Official paper ledger
Product · Paper experiment
TapeEdge
Structured market intelligence for asymmetric trades: a confidence-gated decision-support system that turns live market data, human-curated levels, risk rules, and catalyst checks into disciplined action notes.
The research question is not whether an LLM can predict markets. It is whether deterministic gates, source freshness, proxy disclosure, and outcome logging can constrain model overconfidence in a high-stakes decision workflow.
The performance test is deliberately risk-adjusted: can TapeEdge produce competitive returns while deploying less capital, defining downside in advance, and avoiding large drawdowns versus SPY, QQQ, Magnificent 7, or obvious AI/theme baskets?
Public analytics
Paper-account performance.
Loading sanitized TapeEdge analytics...
Social paper ledger
SPY buy-and-hold benchmark
Stock-sleeve sub-index vs S&P 500
Loading stocks-vs-index comparison...
Loading benchmark chart...
Current spread
- Stock sleeve only-
- S&P 500 / SPY-
- Difference-
This chart is sleeve-only. The full TapeEdge Core account ROI is the headline number above.
Not every comparison is the same kind of risk.
SPY, QQQ, Magnificent 7, and theme baskets are buy-and-hold benchmarks: they usually put the whole $1,000 to work. TapeEdge is a gated decision system: it may hold most of the account in cash and only deploy capital when a setup clears its level, risk, data-confidence, and catalyst checks. The question is whether the system can produce competitive returns with less exposure and tighter planned downside.
Raw return
What happened to the full $1,000 paper account.
Deployed-capital return
What happened to the dollars actually placed into trades.
Planned-risk return
How much was made or lost per dollar of predefined stop-risk.
Cumulative trust model
Loading evidence ladder...
Trust score
-
Paper only
No live allocation while evidence is immature.
Comparison days-
Superior days-
Alpha vs SPY-
Stock trades-
Loading trust gates...
Universes
-
Active universes
-
Trades opened latest run
-
Latest confidence
Loading
Allocated paper capital
-
Paper-account base
Aggregate equity
-
Realised plus open P&L across universes
Total return
-
Total P&L
Max drawdown
-
Peak-to-trough
Aggregate seven-universe equity curve
This combines every public paper universe. The stock-vs-SPY chart above is the cleaner apples-to-apples benchmark.
Loading chart...
Decision system
- Runs logged-
- Decisions logged-
- Buy-interest decisions-
- No-trade decisions-
- Setups opened-
- Open risk-
- Last generated-
Universe accounts
Loading universe accounts...
Universe detail
| Universe | Account | Return | Setups | Open / closed | Open capital | Open risk | P&L | Decisions |
|---|---|---|---|---|---|---|---|---|
| Loading universes... | ||||||||
Benchmark comparison
Benchmark tracking loads after the analytics JSON is available.
| Benchmark | Live proxy | $1,000 buy-hold | Benchmark return | TapeEdge vs benchmark | Status |
|---|---|---|---|---|---|
| Loading benchmarks... | |||||
Recent anonymised setups
| Setup | Universe | Status | Opened | Capital | Risk | P&L | R |
|---|---|---|---|---|---|---|---|
| Loading setups... | |||||||
Method
Decision support before prediction.
TapeEdge is being treated as a product experiment and a possible paper on high-stakes AI decision support. The system architecture matters more than the trading domain: the LLM is constrained by structured inputs, deterministic gates, confidence tiers, and an audit trail.
Structured truth layer
Market state, levels, valuation backdrop, technical signals, and catalyst risk are structured before the model writes anything.
Deterministic gates
The system can force a pass when a level, stop, reward/risk, staleness, or binary-risk condition fails.
Outcome logging
Every paper run updates decisions, paper trades, equity, drawdown, and anonymised public analytics.
Public analytics are sanitized from the TapeEdge paper-account report. Tickers, exact prices, level zones, and trade theses are withheld.