Maple Vaultis processes continuous market data streams through a predictive model, converts the output into risk-adjusted signals, and publishes the resulting logs for independent review. No discretionary trading calls, no hidden ledger.
Every signal the model issues is timestamped before its outcome is known. Members can cross-reference entries against public market data, which is the basis of the verification badge described below.
| Timestamp (UTC) | Asset class | Signal type | Model confidence | Verified outcome |
|---|---|---|---|---|
| 2024-11-03 09:14 | UK Equity basket | Long bias | 61% | Verified |
| 2024-11-03 08:47 | UK Gilts, 10yr | Hold / no action | 54% | Verified |
| 2024-11-02 16:02 | EUR/GBP | Signal withheld — risk filter | — | Filtered |
| 2024-11-02 11:19 | US Sector rotation | Reduced exposure | 58% | Verified |
Illustrative log extract for demonstration purposes. The full, continuously updated ledger is available within the terminal after registration.
A signal is hashed and timestamped at the moment it is generated, before the outcome is known. Once the relevant market data settles, the outcome is appended to the same record rather than overwritten. Members can inspect the sequence and confirm that no entry was edited or removed retroactively. This does not guarantee future accuracy — it simply means past claims can be checked against a fixed record rather than taken on trust.
The platform does not output buy or sell instructions directly from a black box. Each signal passes through four distinct, auditable stages before it reaches a member's terminal.
Price feeds, order book depth and macro releases are pulled continuously from licensed data providers and normalised into a common schema.
A statistical model trained on historical sequences estimates a probability distribution of near-term outcomes, not a single fixed prediction.
Signals below a confidence threshold, or those that would breach a position's exposure limit, are withheld automatically rather than passed through.
What remains is written to the public log with a plain-language rationale, so the reasoning behind the signal is visible, not just the outcome.
In practical terms, this means the model's role is bounded: it proposes, the risk filter constrains, and the log records. No stage operates on full discretion, and every rejected signal is as traceable as an accepted one. Confidence figures shown in the terminal describe statistical likelihood based on historical patterns — they are not a forecast of certainty, and past log entries are not an assurance of future results.
Maple Vaultis was shaped around a specific constraint: users who move between time zones and cannot monitor markets on a fixed schedule need a system that behaves predictably in their absence. That shaped most of the design decisions below, from how automation handles connectivity gaps to how uptime is reported.
The platform does not replace an investor's own judgement. It narrows a large data set down to a small number of risk-checked, logged signals, so that review and decision-making take minutes rather than hours.
Signal generation, risk filtering and logging run server-side on a fixed schedule independent of a member's location. The sections below describe how each benefit is implemented, not just what it is called.
The model runs on UTC-based schedules rather than a local market open, so review windows stay consistent regardless of where a member is based.
| Scheduling basis | UTC, 24h |
| Member action required | Review only |
| Manual override window | 15 min |
The log and terminal are designed to load on constrained connections, with data tables rendering before any secondary assets.
| Core view payload | < 200 KB |
| Offline log cache | Last 50 entries |
| Mobile data mode | Supported |
Exposure limits are set once per account and enforced at the filtering stage, so a missed review window cannot silently increase risk.
| Per-position cap | Set by member |
| Breach response | Signal withheld |
| Adjustable frequency | Anytime |
In practice, this means a member travelling between time zones does not need to adjust alerts or stay awake for a particular market open. The automation logic is fixed in advance, and any change to risk parameters takes effect from the member's next login rather than requiring continuous supervision.
The detail below is provided so technically-minded members can assess the system on its mechanics rather than its marketing.
These are the questions most often raised during onboarding calls. They are answered here in the same terms we would use on a call.
Yes. Confidence scores describe a statistical likelihood based on historical patterns, not a guarantee. Markets can behave outside the range the model was trained on, and the risk filter reduces exposure rather than eliminating it.
No. The model proposes a signal; the risk filter can withhold it; the member decides whether to act on what remains visible in the terminal. There is no stage where funds move without an account-level permission already set by the member.
The outcome is appended to the same log entry as the original signal, including entries where the model was incorrect. This is deliberate: a log that only shows favourable outcomes would not be verifiable.
From licensed third-party data providers covering equities, fixed income and major currency pairs. The platform does not generate its own price data or rely on unverified sources.
Each log entry includes a short, plain-language rationale. For a fuller technical description of the modelling approach, the methodology documentation is available on request.
Access is granted manually, following a short verification step, rather than opened to anyone who signs up. This keeps the member base, and the log itself, aligned with its stated purpose.
Access TerminalNo credit card is required to request access. Registration includes a brief identity verification call before terminal credentials are issued.
