How Uptick Works

Last Updated: January 24, 2026

---

Overview

Uptick provides algorithmic stock recommendations and portfolio rebalancing signals through a proprietary analysis platform. This page explains how our algorithm works, what data we analyze, and how recommendations are generated.

Important: Uptick's recommendations are algorithmic, non-personalized, and for informational purposes only. They do NOT constitute investment advice, and Uptick is NOT a registered investment adviser. See our Disclaimers section and our Investment Adviser FAQ for important information.

---

1. Core Algorithm Overview

What Does Uptick's Algorithm Do?

Uptick's algorithm analyzes publicly available market and financial data to identify potentially attractive stock opportunities and suggest portfolio rebalancing strategies. The algorithm:

• Analyzes Data: Processes market data, corporate financial data, sentiment data, and macroeconomic data
• Identifies Patterns: Looks for statistical patterns in historical data
• Scores Stocks: Assigns numerical scores to securities based on multiple factors
• Generates Recommendations: Recommends actions (buy, hold, sell/avoid)
• Suggests Rebalancing: Proposes portfolio allocation adjustments
• Updates Continuously: Updates recommendations as new data becomes available

What Is Our Optimization Objective?

Uptick's algorithm is designed to maximize the Sharpe ratio—a measure of risk-adjusted returns that balances return potential against volatility.

Sharpe Ratio Definition:

• Ratio of expected return to volatility
• Higher ratio = better risk-adjusted returns
• Accounts for both return and risk taken
• Helps compare different portfolio strategies

Our Approach:

• We optimize for maximum Sharpe ratio based on historical analysis
• Goal is to generate attractive returns while minimizing volatility
• We evaluate this using 5 years of historical data
• We assume this optimization will continue to produce good results (see limitations below)

---

2. Data Sources and Inputs

Market Data

OHLCV Data (Open, High, Low, Close, Volume)

• What it is: Daily stock price and volume information
• Source: Exchange data, market data providers
• Update frequency: Near real-time throughout trading day
• Use in algorithm: Price trends, technical analysis, momentum assessment
• Coverage: All publicly traded U.S. stocks and major international stocks

Technical Indicators Derived from Price Data:

• Moving averages (short-term and long-term price trends)
• Momentum indicators (rate of price change)
• Volatility measures (price fluctuations)
• Volume trends (trading activity changes)

Corporate Financial Data

10-K Annual Reports

• What it is: Comprehensive annual financial statements filed with SEC
• Filed: Within 90 days of fiscal year-end
• Contents: Financial statements, management discussion, risk factors, detailed disclosures
• Use in algorithm: Financial health assessment, growth trends, profitability

10-Q Quarterly Reports

• What it is: Interim quarterly financial statements
• Filed: Within 45 days of quarter-end
• Contents: Quarterly financial statements, updates to annual information
• Use in algorithm: Tracking changes in financial performance, quarterly trends

8-K Current Reports

• What it is: Reports of significant corporate events
• Filed: Within 4 business days of event
• Events: Mergers, executive changes, major contracts, bankruptcy, etc.
• Use in algorithm: Incorporating significant corporate developments

Financial Data Extracted:

• Revenue and earnings
• Profitability metrics (gross margin, operating margin, net margin)
• Growth rates (revenue growth, earnings growth)
• Debt levels and financial ratios
• Return metrics (ROE, ROA, ROIC)
• Cash flow generation
• Capital structure

News and Information Data

Financial News and Announcements

• Sources: Financial news wires, company press releases, SEC filings
• Content: Earnings announcements, contract wins, management changes, acquisitions
• Update frequency: Throughout the day as news is released
• Use in algorithm: Incorporating significant corporate developments and events

Earnings Announcements

• What it is: Quarterly and annual earnings releases
• Timing: Typically within 2 months after quarter/year-end
• Contents: Financial results, forward guidance, management commentary
• Use in algorithm: Assessing actual results vs. expectations, tracking performance trends

Sentiment Data

Social Media and Community Sentiment

• Sources: Social media platforms, financial forums, investor discussions
• Content: Investor sentiment, community opinions, peer discussions
• Update frequency: Continuous as users post
• Use in algorithm: Assessing investor sentiment and perception changes

Market Sentiment Indicators

• Analyst sentiment (bullish/bearish ratings)
• News sentiment (positive/negative tone of news coverage)
• Social media sentiment (aggregated social media tone)
• Options market sentiment (implied by option prices)

Macroeconomic Data

Federal Economic Data

• Source: U.S. Federal Reserve, government agencies
• Key indicators:
  • Interest rates (Fed funds rate, Treasury yields)
  • Employment data (unemployment rate, job creation)
  • Inflation data (CPI, producer prices)
  • GDP growth and economic activity
  • Money supply and credit data
  • Housing starts and construction data

Broader Economic Indicators

• Gross Domestic Product (GDP) growth
• Inflation rates and trends
• Unemployment rates and trends
• Consumer confidence
• Business sentiment
• Interest rate levels and trends

International Economic Data

• Global GDP growth
• International interest rates
• Currency exchange rates
• International trade data
• Geopolitical risk indicators

Data Integration and Processing

How Data Flows Into Algorithm:

1. Data is collected from multiple sources
2. Data is cleaned and validated
3. Missing values are handled appropriately
4. Data is normalized for analysis
5. Technical indicators and features are calculated
6. Algorithm processes all integrated data
7. Recommendations are generated based on total analysis

---

3. Algorithm Methodology

General Approach

Uptick uses quantitative and statistical analysis to identify patterns in market and corporate data. While specific details are proprietary, the general approach includes:

Step 1: Data Analysis and Feature Engineering

What Happens:

• Historical data is analyzed for patterns and trends
• Technical indicators are calculated from price data
• Financial ratios are calculated from corporate data
• Statistical relationships are identified
• Features are engineered from raw data

Example Features:

• Recent price momentum (has the stock been rising or falling?)
• Earnings growth trend (is the company growing faster?)
• Valuation metrics (is the stock cheap or expensive?)
• Financial health indicators (is the company financially sound?)
• Sentiment indicators (what do investors think?)
• Market conditions (is the overall market favorable?)

Step 2: Pattern Recognition

What Happens:

• Algorithm looks for statistical patterns in historical data
• Identifies which stocks have historically performed well
• Identifies which combinations of factors predict good returns
• Tests different analytical approaches
• Ranks factors by predictive power

Examples of Patterns Tested:

• Do cheap stocks outperform expensive stocks?
• Do stocks with high earnings growth perform better?
• Do stocks with strong sentiment perform better?
• Do certain combinations of factors predict outperformance?
• How do market conditions affect relative performance?

Step 3: Model Optimization

What Happens:

• Multiple models are tested on historical data
• Models are optimized to maximize Sharpe ratio
• Weights are assigned to different factors
• Combinations of factors are tested
• Best-performing model is selected

Backtesting Process:

• Historical recommendations are generated
• Performance is measured
• Risk-adjusted returns (Sharpe ratio) are calculated
• Model is refined to improve results
• Process is repeated until optimal performance achieved

Step 4: Recommendation Generation

What Happens:

• Algorithm processes current data for each stock
• Calculates scores based on the validated model
• Generates standardized recommendations
• Updates recommendations as new data arrives
• Produces consistent output for identical inputs

Types of Recommendations:

• Stock-level: Individual stock scores and recommendations
• Portfolio-level: Suggested asset allocation
• Rebalancing signals: When to adjust portfolio composition
• Priority ranking: Which stocks to prioritize

Step 5: Continuous Updates

What Happens:

• Algorithm continuously monitors for new data
• Updates recommendations when material information changes
• Adjusts as market conditions evolve
• Processes earnings announcements and news
• Incorporates latest economic data

---

3.6 Algorithmic Transparency and Decision Making

How to Understand Our Recommendations

One of our key principles is transparency about how our algorithm works. While we keep certain proprietary details confidential (to protect our intellectual property), we disclose the factors, methodology, and reasoning behind our recommendations.

What Factors Does the Algorithm Consider?

Our algorithm evaluates four main categories of factors:

1. Technical Factors (Price and Volume Data)

• Recent price momentum (direction and strength of price trends)
• Price volatility (how much prices fluctuate)
• Trading volume trends (changes in trading activity)
• Relative strength compared to overall market

2. Fundamental Factors (Company Financial Health)

• Earnings growth rates and trends
• Profitability metrics (gross margin, operating margin, net margin)
• Valuation relative to earnings and book value
• Return on invested capital (ROIC)
• Cash flow generation and quality

3. Sentiment Factors (Market Opinion)

• Analyst recommendations (bullish vs. bearish sentiment)
• News sentiment (positive vs. negative tone of coverage)
• Social media sentiment (aggregated investor discussion)
• Institutional positioning and ownership trends

4. Macroeconomic Factors (Overall Market Conditions)

• Interest rate environment (Fed rates, Treasury yields)
• Economic growth outlook (GDP growth, earnings growth expectations)
• Inflation trends (CPI, PPI, inflation expectations)
• Sector performance relative to overall economy

How Are These Factors Combined?

The algorithm assigns weights (importance scores) to each factor based on historical analysis of which factors have been most predictive of future performance. These weights are:

• Regularly reviewed and validated against new data
• Tested against different time periods and market conditions
• Adjusted if historical performance patterns change significantly
• Applied consistently across all users and stocks

Key Point: All users receive the same algorithmic output for identical inputs. The algorithm does not personalize weights or factors based on individual user circumstances.

Why Aren't Specific Weights Disclosed?

The specific numerical weights assigned to each factor are proprietary intellectual property for several reasons:

Competitive Protection:

• Disclosing exact weights would allow competitors to copy our methodology precisely
• Competitors could develop strategies specifically designed to exploit our algorithm
• Third parties could front-run our recommendations

Industry Standard:

• Other algorithmic advisory firms (Betterment, Wealthfront, Robo Global, etc.) use the same approach
• Robo-advisors disclose their factors and methodology but not specific weights
• Asset management firms disclose their general approach but not exact portfolio construction rules
• This is recognized industry-standard practice for protecting competitive advantage

Regulatory Acceptance:

• SEC guidance on Form ADV explicitly allows for proprietary weighting to remain confidential
• Regulators understand the need for intellectual property protection
• Standard practice is to disclose methodology and factors, not proprietary weights

What You CAN Know About Our Algorithm

Complete transparency on:

• All factors we consider - Listed above with explanations
• General categories of analysis - Technical, fundamental, sentiment, macroeconomic
• Our optimization objective - Maximize Sharpe ratio (risk-adjusted returns)
• How recommendations change - As new data arrives, recommendations update
• Our backtesting results - See detailed methodology in Section 5.4
• Key limitations - See Section 6 for comprehensive limitations
• How long data takes to propagate - See Section 5 for update frequencies

What You CANNOT Know About Our Algorithm

For intellectual property and competitive reasons:

• Exact numerical weights of each factor
• Specific mathematical formulas and calculations
• Proprietary algorithmic architecture details
• Training process and optimization specifics
• Exact decision thresholds for buy/hold/sell
• Proprietary indicators or derived metrics
• Specific improvements we've made to past versions

This balance is the global standard: Transparency about factors and methodology, while protecting the specific implementation that represents our competitive advantage and intellectual property.

How the Algorithm Generates Recommendations

The multi-step process:

Step 1: Factor Calculation

• All factors are calculated for each stock
• Data is normalized (converted to comparable scales)
• Indicators are validated for quality and accuracy

Step 2: Weighting

• Each factor is multiplied by its assigned weight
• Weighted factors are combined into a total score
• Scores reflect importance of each factor in final recommendation

Step 3: Ranking

• All stocks are ranked by total score
• Highest scores represent best opportunities
• Lowest scores represent least attractive opportunities

Step 4: Categorization

• Top-ranked stocks receive "Buy" or "Strong Buy" recommendation
• Middle-ranked stocks receive "Hold" recommendation
• Lowest-ranked stocks receive "Sell" or "Avoid" recommendation

Step 5: Explanation

• For significant recommendation changes, we generate explanations
• Explanations indicate which factors changed most significantly
• Help users understand why a stock moved from Hold to Buy, for example

Why Different Stocks Are Recommended for Different Reasons

Different stocks may achieve high scores for different reasons:

Growth Stock Example:

• High score driven by: Earnings growth acceleration + Revenue growth + Technical momentum
• Recommendation: Buy growth opportunity
• Suitable for: Growth-oriented investors

Value Stock Example:

• High score driven by: Low valuation multiple + Strong fundamentals + Dividend yield
• Recommendation: Buy undervalued opportunity
• Suitable for: Value-oriented investors

Momentum Stock Example:

• High score driven by: Strong technical trend + Positive sentiment + Recent outperformance
• Recommendation: Buy trending opportunity
• Suitable for: Trend-following investors

Income Stock Example:

• High score driven by: High dividend yield + Sustainable dividend + Financial stability
• Recommendation: Buy income opportunity
• Suitable for: Income-focused investors

In Your Portfolio: Our recommendations may reflect a mix of these styles. You'll receive a diversified portfolio of recommendations, not concentrated in a single factor or style.

Algorithmic Limitations and What the Algorithm Does NOT Do

The algorithm explicitly DOES NOT:

• Have human judgment or decision-making input (purely algorithmic)
• Consider your personal financial situation
• Consider your investment objectives or goals
• Make personalized recommendations based on your circumstances
• Determine suitability of recommendations for you
• Assess your risk tolerance or capacity
• Apply different logic to different users
• Explain each individual recommendation in detail
• Adjust for individual preferences or values
• Consider non-financial factors (ESG, ethics, personal values)
• Predict black swan events or extreme market disruptions
• Guarantee accuracy of recommendations or their performance

You are responsible for assessing whether recommendations are suitable for YOUR situation.

Transparency About Data Quality and Assumptions

All analysis depends on data quality. We:

• Use high-quality third-party data sources
• Validate data for accuracy and completeness
• Correct data errors when discovered
• Update data continuously as new information arrives
• Maintain audit trails of data changes

However:

• Data may contain errors despite validation efforts
• Corrections may occur after recommendations are made
• Our data sources may have their own quality limitations
• Third-party data providers may issue revisions

See Privacy Policy → Data Governance and Quality Assurance for complete details on data quality processes.

---

4. How Recommendations Are Generated

Stock Scoring

Each stock receives a numerical score based on analysis:

Score Components:

• Technical factors (price momentum, trends, volatility)
• Fundamental factors (earnings, profitability, growth, valuation)
• Sentiment factors (investor sentiment, news tone)
• Macroeconomic factors (economic conditions, rates, inflation)

Score Aggregation:

• Individual factors are weighted based on historical predictive power
• Weights sum to produce a total score
• Higher scores suggest better opportunities
• Scores are relative (best and worst performers ranked)

Recommendation Levels

Stocks receive recommendations at different levels:

Strong Buy / Buy

• Highest scores
• Most attractive based on analysis
• Significant overweight recommended

Hold

• Neutral scores
• Fairly valued based on analysis
• Market-weight or neutral allocation

Sell / Avoid

• Low scores
• Least attractive based on analysis
• Consider underweight or avoiding

Portfolio Rebalancing Recommendations

Based on stock scores, Uptick suggests portfolio adjustments:

Rebalancing Process:

• All stocks are ranked by score
• Portfolio allocation is adjusted to favor higher-scored stocks
• Concentration limits may be applied (maximum/minimum weights)
• Turnover is considered (how much trading is required)
• Recommendations are generated for achieving optimal allocation

Output:

• Suggested % allocation to each stock or sector
• Specific trades to move from current to suggested allocation
• Rationale for rebalancing

---

5. Data Update Frequency and Freshness

Market Data (Stock Prices)

Update Frequency: Near Real-Time

• Stock prices update continuously during trading hours
• Recommendations can be updated intraday
• Latest market information reflected immediately
• Most current price trends incorporated

Trading Hours: 9:30 AM - 4:00 PM ET (regular market hours)

Corporate Financial Data

10-K Annual Filings: Once per year

• Filed within 90 days of fiscal year-end
• Immediately incorporated when available
• Remains relevant for months until next filing

10-Q Quarterly Filings: Every 3 months

• Filed within 45 days of quarter-end
• Incorporated immediately when available
• Tracks quarterly progress between annual filings

8-K Current Event Reports: As events occur

• Filed within 4 business days of significant events
• Incorporated immediately when available
• Captures material developments

Earnings Announcements: Every 3 months

• Usually within 2 months of quarter-end
• Immediately incorporated when released
• Key driver of recommendation changes

News and Sentiment Data

News Data: Continuous throughout the day

• Financial news updated as it's released
• Corporate announcements processed immediately
• News sentiment analyzed continuously

Social Media Sentiment: Continuous

• Social media activity monitored continuously
• Sentiment aggregated and updated regularly
• Reflects changing investor sentiment

Macroeconomic Data

Federal Reserve Data: Monthly to quarterly

• Fed funds rate: Changes announced in FOMC meetings
• Employment data: Released monthly by BLS
• Inflation data: Released monthly by BLS
• GDP data: Released quarterly by Bureau of Economic Analysis

Updates: Data is incorporated into algorithm immediately when released

Lag: Some macroeconomic data is released with 1-month delays (e.g., employment reports released in next month for previous month)

Overall Update Frequency

In Summary:

• Market data is nearly real-time
• Corporate financial data updates quarterly (with annual updates)
• News and sentiment data updates continuously
• Macroeconomic data updates monthly to quarterly
• Recommendations reflect latest available data from all sources

---

5.4 Detailed Backtesting Methodology

How Our Backtesting Works

Backtesting is the process of testing a trading algorithm using historical data to evaluate how it would have performed if implemented in the past. Our detailed methodology:

Step-by-Step Backtesting Process

Step 1: Historical Data Collection (5-Year Period) We collect complete historical data for the prior 5 years:

• Market Data: Daily OHLCV (Open, High, Low, Close, Volume) for all publicly traded stocks
• Corporate Financial Data: SEC filings (10-K annual reports, 10-Q quarterly reports, 8-K current reports)
• Financial Metrics: Extracted from filings (revenue, earnings, margins, cash flow, debt levels, etc.)
• News and Events: Corporate announcements, earnings releases, significant events
• Sentiment Data: Historical news sentiment, analyst recommendations, social media sentiment
• Macroeconomic Data: Historical interest rates, employment data, inflation, GDP, etc.

Data Quality Check:

• Data is cleaned and validated for accuracy
• Missing values are handled consistently
• Errors are corrected where identified
• Data is normalized for analysis

Step 2: Algorithm Development and Optimization

Algorithm is designed based on financial market research and analysis:

• Factor Selection: Based on financial theory and preliminary analysis, we select candidate factors
• Preliminary Testing: Test factors against historical data to identify promising relationships
• Parameter Optimization: Adjust algorithm parameters to maximize Sharpe ratio (risk-adjusted returns)
• Alternative Testing: Test multiple variations and approaches
• Best Version Selection: The version with highest Sharpe ratio over the 5-year period is selected

Critical Point: In this optimization process, we are essentially fitting the algorithm to historical data. This creates inherent bias toward past patterns.

Step 3: Performance Measurement

We calculate detailed performance metrics:

• Daily Returns: Calculated each day based on algorithm recommendations
• Cumulative Returns: Aggregated daily returns over time
• Annualized Returns: Returns expressed as annual percentage
• Volatility: Standard deviation of returns (risk measure)
• Sharpe Ratio: Return divided by volatility (risk-adjusted return)
• Maximum Drawdown: Largest peak-to-trough decline
• Win Rate: Percentage of recommendations that were profitable
• Other Metrics: Beta, Sortino ratio, Calmar ratio, etc.

Results: Historical backtesting may show strong results like 12-15% annualized returns with Sharpe ratio of 1.5+

Step 4: Results Analysis

We examine the backtested performance in detail:

• Overall Performance: How did the algorithm perform over the full 5 years?
• Time Period Analysis: How did it perform in different 1-year and 2-year sub-periods?
• Market Condition Analysis: How did it perform in bull markets vs. bear markets?
• Sector Analysis: Which sectors did well and which underperformed?
• Consistency: Was performance consistent or highly volatile?
• Drawdown Analysis: What was the worst losing period?

Findings May Show:

• Strong overall returns with acceptable drawdowns
• Consistency across most time periods
• Some periods of underperformance
• Some sectors performed better than others

Backtesting Assumptions and Their Impact

All our backtests assume the following:

Market Assumptions:

• Stocks traded at exact closing prices shown in historical data
• Orders executed at these prices with no slippage or delay
• Unlimited liquidity (any size order can be executed immediately)
• No impact on market prices from our trades (large trades move prices)
• Trading during regular market hours (9:30 AM - 4:00 PM ET)

Cost Assumptions:

• Zero trading commissions or fees
• Zero bid-ask spreads (difference between buy and sell prices)
• Zero market impact from executing trades
• No transaction fees or platform charges
• No margin interest or borrowing costs

Tax and Administrative Assumptions:

• No taxes on capital gains or dividends
• No tax-loss harvesting or tax management
• No rebalancing costs or complexity
• Perfect execution of all recommendations
• All dividend reinvestments are perfect

Forward Assumptions:

• Historical patterns will continue into the future
• Market structure and correlations will remain the same
• No major regulatory changes affecting markets
• No structural changes in financial markets
• Economic conditions will follow historical patterns

Why These Assumptions Matter

Each assumption makes backtested results look BETTER than realistic forward performance:

Assumption	Typical Drag on Real Performance
Commission costs	0.3-0.5% annually
Bid-ask spreads	0.2-0.5% annually
Market impact (large trades)	0.3-0.8% annually
Total trading costs	0.8-1.8% annually
Tax drag (capital gains)	0.3-1.0% annually
Implementation challenges	0.2-0.5% annually
TOTAL REALISTIC IMPACT	1.3-3.3% annually

Example:

• Backtested return: 12% annually
• Realistic drag from costs and taxes: -2% annually
• Expected actual return: ~10% annually

This 2% annual difference compounds significantly over time, reducing 10-year returns by about 18%.

Backtesting Biases Explained

Beyond unrealistic assumptions, backtesting introduces systematic biases that make results appear better than they actually are:

1. Data Mining Bias (Selection Bias)

• We tested hundreds of potential factors and combinations
• We selected factors that worked well historically
• We discarded factors that didn't work
• This creates selection bias toward factors that "got lucky"

Example: If we tested 100 factors, even random factors will show some correlation to performance by chance. By selecting only factors with high correlation, we're selecting for luck, not predictive power.

Impact: Can inflate expected returns by 2-5%

2. Optimization Bias (Overfitting)

• Algorithm parameters were optimized for the specific 5-year period
• Parameters that worked great in 2019-2024 may not work in 2025+
• Algorithm is essentially "tuned" to past data

Example: If we test 1,000 different parameter combinations, one of them will have the best backtest results. But that combination was optimized specifically for that historical period.

Impact: Can inflate expected returns by 1-3%

3. Survivorship Bias

• Backtest only includes stocks that existed throughout the period
• Excludes companies that failed, went bankrupt, or were delisted
• Dead stocks don't exist to show losses

Example: If 50 companies went bankrupt during the 5-year period, our backtest would show better results than if we included those massive losses.

Impact: Can inflate expected returns by 0.5-2%

4. Look-Ahead Bias

• Backtests have knowledge of the full 5-year period
• In real trading, we can only use past information
• Could inadvertently use "future" information in calculations

Example: If algorithm references price data from "end of quarter" and earnings are released "beginning of next quarter," there's a timing issue. Backtest might resolve this differently than real-time execution.

Impact: Can inflate expected returns by 0.3-1%

5. Period Selection Bias

• The 5-year period chosen (2019-2024) was favorable to stock market investing
• Included recovery from 2020 COVID crash
• Included strong bull market in 2021-2023
• May not include severe bear markets

Example: If we backtested on 2008-2013 (financial crisis period), results would be very different. By choosing 2019-2024, we chose a favorable period.

Impact: Can inflate expected returns by 1-3%

Total Backtesting Bias: 5-14% of reported returns may be attributable to bias, not actual predictive power.

The Backtesting Performance Gap

This explains why actual forward performance often differs from backtested results:

Reported Backtested Performance:

• Annualized return: 12-15%
• Sharpe ratio: 1.5-2.0
• Maximum drawdown: -15 to -20%

Realistic Expectation for Forward Performance:

• Annualized return: 8-10% (after costs and taxes)
• Sharpe ratio: 1.0-1.3 (lower in changing conditions)
• Maximum drawdown: -20 to -30% (worse in real crises)

Why the difference?

• Unrealistic cost/tax assumptions in backtest
• Backtesting biases (data mining, overfitting, survivorship)
• Market conditions that change over time
• Algorithm parameters not optimal for future
• Luck/favorable period selection in historical data

Conservative Investors Should:

• Discount backtested returns by 2-3% annually
• Assume worst-case drawdowns 1.5x worse than backtest
• Plan for Sharpe ratios 0.3-0.5 lower than backtest

Forward Performance Monitoring

To validate our algorithm against real market conditions, we:

1. Track Forward Performance

• Monitor actual recommendations generated in real-time
• Track implementations by users who follow recommendations
• Measure actual returns achieved
• Compare to backtested expectations

2. Compare to Benchmarks

• S&P 500 index (overall market performance)
• Other algorithmic advisors' published performance
• Passive index funds (passive investing baseline)
• Relevant sector benchmarks

3. Identify Performance Differences

• If forward performance underperforms backtest significantly, we investigate
• Could indicate market conditions have changed
• Could indicate algorithm assumptions no longer valid
• Could indicate parameter choices not optimal for current conditions

4. Algorithm Adjustments

• If performance differs materially from backtest, algorithm may be updated
• Changes only made after rigorous testing
• Changes are made solely to improve forward performance
• Users are NOT notified of internal algorithmic changes (non-material operational detail)
• Changes are tracked and documented for audit purposes

Critical Important Disclaimer

Backtested performance CANNOT be relied upon for predicting future performance.

Reasons:

• Markets change (2008 crisis, COVID, etc., change market dynamics)
• Conditions change (interest rates, inflation, geopolitical events)
• Competitive dynamics change (more algorithms using similar strategies)
• Regulatory changes (new rules affecting markets or trading)
• Algorithm parameters optimized for the past may not be optimal for the future
• Luck and favorable period selection in historical data

Our backtest shows: Our algorithm WOULD HAVE outperformed IF implemented in the past AND IF all assumptions held.

This DOES NOT mean: Our algorithm WILL outperform in the future.

Summary of Backtesting Limitations

• ✓ Shows how algorithm would have performed historically
• ✗ Cannot guarantee future performance
• ✓ Demonstrates consistent methodology
• ✗ Cannot account for unprecedented events
• ✓ Provides reasonable baseline expectations
• ✗ Understates realistic costs and risks
• ✓ Helps evaluate algorithm versus alternatives
• ✗ Inherently biased toward past patterns

---

6. Key Limitations and Assumptions

Important Caveats

6.1 Historical Data Reliance

• Algorithm is trained on 5 years of historical data
• Assumes historical patterns will continue into the future
• If markets change fundamentally, past patterns may not repeat
• Recommendations may fail if market regime shifts

6.2 Model Assumptions

• Algorithm assumes certain relationships between variables
• Assumptions based on historical data may not hold in future
• Unexpected market conditions can break model assumptions
• Black swan events fall outside model's historical experience

6.3 Data Quality Limitations

• Third-party data may contain errors
• Data may be stale or subject to revision
• We cannot verify accuracy of all data sources
• Data corrections occur after recommendations are made

6.4 No Guarantee of Future Performance

• Past backtested performance does NOT guarantee future results
• Algorithms optimized for the past may fail on future data
• Market conditions change; past success doesn't ensure future success
• Recommendations could result in significant losses

6.5 Overfitting and Backtesting Bias

• Algorithm may be overfit to historical data
• Patterns that worked in the past may not repeat
• Backtests do NOT include realistic trading costs
• Backtests assume perfect execution at unrealistic prices
• Backtests do NOT include taxes

6.6 Algorithm Cannot Predict:

• Black swan events (extreme, unprecedented events)
• Geopolitical shocks (wars, terrorism, sanctions)
• Natural disasters (pandemics, earthquakes, hurricanes)
• Technological disruptions
• Market dislocations and crashes
• Rare but impactful events outside historical experience

6.7 Non-Personalized Output

• Algorithm does NOT consider your personal situation
• