AlphaGlyph runs quant and machine-learning strategies on real historical prices with simulated capital, then does what most backtests don't: it checks whether the result is genuine skill or just luck, using the same statistical tests quant funds use. Six strategies, a no-code rule builder, an ML transformer — and an honest verdict on every run.
No code · No account · No real money · Every trade explained in plain English
The whole flow is one page — the Backtester — with everything explained, not hidden behind a number.
Choose from six strategies (or build your own buy/sell rules from indicators), pick stocks and a date range, and simulate it day-by-day on real prices with transaction costs.
The equity curve, each trade with a plain-English reason (what it saw, the regime, why it acted), and an optional animated replay of the whole run.
Monte Carlo, Deflated Sharpe, and Fama-French synthesise into one verdict: Statistically Significant, Promising, or Inconclusive.
Any strategy can look good on historical data by chance. AlphaGlyph runs three independent statistical tests and synthesises them into a single verdict.
1,000 bootstrap paths build a distribution — your result either stands out or it doesn't.
The Deflated Sharpe Ratio penalises testing many strategies, the way good science corrects for multiple comparisons.
Factor decomposition strips out market, size, and value exposure to reveal whether any genuine alpha remains.
# Probability the Sharpe is real after # testing N strategies on noisy data def deflated_sharpe_ratio(returns, n_strategies): sr = annualized_sharpe(returns) skew = scipy.stats.skew(returns) kurt = scipy.stats.kurtosis(returns) # expected max Sharpe from pure luck sr_star = expected_max_sharpe(n_strategies) return psr(sr, sr_star, skew, kurt) # → 0.964 ✓ significant
Every indicator, model, and statistical method is implemented from scratch — no black-box TA libraries.
MA Crossover, RSI, MACD, a patient Dip Buyer, an ML transformer — and a no-code builder to define your own buy/sell rules from indicators.
A multi-modal transformer trained with a chronological 60/20/20 split; backtests run strictly out-of-sample. The win is the rigour, not a magic edge.
ADX, Bollinger Band Width, and realized volatility classify the market into four regimes; the adaptive strategy switches to match.
Wide trailing stops let winners run; the Kelly Criterion sizes positions by the historical edge, not a fixed percentage. No look-ahead — and it's tested.
Rank every strategy on the same data, and compute the Markowitz efficient frontier (max-Sharpe and min-variance portfolios) in the Tools tab.
Monte Carlo, Deflated Sharpe (López de Prado), and Fama-French attribution tell you whether an edge is genuine — or just luck.
Pick a strategy and a few stocks, run it on real history in seconds, and see whether the edge holds up — no account required.