Tuesday, 17 November 2026. TradingView has expanded the tools around Pine development and market measurement this week. The useful question is what can now be checked more precisely: compiler feedback, the classification behind volume delta, the timing of an earnings notification, or the effect of choosing a different bar interval. This report follows last week's actual source cutoff and keeps new capabilities separate from older technical context.

TL;DR — 17 November

  • Pine through MCP: new development tools and daily call allowances make a bounded, recorded experiment more useful than an open-ended optimization loop.
  • Tick-based volume and custom seconds: check classification, available history and session alignment before comparing signals or returns.
  • Earnings alerts: distinguish the issuer's release time from the alert's firing time; do not assume a simulated reaction was executable.
  • Release discipline: no November Pine language entry was verified. The Copilot article has a date conflict, and DAOL's current announcement describes an earlier connection. The detailed sections below preserve those limits.
ℹ️
AI-generated educational briefing — not investment advice.
This report was prepared with AI from the linked public sources. Verify the original material and test implementations independently. It is not a recommendation to buy, sell or trade any investment.

Get TradingView updates in your inbox

Follow Pine Script, screening tools and platform changes alongside MetaTrader in the shared Trading Development newsletter.

Select Trading Development in the free signup flow. Already a member? Sign in and update your email preferences. Unsubscribe anytime.

Choose Trading Development

At a glance

SubjectWhat matters for development and research
Pine tools through MCPThe 16 November announcement adds direct development operations. Record source revisions, settings, errors and a finite trial budget.
Tick-based volumeThe 13 November release changes measurement options. Classification method and incomplete history must remain visible.
Custom second intervalsThe 10 November announcement followed our prior cutoff. Treat interval selection as a research parameter, with a separate holdout.
EPS and revenue alertsThe 12 November feature supports event monitoring. Preserve publication, observation and notification clocks separately.
Copilot modelsThe current article is dated 16 November; the index points earlier. Compare correctness on identical tasks, not vendor rankings alone.
Broker integrationsEquiti has a 12 November announcement. DAOL describes a 5 November connection; verify entity, account and total-cost assumptions.
Embedded libraries 33.1Boundary-day coverage for 10 November. Review widget teardown, overnight sessions and language-independent restrictions.
Pine, screeners and distributionNo new November language entry verified. Existing timing rules still matter, and the 15 November invite-only conversion milestone is current.
SuperTrend researchA 12 November author simulation offers a sensitivity exercise. It is not independently reproduced performance evidence.

1. Pine development through MCP: useful tools need a bounded experiment

TradingView’s 16 November MCP announcement adds Pine validation, script inspection, indicator execution, backtests, script search, permitted source retrieval, and creation or updating of saved scripts. Source access covers your own and open-source scripts; protected and invite-only source remains closed. Starting 17 November, included daily MCP allowances are 10 calls for Essential and legacy plans, 20 for Plus, 50 for Premium and 150 for Ultimate. The paid AI add-on provides 500 daily calls, with 50 per minute. All MCP calls count, not just Pine operations. TradingView describes personal-use restrictions, including no bulk extraction, resale, model training or automated trading bots.

Practical implication: a development assistant can work against actual compiler feedback and a specified test case. That reduces one source of guesswork, but successful compilation establishes only that the program meets the compiler’s rules. It does not establish that the signal uses information available at decision time, that a simulated fill was attainable, or that a result survives another sample.

A sensible first project is deliberately small: write one plain-language rule, record its input values and expected behavior on several known bars, validate the implementation, and inspect the resulting values before requesting a backtest. Save the original source and every accepted revision. Reject a revision that improves the reported result by changing the hypothesis, date range or execution assumptions without an explicit reason.

Treat the call allowance as an experiment budget. Reserve capacity for checking the final candidate and retain the exact symbol, exchange feed, timeframe, session, date range, settings and returned errors. A loop that keeps asking for a better strategy until calls run out selects among many trials while hiding how many failed. Our recommendation is a fixed stopping rule and a separate holdout, not an autonomous search for the best-looking equity curve. These are proposed controls; this briefing did not connect an MCP client, save a Pine script or run a backtest.

2. Tick-based volume tools: classification and completeness come first

The 13 November announcement introduces tick-based calculations for Footprint, Volume Delta and Cumulative Volume Delta. Classification uses exchange-provided aggressor information when available, then bid/ask information, then price changes between ticks. The selected method is displayed. The option is off by default and requires Ultimate; chart timeframes and the CVD anchor must be no greater than one day. Available history is bounded by up to three million ticks. Incomplete Footprint history can produce a partial session with a warning; insufficient history to cover a CVD anchor or at least one complete Volume Delta bar produces an error.

Why this matters: “tick-based” does not make every observation exchange-labelled buyer or seller aggression. Nor does a blank or errored result mean balanced buying and selling. Those distinctions belong beside the measurement before it is used in a trading rule.

