KNOWLEDGE CENTER
What Is Solius
An algorithmic trading model scanner and signal delivery platform.
Solius continuously evolves, validates, and ranks algorithmic trading models using a fully automated pipeline. Promoted models emit live BUY/SELL signals that subscribers receive in real time. The platform covers cryptocurrency and equity markets.
How It Works
From model discovery to live signal delivery in four stages.
The pipeline automatically generates and evaluates thousands of model candidates. Each candidate goes through multiple rounds of stress testing to assess whether it performs consistently across different market conditions. Survivors are promoted to live signal production where they continuously analyze market data. Models that drift in quality are automatically retired.
Credits & Data Sources
Platform credits and acknowledgements.
Solius was designed and built by Ariel. Market data, on-chain metrics, and AI capabilities are sourced from third-party providers.
Disclaimer
Solius provides informational tools, not financial advice.
Solius is an informational platform. Signals, scores, and metrics are algorithmic outputs based on historical and real-time data — they do not constitute financial advice, investment recommendations, or solicitations to trade. Past performance, whether backtested or live, does not guarantee future results. Always do your own research and consult a qualified financial advisor before making investment decisions.
Model Score
Ranks models by recent risk-adjusted outperformance versus buy-and-hold.
The score weighs recent performance more heavily than older results and factors in trading consistency. Higher score = better overall performance relative to the market. This is separate from the five percentile bars (profit, safety, accuracy, consistency, activity), which rank individual traits against all other models.
Percentile Stats
Five bars ranking a model's traits against all peers.
Each model is rated on five percentile dimensions: Profit (total return), Safety (drawdown control), Accuracy (win rate), Consistency (return stability), and Activity (trading frequency). A percentile of 80 means the model outperforms 80% of peers on that trait. These are independent rankings — a model can score high on profit but low on safety.
Model Archetypes
Personality labels classifying a model's trading style.
Each model is classified into one of six archetypes based on its behavior: Trend Rider (trend-following), Sharpshooter (high-accuracy), Fortress (low-risk), Machine Gun (high-frequency), Wildcard (high-return), and All-Rounder (balanced). Archetypes are derived from live metrics like win rate, drawdown, and trade frequency. The detail drawer also shows Market Strengths — conditions where the model historically performs best (e.g., trending, volatile).
Sharpe Ratio
Return earned per unit of risk — a widely used benchmark for comparing models.
The Sharpe ratio divides a model's excess return by its volatility. Above 1.0 is generally considered good, above 2.0 excellent. It lets you compare models on a level playing field: a 50% return with low volatility may be better than 100% with extreme swings.
Max Drawdown
The worst peak-to-trough equity drop — measures worst-case scenario.
Maximum drawdown is the largest percentage decline from a peak to a subsequent trough. A drawdown of -20% means at some point the portfolio lost 20% from its highest value. Lower (closer to 0%) is better. It's a key risk metric for understanding downside exposure.
Win Rate
Percentage of trades that closed in profit.
Win rate is the number of profitable trades divided by total trades, expressed as a percentage. A 60% win rate means 6 out of 10 trades were profitable. Win rate alone doesn't determine profitability — a 40% win rate can still be very profitable if winning trades are much larger than losing ones.
CAGR
Compound Annual Growth Rate — annualized return over the backtest period.
CAGR smooths out a model's total return into a consistent yearly rate. A model with 200% total return over 3 years has a CAGR of about 44%. It makes models with different time horizons directly comparable.
Alpha
Excess return above the buy-and-hold benchmark.
Alpha measures how much a model outperforms (or underperforms) simply buying and holding the asset. Positive alpha means the model has outperformed passive exposure over the measured period. For example, alpha of +15% means the model returned 15 percentage points more than buy-and-hold over the same period.
Profit Factor
Gross profits divided by gross losses — profitability at a glance.
Profit factor divides total profits from winning trades by total losses from losing trades. A profit factor of 2.0 means the model earned $2 for every $1 lost. Above 1.5 is generally considered favorable, above 2.0 is strong. Below 1.0 means the model lost money overall in the test period.
Net Return
Total percentage return over the entire backtest period.
Net return is the cumulative percentage gained or lost across all trades in the backtest period. It is not annualized — a model running for 6 months with +50% and one running 3 years with +50% have the same net return but very different CAGRs. Always consider net return alongside the backtest timeframe.
Total Trades
Number of round-trip trades executed during backtesting.
Total trades counts complete buy-then-sell (or sell-then-buy) cycles. More trades generally provide higher statistical confidence in the model's metrics. A model with 200 trades and a 60% win rate has higher statistical significance than one with 10 trades and the same win rate.
Track Record
Days the model has been tracked on real market data, not just backtested.
Track record measures how long a model has been running on real market data — these are verified, non-backtested days. A longer track record provides more confidence that the model performs in actual market conditions. Performance data is refreshed nightly, not in real time during the day.
Average Hold Time
Average duration a position is held open before closing.
Average hold time shows how long a model typically keeps a trade open — from entry to exit. A model holding positions for minutes is a scalper; hours to days is a swing trader; weeks or more is a position trader. This helps you understand the model's trading style and whether it matches your preferences.
Alerts
Custom price, signal, and indicator notifications.
Alerts let you set conditions on prices, model signals, or technical indicators. When a condition is met, you receive notifications via email, push, or Telegram. You can create, toggle, and delete alerts from the Alerts page or by asking the AI assistant.
Subscriptions
Follow a model to receive its live signals in real time.
Subscribing to a model means you'll receive notifications whenever it changes its trading signal. You choose which delivery channels to use (in-app, email, Discord, Telegram). Subscriptions can be enabled/disabled without deleting them.
Trading Signals
BUY/SELL/NEUTRAL outputs from verified models, refreshed nightly.
Each verified model analyzes market data and emits an algorithmic signal: BUY, SELL, or NEUTRAL based on its indicator readings. Signals and performance metrics are refreshed once per night — not in real time during the day. Signals are informational and do not constitute financial advice. You access them through subscriptions, the model browser, or the AI assistant.
Model Browser
Search, filter, sort, and explore all available models.
The model browser is the main page for discovering models. You can filter by symbol, sort by any metric, and toggle between a visual percentile view and a raw metrics view. Clicking a model opens its detail drawer with full performance data, charts, and subscription controls.
Backtesting
Simulating a model on historical data to estimate how it would have performed.
Backtesting applies a model's trading rules to historical price data to see what returns, drawdowns, and win rates it would have produced. Backtests are re-run nightly against the latest market data, so metrics stay current. However, past performance does not guarantee future results; that's why Solius applies additional stress testing before any model enters verified tracking.
Verified Trading
Models tracked on real market data after passing validation — performance refreshed nightly.
Verified trading means a model has passed the platform's validation criteria and is now being tracked on real market data. Performance metrics are refreshed once per night — they are not updated in real time during the trading day. If performance drifts too far from expected behavior, the model is automatically retired.
Benchmark Comparison
Comparing a model's performance against simply holding the asset.
The buy-and-hold benchmark represents the return you'd get by just buying the asset and holding it for the same period. If a model can't beat buy-and-hold, the active trading adds risk without reward. The comparison page overlays equity curves and key metrics side-by-side.
Robustness Testing
Stress-testing a model to make sure it isn't overfit to past data.
Robustness testing checks whether a model's edge is real or just a statistical fluke. The platform applies multiple stress tests to verify results hold up under varying market conditions. Only models that pass all checks are promoted to live signal production.