scaler subsystem that applies configurable multipliers to mirrored orders, preserves take-profit stack ratios under venue constraints, and records auditable changes. Integrates across engine, storage, API, docs, and tests. | Area | Key changes | | -------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Config | Add `OrderScalerMaxMultiplier` (flag `--order-scaler-max-multiplier`, env `RECOMMA_ORDER_SCALER_MAX_MULTIPLIER`). Validate >0. Wire into API handler and engine. | | Engine | Insert scaling step before submission; skip orders that fall below venue minimums (with audit); preserve TP stack ratios; feed scaled size/price into fill tracker. | | Emitter | Return/propagate Hyperliquid order status; log `requested_size`, `executed_size`, and IOC retry metadata. | | HL constraints | Add metadata cache for coin rounding (`SizeStep`, significant figs, min notional helper). | | Storage | Persist default + per-bot scaler configs; compute effective scaler; store `scaled_orders` audit with rounding delta, actor, skip reason; expose list/query helpers. | | API (OpenAPI) | New endpoints: `GET/PUT /api/v1/order-scaler`, `GET/PUT/DELETE /api/v1/bots/{botId}/order-scaler`, `GET /api/orders/scalers`. Enforce multiplier bounds and vault session. | | Streaming | Emit `order_scaler_config` and `scaled_order_audit` entries on `/sse/orders` with actor and effective config. | | Docs | Architecture + operations runbook for the scaler; navigation updates; README pointer. | | Tests | Unit tests for scaler service, engine integration, emitter logging, fill tracker updates, and API handler. | Notes: * Multiplier clamped by configured max; below-minimum-notional legs are **skipped** and audited with reason. * Effective config resolves per-bot overrides against the global default and is included in stream/audit payloads.

feat(storage): add order scaler persistence

4c3c141

terwey added the codex label Oct 31, 2025 — with ChatGPT Codex Connector

feat(storage): add order scaler wrappers

7796496

chatgpt-codex-connector bot reviewed Oct 31, 2025

View reviewed changes

terwey added 10 commits October 31, 2025 12:57

feat(storage): add order scaler read models and events

16222d6

feat: scale orders before emitting

58887b0

feat(api): add order scaler configuration endpoints

58bd836

fix(orderscaler): record applied multiplier in audits

a8133c9

fix(api): clamp order scaler effective multipliers

d8f319f

docs: add order scaler runbook and architecture

aa3727a

Merge pull request #46 from recomma/codex/add-order-scaler-documentat…

f412fb1

…ion-in-operations docs: document order scaler operations and architecture

Merge pull request #45 from recomma/codex/extend-order-scaler-api-sup…

ceba566

…port feat: add order scaler configuration endpoints

Merge pull request #44 from recomma/codex/introduce-order-scaling-ser…

d847c33

…vice feat: add Hyperliquid order scaling service

Merge pull request #43 from recomma/codex/add-effectiveorderscaler-an…

952d114

…d-recordscaledorder feat(storage): add order scaler read models and events

chatgpt-codex-connector bot reviewed Oct 31, 2025

View reviewed changes

terwey added 2 commits October 31, 2025 15:37

feat: audit skipped scaled orders

3000174

Merge pull request #47 from recomma/codex/update-scaler-audit-model-f…

acc4fa7

…or-skips feat: audit skipped scaled orders

chatgpt-codex-connector bot reviewed Oct 31, 2025

View reviewed changes

terwey added 2 commits October 31, 2025 16:33

fix(engine): preserve take-profit stack ratios

dc44899

fix(storage): use generated take-profit stack query

d21a35d

terwey changed the title ~~feat(storage): add order scaler persistence~~ feat(orderscaler): introduce order scaling with config, API & audits; integrate engine/emitter Oct 31, 2025

terwey merged commit ebcf246 into main Oct 31, 2025
1 check passed

terwey deleted the codex/update-schema-and-queries-for-orders branch October 31, 2025 15:46

feat(orderscaler): introduce order scaling with config, API & audits; integrate engine/emitter #42

feat(orderscaler): introduce order scaling with config, API & audits; integrate engine/emitter #42

Uh oh!

Conversation

terwey commented Oct 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Refs: #43, #44, #45, #46, #47

Uh oh!

terwey commented Oct 31, 2025

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

terwey commented Oct 31, 2025

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

terwey Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

terwey commented Oct 31, 2025

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

terwey Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connector bot Oct 31, 2025

Choose a reason for hiding this comment

Uh oh!

terwey commented Oct 31, 2025

Uh oh!

chatgpt-codex-connector bot commented Oct 31, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

terwey commented Oct 31, 2025 •

edited

Loading