CORE SYSTEMS

How Paper Trading Helps Companies Test Algorithmic Strategies Without Risk¶

Simulation, metrics and best practices for enterprise algorithmic trading

Published: February 12, 2026 Author: CORE SYSTEMS Reading time: 14–18 minutes Category: FinTech, Algorithmic Trading

Algorithmic trading today accounts for more than 70% of volume on major exchanges. Yet many companies deploy strategies into production without thorough verification in a simulated environment. Paper trading — trading on “paper” with fictitious capital — is a critical step that separates successful quant teams from those learning through live losses.

73%

algorithms fail in the first year without paper testing

4–8 weeks

minimum paper trading phase for enterprise strategies

3×

higher survival rate of strategies after thorough simulation

$0

capital risk during entire testing

What is Paper Trading¶

Paper trading is real-time trading simulation with fictitious capital. Unlike backtesting, where strategies are verified on historical data, paper trading runs on live markets — orders are generated but never actually executed on the exchange. The result is an accurate picture of how the strategy would perform, without any financial risk.

Difference Between Backtesting and Paper Trading¶

Many teams confuse these two concepts, even though they serve different purposes:

• Backtesting — testing on historical data. Fast, repeatable, but prone to overfitting and ignores real conditions (latency, slippage, liquidity).
• Paper trading — simulation on live data in real time. Slower, but captures actual market conditions, execution delays and data anomalies.
• Forward testing (out-of-sample) — paper trading on previously unseen data. Gold standard for validation before deploying live capital.

Why Backtesting Alone Isn’t Enough?¶

Backtesting suffers from look-ahead bias (strategy “sees” the future), survivorship bias (only tests companies that survived) and ignores market impact of own orders. Paper trading eliminates these problems because it runs in real conditions — just without actual capital.

Why Paper Trading is Key for Enterprise¶

For companies — whether hedge funds, prop trading firms, banks or insurance companies managing investment portfolios — paper trading is much more than just a “sandbox phase”. It’s an integral part of the algorithmic strategy lifecycle.

1. Eliminating Financial Risk in Early Phases¶

Every new strategy carries inherent uncertainty. Paper trading allows identifying logic errors, edge cases and unexpected market regimes before they can cause real losses. For companies trading tens of millions daily, a single untested bug can mean losses in the millions of crowns.

2. Regulatory Compliance¶

Regulators like ESMA, CNB and SEC require documentation of testing processes for algorithmic systems. MiFID II (Article 17) explicitly requires firms operating algorithmic trading to have established systems for testing algorithms before deployment. Paper trading provides auditable evidence of due diligence.

3. Infrastructure Validation¶

Paper trading doesn’t just test the strategy — it tests the entire stack: data feeds, order management system, risk management modules, latency, failover procedures. It’s a “dress rehearsal” for production.

4. Building Stakeholder Confidence¶

Investors, risk committees and board members require evidence. Paper trading generates a track record that is simulated but based on real market conditions — much more convincing than backtesting results.

Key Metrics: What to Monitor During Paper Trading¶

Raw profit isn’t enough. Enterprise quant teams monitor an entire battery of metrics that together paint a complex picture of strategy quality.

Win Rate (Trade Success Rate)¶

Win rate expresses the percentage of profitable trades out of total trades. Seemingly simple metric, but misleading on its own.

Win Rate = (number of profitable trades / total number of trades) × 100 Example: 430 total trades, 258 profitable Win Rate = 258 / 430 × 100 = 60%

Interpretation: A 60% win rate sounds great, but if the average loss significantly exceeds the average gain, the strategy can be overall unprofitable. Therefore, win rate must always be evaluated in combination with profit factor and risk-reward ratio.

• Trend-following strategies: Typically 35–45% win rate, but with high profit-to-loss ratio.
• Mean-reversion strategies: Often 60–75% win rate, with smaller profits per trade.
• Market-making: 50–55% win rate, profit from bid-ask spread.

Profit Factor¶

Profit factor is the ratio of gross profits to gross losses. It’s one of the most important metrics because it combines both frequency and size of profits and losses.

Profit Factor = Σ profits / |Σ losses| Example: Gross profits: 2,450,000 CZK Gross losses: 1,680,000 CZK Profit Factor = 2,450,000 / 1,680,000 = 1.46

Profit Factor	Rating	Action
< 1.0	Losing strategy	Stop, reevaluate logic
1.0 – 1.2	Marginal	Optimize, consider costs
1.2 – 1.5	Good	Continue testing
1.5 – 2.0	Very good	Candidate for live deployment
> 2.0	Excellent (verify overfit)	Check if too good to be true

⚠️ Beware of “too good” results. Profit factor above 3.0 in paper trading is a warning signal. Either the strategy only works in a specific market regime, or there’s an error in simulation (e.g., unrealistic fill prices).

Maximum Drawdown¶

Maximum drawdown measures the largest decline in portfolio value from peak to trough. It’s a key metric for risk management — it tells you “how much it can hurt” in the worst-case scenario.

Max Drawdown = (Trough Value − Peak Value) / Peak Value × 100 Example: Portfolio reached maximum of 10,000,000 CZK Then fell to minimum of 8,200,000 CZK Max Drawdown = (8,200,000 − 10,000,000) / 10,000,000 = −18%

Why is drawdown more important than total profit? A strategy that earns 50% annually but has max drawdown of −60% is unacceptable for most enterprise clients. Investors and risk management teams typically have hard limits:

• Conservative funds: Max drawdown −10% to −15%
• More aggressive strategies: Max drawdown −20% to −25%
• High-frequency: Max drawdown under −5%

During paper trading, it’s critical to monitor not only maximum drawdown but also recovery time — how long it takes for the portfolio to return to previous maximum.

Sharpe Ratio¶

Sharpe ratio is probably the most used risk-adjusted return metric. It measures how much return above the risk-free rate the strategy generates per unit of risk (volatility).

Sharpe Ratio = (Rp − Rf) / σp where: Rp = average portfolio return (annualized) Rf = risk-free rate (e.g., 3.5% for CZK in 2026) σp = standard deviation of returns (annualized) Example: Annualized return: 18% Risk-free rate: 3.5% Volatility: 12% Sharpe = (0.18 − 0.035) / 0.12 = 1.21

Sharpe Ratio	Rating
< 0.5	Weak — risk inadequately compensated
0.5 – 1.0	Average — comparable to passive strategy
1.0 – 2.0	Good — attractive to most investors
2.0 – 3.0	Excellent — top-tier hedge fund level
> 3.0	Exceptional (verify realism)

Other Metrics Worth Monitoring¶

• Sortino Ratio: Sharpe variant that penalizes only downside volatility — better for strategies with asymmetric returns.
• Calmar Ratio: Annualized return divided by maximum drawdown. Ideal for comparing strategies with different risk profiles.
• Average Trade Duration: How long the average trade lasts — important for assessing capacity and turnover.
• Slippage and Commission Impact: Difference between simulated and realistic execution price. Paper trading should model realistic slippage.
• Exposure Time: What percentage of time the strategy is in the market — low exposure with high Sharpe is ideal.

Practical Examples from Enterprise Environment¶

Example 1: Prop Trading Firm — Momentum Strategy on Stocks¶

Situation¶

A Prague prop trading firm developed a momentum strategy trading the 50 most liquid European stocks. Backtesting on 5 years of data showed Sharpe ratio 2.1 and profit factor 1.8.

Paper Trading (8 weeks)¶

Real-time results revealed three key problems:

• Slippage: Real slippage was 2.3× higher than in backtest due to insufficient liquidity on smaller titles.
• Data feed latency: 150ms delay in data caused entries at worse prices for fast momentum signals.
• Market regime shift: In low-volatility environment, strategy generated too many false signals.

Result After Adjustments¶

After correcting volatility filter and limiting to 30 most liquid titles: Sharpe 1.4, profit factor 1.5, max drawdown −12%. Strategy deployed with 20M CZK capital.

Example 2: Asset Management — Mean-Reversion on FX Pairs¶

Situation¶

Investment company managing corporate client portfolios tested mean-reversion strategy on major FX pairs (EUR/USD, GBP/USD, USD/JPY) for hedging exposure.

Paper Trading (12 weeks)¶

Key findings:

• Win rate 68% (consistent with backtest), but average profit per trade was 15% lower due to more realistic fill prices.
• Max drawdown −8.3% — within the −10% limit set by risk committee.
• Sharpe ratio 1.7 — better than benchmark (passive hedging with forwards, Sharpe 0.3).
• Strategy selectively failed around central bank interest rate announcements.

Result¶

Added filter that suspends trading 2 hours around key macroeconomic events. Final Sharpe 1.85, deployed on hedging portfolio worth 500M CZK.

Example 3: Bank — Algorithmic Execution for Client Orders¶

Situation¶

Czech bank implemented TWAP/VWAP execution algorithms for processing large client orders on Prague Stock Exchange. Goal: minimize market impact for trades above 50M CZK.

Paper Trading (6 weeks)¶

Metrics monitored specifically for execution algorithms:

• Implementation shortfall: Average 4.2 bps vs. benchmark 6.8 bps — significant improvement.
• VWAP deviation: ±1.1 bps from daily VWAP — within tolerance.
• Fill rate: 97.3% of orders completely filled within day.
• Stability: Consistent results even on high volatility days.

Result¶

After successful paper trading, live algorithm deployed. Annual savings for bank clients estimated at 12M CZK on reduced execution costs.

Best Practices for Enterprise Paper Trading¶

1. Realistic Execution Modeling¶

Paper trading is only as good as the simulation is realistic. Key aspects the simulation must capture:

• Slippage model: Use historical spread and liquidity data to model realistic slippage. Minimum 1–3 bps for liquid instruments.
• Partial fills: Not every order fills completely. Simulate partial fills based on available volume.
• Latency: Simulate real latency of your production stack (data feed + order routing + exchange).
• Fees: Include all transaction costs — brokerage, exchange fees, clearing.

2. Define Go/No-Go Criteria Upfront¶

Before starting paper trading, establish clear “pass/fail” criteria:

Example go/no-go criteria¶

MINIMUM_CRITERIA = { “sharpe_ratio”: 1.0, # Minimum Sharpe “profit_factor”: 1.3, # Minimum PF “max_drawdown”: -0.15, # Max -15% “win_rate”: 0.40, # Min 40% “min_trades”: 200, # Statistical significance “recovery_time_days”: 30, # Max recovery time “paper_trading_weeks”: 6 # Minimum testing period } def evaluate_strategy(results): for metric, threshold in MINIMUM_CRITERIA.items(): if metric == “max_drawdown”: if results[metric] < threshold: return “FAIL”, f”{metric}: {results[metric]} < {threshold}” else: if results[metric] < threshold: return “FAIL”, f”{metric}: {results[metric]} < {threshold}” return “PASS”, “All criteria met”

3. Test Across Market Regimes¶

Paper trading should cover different market conditions. If possible, extend testing period to include:

• Low and high volatility periods
• Trending and sideways markets
• Macroeconomic events (central bank meetings, earnings season)
• Exceptional situations (flash crash simulation, liquidity gaps)

4. Automated Monitoring and Alerting¶

Set up real-time dashboards and alerts for key metrics. Paper trading isn’t “set and forget” — it requires active monitoring:

• Daily P&L report with comparison against backtest
• Alert when drawdown limit exceeded
• Data quality monitoring (gaps, stale data, anomalies)
• Weekly review with quant and risk team

5. Documentation and Audit Trail¶

Every trade, parameter and decision must be logged. For regulatory compliance and internal knowledge management:

• Strategy version (git hash) for each paper trade
• Market context at time of trade
• Reasons for all parameter adjustments
• Sign-off from risk management before going live

Paper Trading System Architecture¶

For enterprise deployment, a simple spreadsheet isn’t enough. Robust paper trading infrastructure includes:

┌─────────────────────────────────────────────────┐ │ MARKET DATA FEED │ │ (real-time: Bloomberg/Refinitiv/Exchange) │ └─────────────────┬───────────────────────────────┘ │ ┌─────────────────▼───────────────────────────────┐ │ STRATEGY ENGINE │ │ (signal generation, position sizing) │ └─────────────────┬───────────────────────────────┘ │ ┌──────────┴──────────┐ │ │ ┌──────▼──────┐ ┌────────▼────────┐ │ PAPER OMS │ │ LIVE OMS │ │ (simulated) │ │ (production) │ └──────┬──────┘ └────────┬────────┘ │ │ ┌──────▼──────┐ ┌────────▼────────┐ │ FILL SIM │ │ EXCHANGE │ │ (slippage, │ │ CONNECTION │ │ latency) │ │ │ └──────┬──────┘ └─────────────────┘ │ ┌──────▼──────────────────────────────────────────┐ │ ANALYTICS & REPORTING │ │ (P&L, Sharpe, drawdown, risk metrics, alerts) │ └─────────────────────────────────────────────────┘

The key design principle is that switching from paper to live should be changing one configuration parameter — the entire rest of the stack remains identical. This ensures paper trading results are maximally representative.

Common Mistakes and How to Avoid Them¶

1. Too short testing period: 2 weeks of paper trading has no statistical weight. Minimum 200+ trades or 6+ weeks.
2. Ignoring transaction costs: Strategy profitable without fees may be unprofitable after including real costs.
3. Optimization during paper trading: Changing parameters mid-test invalidates results. Every change = new test from beginning.
4. Survivorship bias in paper portfolio: Don’t forget to include delisted instruments and corporate actions.
5. Missing stress testing: Paper trading in calm market doesn’t reveal how strategy behaves during volatility spike.
6. Unrealistic fill assumptions: Assuming orders always fill at mid-price is a dangerous illusion.

From Paper Trading to Production: Transition Playbook¶

After successful paper trading, transition to live trading happens gradually:

1. Phase 1 — Micro-live (1–2 weeks): Trade with 5–10% of planned capital. Verify live execution matches paper results.
2. Phase 2 — Scale-up (2–4 weeks): Gradually increase to 25%, then 50%, while continuously monitoring metrics.
3. Phase 3 — Full deployment: 100% capital, but with automatic kill-switches and drawdown limits.
4. Phase 4 — Continuous monitoring: Daily comparison live vs. paper (parallel run) to detect degradation.

Kill-Switch Rules (Example)¶

Automatic strategy shutdown when:

• Daily loss > 2% of capital
• Drawdown > 50% of maximum allowed drawdown
• Sharpe ratio over last 30 days drops below 0.5
• Deviation from paper results > 2 standard deviations

Conclusion¶

Paper trading isn’t an optional step — it’s insurance that separates professional algorithmic trading from gambling. For enterprise firms, it represents an investment with the highest possible ROI: for the cost of several weeks of simulation, you gain confidence in the strategy, meet regulatory requirements and significantly reduce the probability of catastrophic loss.

Key takeaways:

• Always test on live data — backtesting isn’t enough.
• Monitor metrics holistically — win rate, profit factor, max drawdown and Sharpe ratio together, never in isolation.
• Define criteria upfront — deployment decisions must be data-driven, not emotional.
• Model realistically — slippage, latency and fees determine whether strategy survives transition to live.
• Document everything — regulators and future you will thank you.

If your company is developing or planning to deploy algorithmic trading strategies and needs help with testing infrastructure architecture, CORE SYSTEMS has experience designing and implementing enterprise-grade paper trading systems.

Contact CORE SYSTEMS¶

Need help with algorithmic trading system architecture, paper trading infrastructure or quantitative analytics? Our experts are ready to help.

Contact us

© 2026 CORE SYSTEMS. All rights reserved.

Expert IT services | FinTech solutions | Algorithmic trading