For a controlled comparison, retain the same instrument, venue, trading session, anchor, chart interval and observation window. Record both calculation modes and the method shown in the legend. Start with a window for which each mode has complete data; then separately investigate incomplete-history behavior. Keep a screenshot and a short log of availability warnings, rather than silently dropping inconvenient sessions.

A useful research question is whether a predefined signal is stable when its measurement changes. Count disagreements and locate them in time before examining returns. A disagreement concentrated at session boundaries may expose an anchoring assumption; one concentrated in thin trading may warrant closer examination of classification and data coverage. Those are hypotheses to investigate, not conclusions demonstrated here. If the test later becomes a strategy, preserve a genuine out-of-sample period and realistic costs. A more granular input is not, by itself, evidence of a profitable edge.

3. Custom seconds: more resolution also means more opportunities to overfit

Custom second intervals were announced on 10 November at 13:44 UTC, after last week’s 06:14 UTC research cutoff. Ultimate can select intervals from 1 to 299 seconds. The standard 1, 5, 10, 15, 30 and 45-second choices remain available from Plus. TradingView lists support across web and mobile charts, Pine indicators and strategies, alerts and Bar Replay.

Development implication: an interval such as 17 or 90 seconds is now a parameter that deserves the same discipline as a moving-average length. Searching many intervals and publishing only the winner makes ordinary noise look unusually persuasive.

Choose a small set of intervals for an operational reason before inspecting performance. Specify how bars align to the exchange session, how the first and last partial intervals are handled, and whether the rule acts at close or during a forming bar. Compare equivalent elapsed-time windows, not simply an equal number of bars. One hundred bars represent very different exposure on different resolutions.

Retain the unavailable-data cases as well. A menu option does not guarantee identical history, entitlement, latency or coverage for every instrument. For execution research, document spreads, fees and slippage at the times the strategy would actually trade. Shorter bars can increase apparent responsiveness while making those omitted costs more consequential. Keep the final interval choice fixed for the holdout; if it changes after inspection, that period has become part of development.

4. Earnings alerts: measure the notification clock, not an imagined instant

TradingView’s 12 November EPS and revenue alerts support individual NYSE, NASDAQ and AMEX symbols. Conditions include publication, an earnings beat or miss, and a value above or below a threshold. They consume the technical-alert allowance. Essential and higher plans can expire an alert after the next report; open-ended operation requires Premium or Ultimate. Multi-condition alerts, watchlist alerts and presets are not supported for this feature. The notification supplies actual versus estimated figures and its firing time.

Research implication: a useful event study needs several clocks. Keep the issuer’s publication time, the data observation time, the alert’s firing time and the earliest price at which the proposed rule could act. These are distinct observations. A chart move around an earnings announcement is not proof that an alert could have captured it.

Before setting a rule, define which metric is being compared and the estimate available beforehand. A surprise calculated with a later revised estimate is not the same experiment. Define handling for missing estimates, delayed reports, after-hours sessions and a condition that never fires. A next-report expiry is a lifecycle choice, not a promise that a qualifying event will occur.

For a paper exercise, select a fixed set of upcoming companies, register the condition before each report and collect every outcome. Do not keep only the dramatic winners. Report notification delay separately from subsequent price movement, and distinguish association from causation: guidance, margins and other simultaneous disclosures can matter too. No earnings alerts were created or triggered as part of this briefing.

5. Copilot choices: compare correctness on the same task

The current Copilot model-choice article is dated 16 November and lists DeepSeek V4.1 Flash, GPT-6 Luna and Claude Sonnet 5.5, with reasoning selectable per chat. TradingView’s comparative quality and speed figures are its own assessments. A purchased refill resets the allowance rather than adding to an unused balance; its stated conditions include auto-renewal, and the billing date stays unchanged.

Date caveat: the blog index and older comments associate this item with 9 November. We therefore describe the current dated article, not a verified first launch on 16 November. That distinction prevents an edited announcement from being counted as an entirely new release.

Our suggested evaluation is deliberately independent of model branding. Give each candidate the same small specification, reference material and expected examples. Score whether it identifies missing assumptions, keeps the requested inputs unchanged, explains timing correctly and passes external checks. Keep unsupported claims and silent changes to the strategy definition in the failure log. Polished explanations are useful, but they should not offset an incorrect implementation.

Separate the cost of answering from the cost of verification. The former is visible in an allowance; the latter includes reviewing source, compiling, inspecting values and testing failure cases. This briefing did not compare the listed models experimentally, purchase a refill or alter account/model defaults. The practical recommendation is a reproducible evaluation, not a claim that one option is best for every Pine task.

Keep improving your TradingView workflow

Get the next research edition with concrete platform changes, source links and limits worth testing. TradingView and MetaTrader share the Trading Development newsletter.

Select Trading Development in the free signup flow. Already a member? Sign in and update your email preferences. Unsubscribe anytime.

Follow Trading Development

6. Broker coverage: Equiti is fresh; DAOL needs its earlier connection date

Equiti’s 12 November announcement describes forex and CFD trading from Supercharts and quotes a commission of $3.50 per lot per side. This is an advertised commission component, not an all-in trading-cost estimate. Availability and the contracting entity must be checked for the reader’s own jurisdiction and account.

The current DAOL article, dated 12 November, concerns Korean and US equities and ETFs, but explicitly dates the connection to 5 November. It belongs here as a current explanation of earlier availability, not a newly connected broker this week. Cached index dates also differ from the current article.

Practical implication: a chart integration is only one layer of the trading arrangement. Before assessing a broker, verify the exact legal entity, eligible instruments, account currency, market-data entitlements, supported order types and the contract specification behind “one lot.” Then account for spreads, financing, exchange charges and currency conversion as applicable. A commission headline cannot settle those questions.

For developers, keep research prices and executable prices distinguishable. A strategy observed on one venue’s data cannot be assumed to fill at that same price through a different account. Record the mapping between chart symbol and broker instrument and investigate rejected or unsupported orders in a controlled environment before considering live use. We did not open, connect or operate a brokerage account, and this coverage is not a broker recommendation.

7. Embedded chart developers: version 33.1 has lifecycle and session work

The official Advanced Charts and Trading Platform 33.1.0 notes are dated 10 November. No publication time was available in the checked material, so this is a boundary-day catch-up omitted from the previous report, not a claim that it appeared after that report’s cutoff. These are licensed embedded libraries, distinct from a new Pine language release or a universal Supercharts rollout.

The notes include overnight/24-hour subsessions, the Trading Platform ntc quote field for overnight prices, stable drawing identifiers for restrictions across languages, and a Broker API destroy lifecycle method. Fixes address areas including study limits, Save As behavior and daily resetting of Session VWAP.

Implementation review: test the whole widget lifecycle, not just the first successful chart load. Mount a widget, subscribe, navigate away, remove it and return. Confirm that teardown releases the connections and listeners the application owns, and that remounting does not double callbacks. Log enough identifiers to distinguish a new subscription from a stale one without capturing account secrets.

For overnight sessions, specify the exchange calendar, timezone and daylight-saving behavior. Include bars on either side of midnight and a session change in the test data. For drawing restrictions, test each supported language against the same intended restrictions rather than matching translated display text. These are proposed regression checks derived from the release’s scope; no application upgrade or integration test was performed for this report.

8. Pine and screeners: no verified new language entry, and old timing hazards remain

The checked Pine release notes still end with October 2026. The screeners category did not show a newer November announcement at the research cutoff. That is a statement about the sources checked, not proof that every user-facing component is unchanged. MCP tooling, new chart intervals and volume calculations should not be relabelled as a newly released Pine language version.

Confirmed higher-timeframe context: TradingView’s repainting documentation explains the paired use of an offset expression and barmerge.lookahead_on for confirmed higher-timeframe requests. The offset and lookahead setting belong together; omitting the offset can introduce future information on historical bars. Waiting for the chart’s bar close alone does not confirm a still-forming higher-timeframe bar. Using confirmed information introduces delay, which must remain visible in the strategy definition.

Running-alert context: the alerts documentation describes a server-side snapshot of the script and its inputs, symbol and timeframe when an alert is created. Editing the chart afterwards does not update that running instance. Keep a record of which revision each alert is meant to execute; verify a replacement’s conditions before retiring an old instance. This is existing behavior, not a new November feature.

A useful acceptance test writes down the earliest permitted decision time for every input. Compare that specification with what the code actually requests and when its state changes. Include startup, missing history and session boundaries. If a revision moves a signal earlier, ask what additional information it used before calling the change an improvement.

9. Invite-only distribution: the 15 November milestone is now current

The current invite-only script rules say conversion of private invite-only publications to public visibility is available from 15 November, subject to moderation. The stated 1 December change restricts adding users or updating a private script with more than five users until it is made public or reduced to that limit. This follows the 6 November terms announcement covered last week; the fresh point is the milestone becoming current, not a second policy launch.

Public visibility for an invite-only publication does not mean its source code becomes open-source. Keep publication discoverability, code visibility and access entitlement as separate concepts when explaining a product to users.

Our suggested maintenance exercise is an inventory, not an immediate migration: record each script’s visibility, user count, dependencies, description quality and moderation readiness. Confirm which behavior is actually restricted before changing access. Communicate any approved change precisely and retain the old publication’s identity. This briefing did not change script access, enroll users or move subscribers; it does not interpret the rules as legal advice.

10. Community research: use SuperTrend as a controlled sensitivity exercise

A 12 November ProEA article on SuperTrend settings compares ATR lookback, multiplier, smoothing and close-versus-wick triggers. Its worked exercise uses 96 simulated 15-minute bars after a 200-bar warm-up. This is the commercial author’s own illustrative simulation, not an independently reproduced market backtest. It discusses existing implementations; it is not evidence that SuperTrend itself is new.

The linked TradingView script pages were unavailable to this research environment because of their access restrictions. We therefore did not independently inspect their source and do not endorse the article’s broader implementation census. The accessible article is evidence for what its author demonstrates and claims, with that boundary intact.

A worthwhile research approach: reconstruct a tiny, fully specified example in an environment where you can inspect every intermediate value. Fix prices, warm-up length, ATR initialization and smoothing, band carry-forward rules and the exact reversal trigger. Change one factor at a time. Explain each differing signal before aggregating results. If two implementations disagree, a parameter name that looks identical is not enough to establish equivalence.

Only after the implementation matches its specification should the study move to real data. Register the market universe, dates, cost assumptions and evaluation measures before running it. Treat every variant tried as part of the research history. An indicator’s visual reversals are not executable fills, and a useful explanatory simulation does not demonstrate future profitability. This report proposes the exercise; it does not claim to have run or validated it.

11. A repeatable research sequence for this week

The new tools make it easier to generate variations. The bottleneck remains deciding what would count as evidence. The following is an editorial research framework, not a TradingView release claim or a recommendation to trade.

  1. Write the hypothesis before opening the results. State the observation, decision time, expected direction and circumstances that would falsify it. Decide which markets and sessions belong in scope and why.
  2. Freeze the measurement contract. Record venue, symbol, timezone, session, chart resolution, anchor, data availability and calculation mode. Keep missing observations separate from numerical zeros. For earnings, retain the relevant estimate and separate issuer, feed and alert times.
  3. Verify values before returns. Check a small number of bars or events by inspection. Include a boundary, a missing-data case and an example where the condition should remain false. Keep compiler success separate from semantic correctness.
  4. Specify execution assumptions. Decide when an order becomes eligible, which price could fill it and what costs apply. Document each assumption that cannot be directly measured. An attractive result under idealized fills is a sensitivity case, not the base conclusion.
  5. Separate development from evaluation. Keep an untouched chronological holdout and record all trials. If the holdout influences a parameter choice, mark it as used and obtain a new evaluation sample rather than continuing to call it independent.
  6. Report failure as evidence. Retain rejected hypotheses, unstable parameters, unavailable histories and timing mismatches. Explain why a candidate was discarded so the next iteration does not unknowingly repeat the same search.

The official strategy documentation is the implementation reference for broker-emulator behavior and settings such as commission, slippage and limit-fill assumptions. Those controls describe a model of execution; they do not certify a live fill. Make the model explicit, compare plausible alternatives and avoid treating the most favorable setting as the expected outcome.

A compact final experiment record should include the source revision, full configuration, data window, trial count, result and limitations. For MCP-assisted work, also retain which steps the assistant proposed and which were independently checked. For Copilot comparisons, score the same task rather than comparing different demonstrations. For custom intervals or tick-based inputs, preserve the original measurement alongside the candidate so a difference can be investigated.

12. Source window, date handling and limits

This edition continues from the previous report’s research cutoff of 10 November 2026, 08:14:27 SAST (06:14:27 UTC). Its current cutoff is 17 November 2026, 08:21:05 SAST (06:21:05 UTC). We read the linked official announcements, release notes and documentation in full and retained dated source evidence. The community item is identified separately as the author’s own work.

Dates need interpretation. The custom-second announcement demonstrably follows last week’s cutoff. The embedded-library release shares that calendar date but lacks a verified time. The Copilot index conflicts with the current article date, and DAOL’s current article describes an earlier connection. We preserve these distinctions rather than manufacturing a uniform list of brand-new releases. Earlier policy and technical documentation are labelled as context.

Source availability and product entitlements may differ by plan, region, instrument and account. We did not execute Pine programs, inspect inaccessible protected code, measure provider latency, validate vendor model benchmarks or place orders. Proposed tests and practical implications are editorial analysis, not completed experiments. None of the indicators, integrations or research approaches described here establishes profitability or suitability for a particular reader.

Use the inline original sources to verify a behavior before relying on it, preserve your own configuration and evidence, and keep the distinction between an announcement, an implementation, a simulation and an observed live result.

Forwarded this TradingView edition?

Get your own copy through Trading Development, our shared newsletter for TradingView, Pine Script, MetaTrader and MQL.

Select Trading Development in the free signup flow. Already a member? Sign in and update your email preferences. Unsubscribe anytime.

Join Trading Development
Share this post
The link has been copied!