The short version
- TTSEC launched its Data Hub on 2 September 2026. The regulator already published downloadable Excel data; the new pages also display Excel and CSV export options.
- The date of a new dashboard is not the date of every observation inside it. Check the selected series and period.
- TTSE's 8 September figure of TT$3.71 million is a Composite total. The report lists SME trading separately.
- An AI summary needs checked inputs and traceable calculations. A public dataset does not establish a reliable trading strategy.
If you research Trinidad and Tobago's capital market, a central place to find the regulator's statistics should save some searching. The useful question is what you can do with the records once you find them. Can you identify the period, understand the units and reproduce a result without guessing what a column means?
The Trinidad and Tobago Securities and Exchange Commission's new Data Hub is worth examining on those terms. It should not be dismissed as a collection of pictures that software cannot read. Equally, finding a spreadsheet does not prove that a model trained on it can make useful investment decisions.
The release says Excel data already existed
TTSEC's 2 September 2026 launch release says securities-market data had been available through downloadable Excel spreadsheets and its Semi-Annual Securities Market Bulletin. The new hub adds an interactive way to locate and compare statistics. Describing the earlier service as scattered PDFs alone would miss an important part of what the regulator provided.
The release lists collective-investment-scheme assets and flows, repurchase agreements, government bond issues, market capitalisation, over-the-counter transactions and private placements. Those are different kinds of financial activity. You need to choose the series that answers your question, rather than treating every value on the site as a stock-exchange trade.
On 13 October, the Securities Market Statistics page displayed a working chart and table, with export-menu options for Excel, CSV and a chart image. The CIS Market Statistics page also lists export options. This review checked the controls and displayed data. It did not download a data file or confirm a documented public API.
In the securities page's initial view, the Average CIS Yield table ran from 30 September 2022 to 30 June 2025. That is a useful reminder to inspect the period before calling a figure current. It is an observation about the view checked for this article, not proof that every series on the hub ends in June 2025.
If you need a later period or a definition that the screen does not explain, check the accompanying material or ask TTSEC. Avoid making your own assumptions about whether a yield is annualised, how an average was calculated or when a historical figure was revised. An attractive chart cannot answer those questions by itself.
Read the exchange's totals with their labels
The TTSE Daily Market Quote for 8 September 2026 reports Composite trading of 495,864 shares worth TT$3,708,930.34 across 157 trades. In a separate SME row, it reports 2,840 shares worth TT$26,673.08 across eight trades. Calling the Composite row the entire exchange would erase that distinction.
The report also separates ordinary shares, preference shares, SME shares and other categories, with fields for last sale date and volume. Its historical date selector matters: a page header or a current ticker can refer to a different day from the report you requested. Check the date above the table you actually use.
Market capitalisation and trading value answer different questions. The first values outstanding shares at the relevant prices; the second records the value traded during a period. You cannot expect the whole market's capitalisation to change hands each day. To discuss turnover, you need a clearly defined ratio, matching coverage and observations across time.
One session also cannot establish what is normal or explain why investors traded. You would need a longer record and evidence for any proposed cause. Claims about controlling families, investors' motives or how a local model performs should not be inferred from one day's total.
New listings need the same care. The exchange's 21 July 2026 West Indian Traders announcement identifies it as the fifth company on the SME Market, after an offer of 5,062,500 ordinary shares at TT$2 each. A count of recent listings needs to include the SME market. A new listing still does not guarantee frequent subsequent trading.
Keep the AI task narrow enough to check
There are useful research tasks short of forecasting prices. You might want help locating the definition of a series, drafting code to combine public tables or identifying a mismatch between a chart and a spreadsheet. Each task has an output you can inspect against the source.
Choose one of those tasks before selecting a tool. If you want a monthly report, write down the questions your team must answer and the records it uses now. A simple spreadsheet may be enough. If you try an AI assistant, judge whether it reduces the work of producing a correct report, including the time spent checking and fixing its output.
Keep the original records unchanged and work on a copy. Ask the tool to show which source row supports each number it uses. Recalculate totals and percentage changes yourself or with checked formulas. A fluent explanation can hide a wrong unit or a denominator taken from a different period.
Do not ask an assistant to fill a missing observation with a plausible value. Keep missing data visibly separate from zero. If you use an estimate for a research exercise, label it, explain the method and keep the source value available so a colleague can see what changed.
Robo-advice is a separate question
Guardian's official Genius page describes algorithmic portfolio management using pre-selected exchange-traded funds, with a questionnaire used to assign a model portfolio and automatic rebalancing. It lists both USD and TTD portfolio options. That description is broader than a dashboard showing holdings, and it does not establish a system for predicting individual TTSE share prices.
You cannot diagnose that product's performance from TTSE turnover alone. A fair assessment would need its actual mandate, holdings, costs and results over a stated period. This article has not audited Genius, compared it with competing services or assessed whether it is suitable for any reader. It is mentioned to clarify the kind of automation being discussed.
Build a record someone else can reproduce
For a first data review, choose one public series and a clearly bounded period. Save the source link, the retrieval date and the exact selection you made. Record the units and any notes about coverage. If you receive a revised file later, keep the earlier version so you can explain why your report changed.
Check a handful of rows against the displayed source before importing the full file. Look at the earliest and latest dates, a blank cell and an unusually large value. This is an initial spot check, not proof that every row is correct. Extend it if you find a mismatch.
When combining tables, check the period and category in each one. A quarterly fund statistic cannot be treated as a daily share price. Keep TT-dollar and US-dollar amounts distinct. If you convert currencies, document the rate, date and reason, and retain the original amounts beside the converted values.
For a share-price study, keep the last trade date and trading volume alongside the displayed price. An unchanged quoted price does not necessarily mean a trade occurred at that price that day. Check how the source handles non-trading days, suspended securities and changes such as share splits before interpreting a series.
If you go further into model testing, state what the model is meant to predict and when the necessary inputs would have been available. Keep later information out of earlier predictions. Compare against a simple method, use a separate evaluation period and report the cases where the model fails. These are research safeguards, not evidence that a particular strategy will earn money.
Keep confidential portfolios, client identities and account details out of an unapproved service. Public market data is enough to test whether a tool can preserve labels and calculate correctly. Access to a public website also does not settle every question about automated collection or redistribution; check the source's terms and seek permission where needed.
What this review can and cannot conclude
The strongest conclusion is about source use. TTSEC already provided machine-readable files, the new interface displays export options, and the exchange separates trading categories with different totals. Those details change how you should describe the data before making any claim about AI.
The larger questions remain open. This review did not test the hub's complete history, all export formats or an API, and it did not build a trading model. It cannot establish a typical daily turnover rate or the causes of market liquidity. Those would require a separate, documented study.
For a research team, the next useful result is a small report that a colleague can reproduce from the public source. Show the series, date range and calculations, and list any unresolved definitions. If an AI tool helps you produce that report more accurately or with less checked effort, you have a specific benefit to assess.
None of this is a recommendation to buy a security or use a particular investment service. If you are making an investment decision, consider your circumstances and seek advice from an appropriately authorised professional. A convincing chart, export file or AI-generated answer cannot remove the risk of loss.
Frequently asked questions
Does TTSEC provide machine-readable data?
Its September launch release says downloadable Excel spreadsheets were already available. The current hub pages also display Excel and CSV export options. The review did not test every export or confirm a documented public API, so check the specific dataset you need.
Was TT$3.71 million the entire exchange total on 8 September?
That amount is the Composite trading value shown in the dated report. SME trading appears in a separate row. Keep the report's categories and dates attached to the figures, rather than presenting one subtotal as every transaction on the exchange.
Can one quiet session prove that an AI model cannot work?
No. A single session does not establish normal trading conditions or test a model. You need to define the task, inspect the relevant history and evaluate the output. An assistant that helps check records is doing a different job from predicting prices.
What should I check before using an AI market summary?
Check the source, date, units and category for each figure, then verify the calculations. Keep missing observations separate from zeros and review any assumptions. Do not use an unchecked summary as an instruction to trade or upload private account information.
Sources and disclosure
- TTSEC Data Hub launch release, 2 September 2026
- TTSEC Securities Market Statistics
- TTSEC CIS Market Statistics
- TTSE Daily Market Quote, 8 September 2026
- TTSE West Indian Traders listing announcement, 21 July 2026
- Guardian official Genius product description
Updated 13 October 2026 after reviewing the official release, live data interface and dated exchange report. This article is about checking public data, not personalised investment advice.