BUY vs SELL Signal Accuracy
We analyzed every BUY and SELL signal SultraxAI has logged. Here's how they performed at 1h, 4h, and 24h — broken down by direction.
The headline numbers
Across 2,365 BUY/SELL signals logged on SultraxAI, the platform-wide 1-hour win rate sits at 52.7%. At 4 hours it's 51.5%; at 24 hours 53.3%. Higher win rates at longer horizons reflect both the model and broader market drift effects.
Win rate by signal direction
BUY signals: 35.4% win rate across 1,089 signals. SELL signals: 68.5% win rate across 1,224 signals. The differential is notable — SELL signals consistently outperform BUY signals at the 1-hour horizon, suggesting the model identifies downward momentum more cleanly than upward momentum on the symbols we cover.
Average return per signal
1-hour average return: 0.02%. 4-hour: 0.11%. 24-hour: 0.09%. Note: these are simple averages including losses. Combined with the win rates above, expectancy can be computed directly.
What this means in practice
A 52.7% win rate at 1h with average return per signal of 0.02% gives positive expectancy when paired with disciplined risk/reward management (1:1 or better targets). The data does NOT suggest 'follow every signal blindly' — the win rate is modest, and individual signals require contextual judgment. Use the published numbers as a baseline, not a guarantee.
Methodology
Every BUY/SELL signal is timestamped at generation with the underlying price and indicator values. Outcomes are recorded at 1h, 4h, and 24h after the signal. A 'win' for a BUY = price higher at horizon. A 'win' for a SELL = price lower. No exclusion of signals; all outcomes count. See methodology page for full details.
Limitations and caveats
- Slippage and execution costs are not modeled. Real trader P&L will be lower than signal-price outcomes.
- Sample period covers signals from platform inception. Win rates may differ in different market regimes.
- Universe is curated (~100 US stocks + 12 cryptos). Less-liquid names are not covered.
- Win rate alone is not a profitability metric — expectancy depends on win/loss ratio per trade.
Cite this research
If you use these findings in your own writing, please cite us. Here's the citation format:
About this data
All data in this research comes from the live SultraxAI signal log. Methodology is documented at /about/methodology/. Raw aggregate statistics are publicly viewable on the platform — feel free to reproduce or extend our analysis.