@battlegrid/mcp-server 32.0.67 → 32.0.70
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +104 -0
- package/SKILL.md +1 -0
- package/dist/index.d.ts +1 -1
- package/dist/index.js +1 -1
- package/package.json +1 -1
- package/skills/EXPORT.json +5 -4
- package/skills/battlegrid-onboarding/SKILL.md +52 -0
- package/skills/battlegrid-strategy-authoring/SKILL.md +31 -5
- package/skills/battlegrid-strategy-examples/SKILL.md +13 -6
- package/skills/battlegrid-strategy-examples/references/tradingview-ports.md +14 -2
package/README.md
CHANGED
|
@@ -38,6 +38,110 @@ Seeing package `31.x` alongside handshake `battlegrid@33.x` — the package **be
|
|
|
38
38
|
|
|
39
39
|
**What this changes for you:** nothing about how you call anything. Upgrading the package no longer waits on a server deploy, and a server deploy no longer strands you on a package that names the wrong contract — reconnect and the announcement follows. **Contract breaking-change notes are no longer keyed to package versions**, since a contract move is no longer a release here; the v11-and-earlier notes below are kept as history, and the live vocabulary is always discovery.
|
|
40
40
|
|
|
41
|
+
## Contract history — v146.1 to v146.12 (1,000-bar history, author-set lengths, pivots on five periods, market structure)
|
|
42
|
+
|
|
43
|
+
**Nothing refused at v146.1 through v146.12 that v146.0.1 accepted, with one exception for a v146.11 value no client
|
|
44
|
+
had read.** Twelve MINORs: eleven widen the metric catalog, its parameters and its served values, and one
|
|
45
|
+
(v146.12) narrows two values that v146.11 had just added. Schema hashes move only where the metric enum or the
|
|
46
|
+
column shape is inlined, and no tool is added or removed. Reconnect the proxy, or restart it, so discovery reads the
|
|
47
|
+
widened enums.
|
|
48
|
+
|
|
49
|
+
### Narrowed at v146.12 — two values v146.11 added
|
|
50
|
+
|
|
51
|
+
- **`INTERNAL_STRUCTURE_TREND` and `SWING_STRUCTURE_TREND` read `bullish` or `bearish` only.** v146.11 also served
|
|
52
|
+
`none`, which the server never published; a condition clause that names `none` is now refused.
|
|
53
|
+
- **`swingLength` stops at 100.** The served `maximum` is the lesser of 100 and the depth bound (it was 467 at the
|
|
54
|
+
shipped lookback cap). 101 is refused with a message naming the range 10 to 100. `equalBars` is unchanged.
|
|
55
|
+
|
|
56
|
+
### Changed values — the same schema serves a different value
|
|
57
|
+
|
|
58
|
+
- **v146.1: the readable history is 1,000 bars per timeframe, up from 210.** Every candle-based header's served hold
|
|
59
|
+
domain widens with it: `maxWindow` and `maxAtLeast` are the 1,000-bar floor less the column's own lookback (985
|
|
60
|
+
for an `RSI14 × value` column, where it was 195). Settled values of the seed-converging and path-dependent
|
|
61
|
+
indicators (the EMAs, RSI, Stoch RSI, MACD, PPO, Keltner, the TTM squeeze channel, WaveTrend, QQE, ATR, ADX and
|
|
62
|
+
DI, Supertrend, Parabolic SAR) are recomputed over 1,000 bars and move by small amounts. Finite-window values do not
|
|
63
|
+
move. The Donchian break memory, the swing pivots' reach and CCI's flat carry hold 979, 997 and 980 bars, and the
|
|
64
|
+
swing label reads no value until 1,000 closed bars have been seen. The move is listed in `METRIC_VALUE_MOVES`.
|
|
65
|
+
- **v146.2: `get_onboarding_requirements`' `next` follows the setup path.** The path is
|
|
66
|
+
`TRADING_WALLET → STRATEGY → AGENT → DEPOSIT_USDC → …`, and a rung waiting only on a deposit no longer holds `next`,
|
|
67
|
+
so a new account builds a strategy and an agent before depositing.
|
|
68
|
+
- **v146.4: `budgets.conditionFrameReads` reads 2,048, up from 256.** `CONDITION_FRAME_READ_BUDGET_EXCEEDED` names
|
|
69
|
+
the new figure.
|
|
70
|
+
- **v146.7: an `aggregate` column over a series shorter than its window reads `null`** (TradingView's `ta.sma`),
|
|
71
|
+
where it read the mean of the elements present. This includes `FUNDING_RATE_AGG24` and `OI_AGG24` at a shorter
|
|
72
|
+
fetch. An `aggregate` column's `lookback.used` counts only its newest bar plus its metric's reach.
|
|
73
|
+
|
|
74
|
+
### Widened catalog — metrics and studies
|
|
75
|
+
|
|
76
|
+
- **v146.5: TradingView's six Pivot Points Standard types on the previous UTC day.** These are Traditional R4/R5/S4/S5
|
|
77
|
+
plus Fibonacci, Classic, Camarilla, Woodie and DM: 34 keys, each a level set of `floorPivots`. Its existing level set
|
|
78
|
+
`Pivots` is renamed `Traditional`. A level a type shares with Traditional is read from the Traditional key.
|
|
79
|
+
- **v146.8: the six types on the previous ISO week and UTC month.** `PIVOT_WEEK_*` and `PIVOT_MONTH_*`, 41 keys each,
|
|
80
|
+
with the studies `weeklyPivots` and `monthlyPivots`. Each key is absent until its period is complete.
|
|
81
|
+
- **v146.10: the six types on the previous UTC quarter and year.** `PIVOT_QUARTER_*` and `PIVOT_YEAR_*`, 41 keys each,
|
|
82
|
+
with the studies `quarterlyPivots` and `yearlyPivots`. They are composed from completed UTC months, so quarterly
|
|
83
|
+
levels first read on 2027-01-01 and yearly levels on 2028-01-01, and read absent until then.
|
|
84
|
+
- **v146.11: LuxAlgo Smart Money Concepts market structure, the study `marketStructure`.**
|
|
85
|
+
- `INTERNAL_STRUCTURE_EVENT` / `SWING_STRUCTURE_EVENT` read `BOS↑`, `BOS↓`, `CHoCH↑` and `CHoCH↓`.
|
|
86
|
+
- `INTERNAL_STRUCTURE_TREND` / `SWING_STRUCTURE_TREND` read the trend.
|
|
87
|
+
- `INTERNAL_STRUCTURE_LEVEL` / `SWING_STRUCTURE_LEVEL` read the broken level, which composes over `distance` and
|
|
88
|
+
`levelInteraction`.
|
|
89
|
+
- `EQH` and `EQL` flag equal highs and lows.
|
|
90
|
+
|
|
91
|
+
Each reads `null` where the column's own bars cannot state the full series' value.
|
|
92
|
+
|
|
93
|
+
### Widened input — lengths an author sets
|
|
94
|
+
|
|
95
|
+
- **v146.6: `rollingMax` and `rollingMin` take an `offset`.** At offset 1 or more, a rolling extreme chains into
|
|
96
|
+
`distance` as the breakout level (`HIGH × rollingMax(55, offset 1)` is `ta.highest(high, 55)[1]`, header
|
|
97
|
+
`high_o1_max55`). The chain-outer vocabulary gains `distance`, and `chainedOuters[outer].offsetMinimum` is served.
|
|
98
|
+
- **v146.7: the `ema` transform** (TradingView's `ta.ema`, header `_ema{n}`) with the window unit
|
|
99
|
+
`smoothingLength`. `aggregate`'s window maximum widens from the lookback cap to the depth maximum (936 at the shipped
|
|
100
|
+
cap). Every served bounded window carries `bound` (`lookback` | `depth`).
|
|
101
|
+
- **v146.9: swing `leftBars` and `rightBars`** (default 2, header `_l<left>r<right>` off 2/2) on the seven swing
|
|
102
|
+
metrics. Every served metric now carries `parameters`, and every transform entry carries `metricParametersAccepted`.
|
|
103
|
+
- **v146.11: `swingLength`, `confluenceFilter`, `equalBars` and `equalThreshold`** on the market-structure metrics.
|
|
104
|
+
They print as header tokens `_s<n>`, `_cf`, `_e<m>` and `_atr<threshold>` off their defaults. A condition's level
|
|
105
|
+
reference carries `swingLength` and `confluenceFilter`.
|
|
106
|
+
|
|
107
|
+
### Widened output
|
|
108
|
+
|
|
109
|
+
- **v146.3: every rung `get_onboarding_requirements` serves carries `remedy`** (`{ action, actor }`, `null` when the
|
|
110
|
+
rung is `COMPLETE`).
|
|
111
|
+
|
|
112
|
+
### New refusal
|
|
113
|
+
|
|
114
|
+
- **v146.7: `REPORT_COLUMN_DEPTH_EXCEEDED`.** A window, swing pair or structure length whose one-value read does not
|
|
115
|
+
fit the 1,000-bar history less the column's lookback is refused, and the refusal names the knob. A pair off 2/2 or a
|
|
116
|
+
structure metric is refused under `rank`, `relativeTo`, `aggregate` and `ema`.
|
|
117
|
+
|
|
118
|
+
### Digest move
|
|
119
|
+
|
|
120
|
+
- `REPORT_CATALOG_SCHEMA_VERSION` 34 → 38. The strategy preview's `vocabularyDigest` and the authoring-contract
|
|
121
|
+
digest move with each catalog widening. `stage_strategy_draft`'s input schema grows from 15,470 to 19,972 bytes.
|
|
122
|
+
|
|
123
|
+
## Contract history — v146.0.1 (the pegs' and breadth's history starts before the deploy)
|
|
124
|
+
|
|
125
|
+
**Re-worded text at v146.0.1, behind unchanged schemas.** v146.0.1 is the contract on top of v146.0.0. Nothing a
|
|
126
|
+
client sends is refused that v146.0.0 accepted, and no schema moves. Reconnecting is not required.
|
|
127
|
+
|
|
128
|
+
### Re-worded text — the `PAR_DEV` gloss
|
|
129
|
+
|
|
130
|
+
- **`PAR_DEV`'s gloss no longer says its series is appended from deploy.** It says the pair's history is seeded at
|
|
131
|
+
boot from the pair's venue candles, so a lookback beyond the history the store holds reads absent. The preview's
|
|
132
|
+
`vocabularyDigest` moves with it; `REPORT_CATALOG_SCHEMA_VERSION` does not.
|
|
133
|
+
|
|
134
|
+
### Changed values — the same schema serves a value where it read absent
|
|
135
|
+
|
|
136
|
+
- **A pair section's lookback reads values from before the v146 deploy.** Each declared pair's warmup windows are
|
|
137
|
+
seeded at boot from its venue's own candles: Binance's for `USDC_USDT`, and Coinbase's for `USDT_USD`, whose 4h bars
|
|
138
|
+
are folded from its 1h candles because Coinbase serves no 4h. A quiet minute is carried only while the pair traded
|
|
139
|
+
within its freshness target, as on the live tape. A slot neither the venue nor that rule can decide still reads
|
|
140
|
+
absent.
|
|
141
|
+
- **A market section's lookback reads breadth from before the v146 deploy.** A bar the live pass left unwritten is
|
|
142
|
+
healed every hour from the stored per-coin bars by the same fold, and only where every coin of the roster can
|
|
143
|
+
measure it.
|
|
144
|
+
|
|
41
145
|
## Contract history — v146 (market breadth and the stablecoin pegs are series a section reads)
|
|
42
146
|
|
|
43
147
|
**Breaking at v146: inputs refused, one section field required, inputs and vocabularies widened.** Market breadth
|
package/SKILL.md
CHANGED
|
@@ -15,6 +15,7 @@ from BattleGrid's own server so they name exactly the tools you reach here:
|
|
|
15
15
|
|
|
16
16
|
| For | Activate |
|
|
17
17
|
|---|---|
|
|
18
|
+
| Setting up an account step by step, or finding what to do next — from nothing to an agent armed on Radar | `battlegrid-onboarding` |
|
|
18
19
|
| Playing Market Grid sessions | `battlegrid-arena-play` |
|
|
19
20
|
| Building or changing a strategy | `battlegrid-strategy-authoring` |
|
|
20
21
|
| Composing beyond a bare template — report sections, conditions, trade levels, playbooks | `battlegrid-strategy-examples` |
|
package/dist/index.d.ts
CHANGED
|
@@ -47,7 +47,7 @@ import { Client, StreamableHTTPClientTransport } from '@modelcontextprotocol/cli
|
|
|
47
47
|
* being asked. Move it for a change to THIS package — a proxy fix, a dependency bump, a docs
|
|
48
48
|
* correction. Never move it to track the server.
|
|
49
49
|
*/
|
|
50
|
-
export declare const PACKAGE_VERSION = "32.0.
|
|
50
|
+
export declare const PACKAGE_VERSION = "32.0.70";
|
|
51
51
|
export declare const DEFAULT_URL = "https://mcp.battlegrid.trade/mcp";
|
|
52
52
|
export interface EnvConfig {
|
|
53
53
|
apiKeys: string[];
|
package/dist/index.js
CHANGED
|
@@ -49,7 +49,7 @@ import { Client, SdkHttpError, StreamableHTTPClientTransport } from '@modelconte
|
|
|
49
49
|
* being asked. Move it for a change to THIS package — a proxy fix, a dependency bump, a docs
|
|
50
50
|
* correction. Never move it to track the server.
|
|
51
51
|
*/
|
|
52
|
-
export const PACKAGE_VERSION = '32.0.
|
|
52
|
+
export const PACKAGE_VERSION = '32.0.70';
|
|
53
53
|
export const DEFAULT_URL = 'https://mcp.battlegrid.trade/mcp';
|
|
54
54
|
const MAX_RETRIES = 3;
|
|
55
55
|
const RETRY_DELAYS_MS = [2000, 4000, 8000];
|
package/package.json
CHANGED
package/skills/EXPORT.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"generator": "battlegrid-app/server/scripts/export-mcp-skills.mjs",
|
|
3
|
-
"contractVersion": "146.0
|
|
3
|
+
"contractVersion": "146.12.0",
|
|
4
4
|
"files": {
|
|
5
5
|
"battlegrid-agent-management/SKILL.md": "324ef8181bfcfc99b955d9644ed543c6ab91071153d43f77ccf0a4979b896b09",
|
|
6
6
|
"battlegrid-agent-management/references/agent-stage-examples.md": "dffcc353d61dfc7b5bb6a2b3077fc6bf8c760a34c287cafb3b59b99306f24195",
|
|
@@ -9,16 +9,17 @@
|
|
|
9
9
|
"battlegrid-market-analysis/SKILL.md": "889ba8a805585db5bfed0c72722d8b177174e8b145fb054619a5dc792cc92b09",
|
|
10
10
|
"battlegrid-officers/SKILL.md": "0ac2173d0026b7329171fe24d38185e7717196340778d402c7af1aa74ea34bc8",
|
|
11
11
|
"battlegrid-officers/references/officer-stage-examples.md": "ef46ba26f4b36c9964b5bde95fdf98e5952c3786b77be5f136056cac792f0aec",
|
|
12
|
+
"battlegrid-onboarding/SKILL.md": "e6b078cfdf13f6c66d2d964d2c0e6b9635f26adb164a8e0914849ef5671b9e3b",
|
|
12
13
|
"battlegrid-radar-deployment/SKILL.md": "ef291ec096461ed2b78915adbfa41000b959b5803ae4ebef3084bff4c1e1a746",
|
|
13
14
|
"battlegrid-radar-deployment/references/arena-stage-examples.md": "65af54b67c3248c53fb8911df769ad32824e1f1f1d83ffc007b42a888d5e3076",
|
|
14
15
|
"battlegrid-radar-deployment/references/radar-stage-examples.md": "7a4195011aa764b106567606f7a2a8ae8e9b2bf74c36b3c74a72379969d70af8",
|
|
15
|
-
"battlegrid-strategy-authoring/SKILL.md": "
|
|
16
|
+
"battlegrid-strategy-authoring/SKILL.md": "11869b767def2c3951d0f046e8dadeb5a549568399d88bb2df3844e65e3eaf73",
|
|
16
17
|
"battlegrid-strategy-authoring/references/strategy-stage-examples.md": "f10ecdc91f3660a708eee7a7bacefce2acbb46c9cf7e0ac1454bd54f839f39fd",
|
|
17
18
|
"battlegrid-strategy-doctor/SKILL.md": "4be392e51f3b5e76ff4aa22f61700236e8b1e19dfe286aeda8ed98e48a8457ba",
|
|
18
|
-
"battlegrid-strategy-examples/SKILL.md": "
|
|
19
|
+
"battlegrid-strategy-examples/SKILL.md": "ac54a2171b3a7bcc0d3e6fa5c75bcf1ee04d5bb2cbfc0d050032d44db5a734d3",
|
|
19
20
|
"battlegrid-strategy-examples/references/playbooks.md": "59996c37eea08e7ce398e49304171939e55c968400a1603597d6737faf9e1771",
|
|
20
21
|
"battlegrid-strategy-examples/references/recipes.md": "ee1b35f5341d3878ef8e8b045d0425a06fb25c76149fe841f2afe075909135f4",
|
|
21
|
-
"battlegrid-strategy-examples/references/tradingview-ports.md": "
|
|
22
|
+
"battlegrid-strategy-examples/references/tradingview-ports.md": "ed977b6c190a74777a735cd482e65e65b92b9e8768157b65ea916cabafdf3cd9",
|
|
22
23
|
"battlegrid-trade-analysis/SKILL.md": "ec0b543bc62fe66d629833b10e4bc58548be1fee57f29c04f41f036e860af1ca",
|
|
23
24
|
"battlegrid-trade-proposal/SKILL.md": "8d833022f753dfc5a765eca2fcd272c4f29d22b8444bc00c14ce63571646ba71"
|
|
24
25
|
}
|
|
@@ -0,0 +1,52 @@
|
|
|
1
|
+
---
|
|
2
|
+
name: battlegrid-onboarding
|
|
3
|
+
description: Walk the player through setting up their account — from nothing to an agent armed on Radar — one step at a time, in the order the server serves. Reads the player's readiness, does each step the Commander can do through the skill that owns it, hands each step only the player can do to the app, and stops when the account is battle-ready. Activate whenever the player asks to get set up, to start, what to do next, to walk them through setup, or why their agents cannot trade yet.
|
|
4
|
+
---
|
|
5
|
+
|
|
6
|
+
# Onboarding
|
|
7
|
+
|
|
8
|
+
You conduct the player's setup. You own no act of your own: every step is done by the skill that
|
|
9
|
+
owns it, or by the player in the app. Your job is to read where the player stands, take the one
|
|
10
|
+
step the server says is next, and keep going until the account is battle-ready.
|
|
11
|
+
|
|
12
|
+
## The one rule: the ladder is the only truth
|
|
13
|
+
|
|
14
|
+
Call `get_onboarding_requirements` before you say anything about setup, and again after every
|
|
15
|
+
step you complete. Read its `status`, its `next` and the `remedy` on the step `next` names, as
|
|
16
|
+
served. Never judge readiness from anything else — a deposit you saw, an agent you created, a
|
|
17
|
+
policy you read — never name or count the steps, and never decide an order: the server already
|
|
18
|
+
chose it, and the app's setup bar names the same step from the same read.
|
|
19
|
+
|
|
20
|
+
## Each turn
|
|
21
|
+
|
|
22
|
+
1. **Read the ladder.** If `status` is battle-ready, say the account is set up and stop.
|
|
23
|
+
2. **Take the step `next` names.** Say what it is in the step's own served sentence, then act on
|
|
24
|
+
its `remedy` by who performs it (`actor`):
|
|
25
|
+
- **`COMMANDER` — you do it, with the player's go-ahead.** Ask once, then turn to the skill
|
|
26
|
+
that owns the act and follow that skill's own procedure to the end: the `strategy-authoring`
|
|
27
|
+
skill builds a strategy; the `agent-management` skill creates an agent or resumes a paused
|
|
28
|
+
one; the `radar-deployment` skill puts an agent on Radar or Arena duty. A resume is the act
|
|
29
|
+
to try, not a promise: if the owning skill reports the stop is still breached, offer its
|
|
30
|
+
exits and do not repeat the resume.
|
|
31
|
+
- **`PLAYER` — only the player can do it, in the app.** It is a deposit, a signature, a
|
|
32
|
+
payment or an API key. Say what it is and why it is next. On the web, the control that does
|
|
33
|
+
it sits directly under your reply; elsewhere, tell the player where in the app it lives.
|
|
34
|
+
Never ask for a key, a seed phrase or a signature in chat, and never say it is done until the
|
|
35
|
+
ladder says so.
|
|
36
|
+
- **`PLATFORM` — the platform does it.** Say so. Account activation in particular is done by
|
|
37
|
+
the player's signed-in web app, which also holds the retry if it fails.
|
|
38
|
+
- **`NONE` — the step could not be read.** Say its status is undetermined right now, never
|
|
39
|
+
that it is done, and offer to check again.
|
|
40
|
+
3. **Re-read the ladder** after any step completes, and continue from whatever it then serves.
|
|
41
|
+
|
|
42
|
+
## Before the first deposit
|
|
43
|
+
|
|
44
|
+
While the deposit is still open and the player's agent already exists, prepare the radar ahead
|
|
45
|
+
of it: have the `radar-deployment` skill stage the deployment as a draft and show its preview,
|
|
46
|
+
and do **not** commit it. Committing arms the agent, and an armed agent with no equity trades
|
|
47
|
+
nothing. Arm it when the ladder's `next` reaches it, after the deposit lands.
|
|
48
|
+
|
|
49
|
+
## Tone
|
|
50
|
+
|
|
51
|
+
One step at a time. Lead with what the step gets the player, not with mechanics. When a step
|
|
52
|
+
belongs to the player, hand it over plainly and wait; do not stack the next step on top of it.
|
|
@@ -173,13 +173,39 @@ question you would otherwise guess — and a refusal you would otherwise earn:
|
|
|
173
173
|
transform (and a chained transform) reads its series: the read shape, the window's unit, the price
|
|
174
174
|
basis and what an absent bar does. `seriesHome` says where the metric's series is kept — `candle`,
|
|
175
175
|
`history`, or `null` when the metric has no series, which is why a series transform does not
|
|
176
|
-
compose on it. Window and offset carry a declared `minimum` and a served `maximum` on the
|
|
177
|
-
entry
|
|
176
|
+
compose on it. Window and offset carry a declared `minimum`, a `bound` and a served `maximum` on the
|
|
177
|
+
transform entry, or `null` when none fits. A `lookback` bound is the largest value whose read size
|
|
178
|
+
fits the lookback cap. A `depth` bound — the length of `aggregate` (the SMA, `ta.sma`) and `ema`
|
|
179
|
+
(`ta.ema`) — is the longest average whose one-value read fits the readable depth beside the
|
|
180
|
+
lookback cap (936 for an SMA, 269 for an EMA at the shipped cap); a longer one is refused with
|
|
181
|
+
`REPORT_COLUMN_DEPTH_EXCEEDED`, and `lookback.used` counts only the average's newest bar. For a
|
|
178
182
|
chained column the outer's declaration is `chainOuterParameters[outer]` (served beside `metric`)
|
|
179
|
-
and its bound is `chainedOuters[outer].windowMaximum
|
|
180
|
-
|
|
181
|
-
|
|
183
|
+
and its bound is `chainedOuters[outer].windowMaximum`; `chainedOuters[outer].offsetMinimum` is the least
|
|
184
|
+
offset the chain accepts — `1` for a rolling extreme chained into `distance` (the breakout level), `null`
|
|
185
|
+
on every other chain — so stage the inner's offset at or above it. A chain outer never takes an offset:
|
|
186
|
+
the offset belongs to the inner. The window carries a `unit`
|
|
187
|
+
(`seriesElements`, `seriesChanges` for `efficiency`, which reads one bar more than its window,
|
|
188
|
+
`priorElements` for `pctl`, or `smoothingLength` for `ema`, whose length reads more bars than it
|
|
189
|
+
names — `ema(20)` reads 73). An average reads null until its whole window is present. Stage at most
|
|
190
|
+
that maximum. A chained column's operands come from
|
|
182
191
|
`chainedOuters[outer].operands`, never from the metric's full `spreadOperands`.
|
|
192
|
+
- A metric can declare its own knobs on `parameters` (served on the metric entry, `null` when it
|
|
193
|
+
declares none). The seven swing metrics declare `leftBars` and `rightBars` — the bars before and the
|
|
194
|
+
closed bars after a pivot — default 2/2, minimum 1, `bound: 'depth'`: the pair reads
|
|
195
|
+
`left + right + 1` bars to decide one pivot and must fit the readable depth beside the lookback cap
|
|
196
|
+
(each side's served `maximum` is at the other's default, 933 at the shipped cap); a wider pair is
|
|
197
|
+
refused with `REPORT_COLUMN_DEPTH_EXCEEDED` on `leftBars`. Stage the pair on the column beside its
|
|
198
|
+
transform knobs, and leave it off for 2/2 — an explicit 2/2 is the same column. Off 2/2 the header
|
|
199
|
+
carries `_l<left>r<right>` after the code (`fractalHi_l5r5_1h`). Stage a pair only on a transform
|
|
200
|
+
(or chained outer) whose entry serves `metricParametersAccepted: true`; `rank`, `relativeTo`,
|
|
201
|
+
`aggregate` and `ema` read the stored 2/2 and refuse it. A condition hold over a column off 2/2 is
|
|
202
|
+
served a window of 1: each of its bars needs the full readable depth behind it.
|
|
203
|
+
- The eight market-structure metrics (`INTERNAL_STRUCTURE_*`, `SWING_STRUCTURE_*` event, trend and
|
|
204
|
+
level, `EQH`, `EQL`) declare their own knobs, each served with its `bound`: `swingLength` (`depth`,
|
|
205
|
+
default 50, minimum 10, maximum 100), `confluenceFilter` (`flag`, internal structure only), `equalBars` (`depth`,
|
|
206
|
+
default 3) and `equalThreshold` (`range`: stage exactly one of the served steps, 0 to 0.5 by 0.1).
|
|
207
|
+
Leave a knob off at its default. They read `null` until the column's bars prove a value, so a
|
|
208
|
+
short history reads more `null`. A trend reads `bullish` or `bearish`. They never take `rank`, `relativeTo`, `aggregate` or `ema`.
|
|
183
209
|
- `get_metric_construction_hints` → `rankOrderings`. Present only when rank is composable on that
|
|
184
210
|
metric, already range-gated server-side — read the offered set rather than deriving one from
|
|
185
211
|
the metric's native output.
|
|
@@ -44,7 +44,7 @@ Before committing a CREATE, every "no" here is a decision to state, not an omiss
|
|
|
44
44
|
| `stddev` wN | `<code>_sd<N>` (population σ, the input's units) | `close_sd20` |
|
|
45
45
|
| `linregSlope` wN | `<code>_slope<N>` (least-squares slope per bar) | `EMA20_slope20` |
|
|
46
46
|
| `pctl` wN | `<code>_pctl<N>` (0–100 against the N prior bars) | `RSI14_pctl20` |
|
|
47
|
-
| `rollingMin` / `rollingMax` wN | `<code>_min<N>` / `<code>_max<N>` | `low_min20`, `high_max20` |
|
|
47
|
+
| `rollingMin` / `rollingMax` wN | `<code>_min<N>` / `<code>_max<N>`; at `offset` K > 0 `<code>_o<K>_min<N>` | `low_min20`, `high_max20`, `high_o1_max55` |
|
|
48
48
|
| `classifyZone` / `classifyState` | `<code>_zone` / `<code>_state` | `RSI14_zone`, `ADX14_state` |
|
|
49
49
|
| `levelInteraction` | `<code>_lvl`, `<code>_lvl_b<N>` at a buffer of N bps | `bbUpper20_lvl`, `bbUpper20_lvl_b25` |
|
|
50
50
|
| any, stored `"bars": "all"` | `<code>_dev`, then the transform's suffix | `RSI14_dev_t1` |
|
|
@@ -52,12 +52,17 @@ Before committing a CREATE, every "no" here is a decision to state, not an omiss
|
|
|
52
52
|
|
|
53
53
|
The code carries its tokens in this order: `<code>[_<tf>][_dev][_o<N>]`, then the transform's suffix.
|
|
54
54
|
`_dev` follows the STORED selector — the author chose Developing — whatever bar set the column then
|
|
55
|
-
reads; `_o<N>` appears on a `value` at offset N > 0.
|
|
55
|
+
reads; `_o<N>` appears on a `value` or a rolling extreme at offset N > 0.
|
|
56
56
|
|
|
57
57
|
Non-anchor rungs affix `_ltf` (lower) / `_htf` (regime): `MAalign_htf`,
|
|
58
|
-
`zones_htf_support_dist`. Chains are bounded at two
|
|
59
|
-
`
|
|
60
|
-
`EMA5_EMA13_spread_now` + `_trend`)
|
|
58
|
+
`zones_htf_support_dist`. Chains are bounded at two and take one of two shapes: inner
|
|
59
|
+
`distance`/`spread`/`relativeTo` → a series outer (`trajectory`, `aggregate`, …) or `rank`
|
|
60
|
+
(`EMA5 spread EMA13 × trajectory` → `EMA5_EMA13_spread_now` + `_trend`); or a rolling extreme
|
|
61
|
+
`rollingMax`/`rollingMin` → `distance`, the breakout level (`HIGH × rollingMax(55, offset 1) ×
|
|
62
|
+
distance` → `dist_high_o1_max55`). The level chain needs a price base with a per-bar candle value and
|
|
63
|
+
`offset` ≥ 1 (the catalog serves it as `chainedOuters.distance.offsetMinimum`): at offset 0 the extreme
|
|
64
|
+
includes the bar measured, so a close can never pass it. The offset belongs to the rolling extreme,
|
|
65
|
+
never to a chain outer.
|
|
61
66
|
|
|
62
67
|
**Timeframe references are two families.** *Relative* (`anchor`/`lower`/`regime`) re-resolve
|
|
63
68
|
when the strategy timeframe changes — `regime` is the anchor's ladder successor (1d for a 4h
|
|
@@ -331,7 +336,9 @@ Donchian band, and every candle label classifier (`MA_ALIGN`, `PRICE_ZONE`, `BB_
|
|
|
331
336
|
against the frame's current price, `LAST` — the strategy bar's close under a decision, the last
|
|
332
337
|
traded price on display — not against the higher-timeframe bar's own close, which is up to one
|
|
333
338
|
higher-timeframe bar stale. A distance CHAINED into a series transform keeps every slot on its own
|
|
334
|
-
bar's close, so one series carries one reference basis.
|
|
339
|
+
bar's close, so one series carries one reference basis. A breakout chain (`rollingMax(n, offset 1) ×
|
|
340
|
+
distance`) ends at the distance, so its newest bar is measured from `LAST` too, as the distance to the
|
|
341
|
+
catalogued `DONCHIAN_UPPER_PRIOR` is, and a condition over it is UNRESOLVED on a frame with no last price.
|
|
335
342
|
|
|
336
343
|
**The lane a strategy is deployed to.** The one report-level scalar family is the **session field**:
|
|
337
344
|
five scalars (`fieldPlayers`, `fieldUpBias`, `fieldBiasDir`, `captConc`, `picksSpread`, each headed
|
|
@@ -74,6 +74,13 @@ rules about the columns themselves, not about porting — they live in the skill
|
|
|
74
74
|
the run, and adds nothing to the crossing; `"breakdown low"` mirrors it.
|
|
75
75
|
**Daily-breakout variant on any anchor:** pin the rail at 1d — `dist_donchianHiPrior20_1d crossesAbove 0`
|
|
76
76
|
(`gt 0` for the held state).
|
|
77
|
+
**Any N-bar breakout (Turtle 55, a 10-bar exit channel):** the prior-bars level is a rolling extreme read
|
|
78
|
+
at offset 1 — `HIGH × rollingMax(55, offset 1)` is `ta.highest(high, 55)[1]`, and chained into `distance`
|
|
79
|
+
it is the breakout column: `dist_high_o1_max55 crossesAbove 0` (the breakout bar; `gt 0` holds on every
|
|
80
|
+
close of the run). Mirror with `LOW × rollingMin(N, offset 1) × distance` → `dist_low_o1_minN crossesBelow 0`.
|
|
81
|
+
At N = 20 it reads exactly the stored `DONCHIAN_UPPER_PRIOR` / `DONCHIAN_LOWER_PRIOR` rails. The offset
|
|
82
|
+
must be at least 1 under `distance` (the save refuses offset 0, whose extreme includes the bar measured),
|
|
83
|
+
and window plus offset must fit the column lookback cap.
|
|
77
84
|
**Literal previous-day levels are native**: `dist_PDH gte 0` composes directly. They are TIMELESS
|
|
78
85
|
and take `{rel: 'anchor'}` — see `## Header grammar` in the skill body. The previous UTC day is the
|
|
79
86
|
00:00–24:00 UTC day, and a level read against a bar of the next day is `null`.
|
|
@@ -119,8 +126,13 @@ rules about the columns themselves, not about porting — they live in the skill
|
|
|
119
126
|
and signals (`MACD_LINE`/`MACD_SIGNAL`, `PPO_LINE`/`PPO_SIGNAL` — the line's zero cross is
|
|
120
127
|
`MACD_LINE × crossDetect`, and the signal-line cross is `MACD_HIST × crossDetect`, the histogram being
|
|
121
128
|
line minus signal), the 15-bar rate of change (`ROC15`), the Donchian break memory
|
|
122
|
-
(`DONCHIAN_BROKEN_HI`/`DONCHIAN_BROKEN_LO`), daily pivots
|
|
123
|
-
(`PIVOT_P`/`PIVOT_R1`–`
|
|
129
|
+
(`DONCHIAN_BROKEN_HI`/`DONCHIAN_BROKEN_LO`), daily pivots of
|
|
130
|
+
all six Pivot Points Standard types (Traditional `PIVOT_P`/`PIVOT_R1`–`R5`/`PIVOT_S1`–`S5`, then
|
|
131
|
+
`PIVOT_FIB_*`, `PIVOT_CLASSIC_*`, `PIVOT_CAM_*`, `PIVOT_WOODIE_*`, `PIVOT_DM_*` — a level one type shares
|
|
132
|
+
with Traditional is read from the Traditional key) and the same six types on the previous ISO week,
|
|
133
|
+
UTC month, UTC quarter and UTC year (`PIVOT_WEEK_*`, `PIVOT_MONTH_*`, `PIVOT_QUARTER_*` and `PIVOT_YEAR_*`:
|
|
134
|
+
the daily key with `WEEK_`, `MONTH_`, `QUARTER_` or `YEAR_` after `PIVOT_`, so `timeframe.change("W")`'s R1
|
|
135
|
+
is `PIVOT_WEEK_R1` and `timeframe.change("3M")`'s is `PIVOT_QUARTER_R1`), Williams %R (`WILLR14`), Stoch RSI
|
|
124
136
|
(`STOCH_RSI_K`/`STOCH_RSI_D` — TradingView's "Stoch RSI" %K and %D at 3, 3, 14, 14, on 0–100), the TTM squeeze (`KC_SQUEEZE`), Connors RSI-2 (`RSI2`), the 8/9/21/50 EMAs
|
|
125
137
|
(`EMA8`/`EMA9`/`EMA21`/`EMA50` — `ta.ema(close, 8)` is `EMA8`, its distance `dist_EMA8`), literal
|
|
126
138
|
previous-UTC-day levels (`PDH`/`PDL`/`PDO`), fractal swing structure
|