Built first
Reverse
What would have happened if you had done the opposite? Two portfolios side by side — yours and its mirror — compared on net result, equity curve, win rate, drawdown, Sharpe and losing streaks.
Decision support · Statistical analysis
See the trade you didn’t make.
ReverseEdge takes your real trading history and builds the portfolio that did the exact opposite — every buy became a sell, every sell became a buy. Then it shows you which one came out ahead.
No predictions. No signals. Just arithmetic on trades that already happened.
Reverse — compare with the mirror
Load the trade history exported from your broker as a CSV file. The tool matches your buys and sells into closed positions (first in, first out), builds the mirror portfolio, and compares the two.
Three modes
Next
Analyze
Where an instrument stands right now: trend, volatility, volume, RSI, MACD, moving averages, support and resistance, momentum, and its correlation with the index.
Later
Predict
Historical situations that looked like this one, and what followed them after 1, 3, 5, 10 and 20 days. Descriptive statistics about the past — never a forecast dressed up as certainty.
How the numbers are made
Market data → Indicators → Feature vectors → Similarity index → Statistics → plain-language explanation
- The engine computes, the language model only explains. Every figure on screen comes from a deterministic, reproducible, test-covered calculation. The AI layer writes the sentence around numbers that already exist. It never invents a probability, a percentage or a price.
- Every probability ships with its base rate and sample size. “64% up” on its own is misleading — markets rise on roughly 54% of days anyway. If a signal does not beat the base rate, this app says so instead of hiding it. Confidence intervals everywhere.
- Weights are fitted, not chosen. A score built from hand-picked weights is astrology with numbers. Until a model is calibrated against real historical outcomes, its score is not shown.
- Probability is not a guarantee. A 64% historical success rate means roughly one in three of those situations went the other way.
Project status
Version 0.2.0 — the Reverse calculation engine and its
API are in place and tested; the screen that uses them is still being
built. Analyze, Predict and Behavior are not started. Nothing here is a
finished product yet.