Auspex Desk

Gold engine · XAU/USD

Osprey Gold trading engine on Auspex Desk

Hunts breakouts from key price structures and lets winners run toward distant targets.

Track record

Simulated paper trading since 30 Sept 2026 · updated hourly

Net return

+5.0%

+$50 on $1,000

Win rate

59%

10 of 17 trades

Profit factor

1.99

gross win / gross loss

Closed trades

17

since 30 Sept 2026

Equity of the simulated $1,000 account

Simulated (paper) trades on live prices with modelled costs. Small samples carry wide uncertainty. Past performance does not guarantee future results.

How Osprey works

A probability-first engine: a setup must beat the odds a fair game would give it before it trades.

Osprey starts from a strict premise: treat price as a fair game. For any stop and target, a random walk already implies a base probability of reaching the target first, set purely by the geometry of the trade. A setup earns a position only when accumulated evidence says it beats that base rate by a clear margin.

The evidence lives in a Bayesian ledger that records, for each setup family and target distance, how often the target was reached before the stop. Exits are clean barriers, stop or target and nothing discretionary in between, so every resolved trade is an unambiguous data point and the probabilities sharpen with each one.

Two live arms carry Osprey's orders. F2 buys the breakout of a falling trendline on its own structural timeframe while the higher-timeframe trend is rising, with a structural stop, and gives the trade a few days. FF trades intraday: the session's first confirmed fractals, on an interval calibrated specifically for the arm, become the breakout levels.

Market
XAU/USD (spot gold)
Account
Paper, 10x leverage
Style
Probability-gated breakouts
Direction
F2 long only, FF long and short
Holding
FF intraday, F2 a few days
Entries
F2 scheduled market buy, FF first-fractal breakout
Exits
Structure-based protection, fixed target, time exit
Decision maker
Rules and Bayesian evidence, no AI

Fair-game baseline

Every trade is measured against the probability a random walk would give it, so geometry alone can never pass as skill.

Bayesian evidence

Outcomes accumulate in a ledger per setup family and distance; conviction grows only as fast as the evidence does.

Clean barrier exits

Stop and target are set together at entry. No discretionary exits means no ambiguity in what each trade proved.

Two live arms

A multi-day trendline-breakout buyer and an intraday first-fractal breakout trader place the live orders.

Decision pipeline

  1. 1

    Candidate setups

    Setup families propose trades with a defined stop, target and horizon, ranked by conservative expected value.

  2. 2

    Base rate

    The stop-to-target geometry sets the probability a fair game would assign the target: the bar to beat.

  3. 3

    Evidence gate

    The Bayesian ledger must clear that bar with margin, after costs, before a setup family is allowed to trade.

  4. 4

    Live arms

    F2 and FF place the live paper orders under their own structural rules, one position at a time.

  5. 5

    Barrier exits

    Stop and target are fixed at entry; beyond them only time and safety exits, never discretion.

The two live arms

F2 waits for gold's higher-timeframe trend to point up, then looks for a trendline drawn across falling highs on its own structural timeframe. When price breaks that line upward, F2 buys at its scheduled entry window with a stop at the structure that invalidates the idea, aims for a profit objective calibrated in-house and gives the trade a few days to work. FF watches the opening phase of each session on an interval calibrated for the arm. The first confirmed upper fractal becomes the buy level and the first confirmed lower fractal the sell level, and a cross of either level triggers the trade, with tight volatility-aware protection and a distant target, closed the same day.

Create a free account to see live positions, trade history and alerts

  • Osprey's open positions, the moment they open
  • Every closed trade, wins and losses
  • Telegram alerts when the engines you pick act

Free during early access · No card required · Not investment advice

All strategies

Not investment advice. The agents running here are a decision-support tool; this is not investment advice. Results are from simulated (paper) trading. Signals are statistical assessments based on historical data and do not guarantee future performance. Leveraged trading can result in the loss of your entire capital.