Software
What the software is, and what stays private.
The signals come from machine-learning software developed by our broker partners. Its internals are covered by an NDA. This page explains what that means for you and what you can check instead.
The software produces signals. It does not trade anyone's account and it never touches client money: you place your own trades in your own Vantage account.
01 · Boundaries
What we keep private, and what we show instead.
Signal software is the core asset of the people who build it, and ours is covered by a non-disclosure agreement. Instead of the code we publish the record it produced, the rules around it and the maths it is built on.
What we show
- ✓ The full signal log, every win and every loss
- ✓ The rules applied to every signal: stop, risk, management
- ✓ What the common analysis metrics mean and why a trader cares
- ✓ Calculators and stress tests that run on the real record
What we keep private
- – Source code, models and training data
- – Parameters, entry formulas and thresholds
- – Anything covered by the non-disclosure agreement
02 · Metrics
What analysis software of this kind measures
Generic definitions of common quantitative metrics. They explain the ideas; they do not reveal how our software uses them.
- β (Kalman)
- Adaptive hedge ratio between gold and silver, re-estimated as new prices arrive. Why it matters: a stale ratio makes the spread look wrong when it is not.
- Z-score
- How far the spread is from its mean, in standard deviations. Why it matters: a large value says 'unusual', not 'trade now'.
- Half-life
- Typical time for the spread to close half the gap to its mean. Why it matters: it sets how long an idea may need.
- θ, κ, σ
- Long-run mean, speed of mean reversion and volatility of the spread (Ornstein–Uhlenbeck parameters).
- ADF p-value
- Statistical test of stationarity. Low values support the idea that the spread returns to its mean.
- ε_t formula
- How the spread is computed from the two prices and the hedge ratio.
- Spread chart, ±2σ
- The spread over time with dashed bands at two standard deviations.
- Signal state
- The model's current state. It is not an order: you decide whether to take a trade.
- Itô's lemma
- Rule for how a function of a random price evolves. Background maths behind volatility models.
- SKEW
- Lopsidedness of the return distribution.
- EX-KURT
- Fat tails: how often extreme moves happen compared with a normal distribution.
- HURST
- Below 0.5 mean-reverting, above 0.5 trending, 0.5 random walk.
- ACF(1)
- Correlation between consecutive values: reversal or momentum from one step to the next.
Try the ideas yourself in the Math lab.
03 · Test us
How you can test us without trusting us
- 01Download the trade log and recompute every statistic in your browser on the Method page
- 02Open a demo account and follow the signals without risking money
- 03Ask us on Telegram about any signal in the log
Recompute everything on the Method page.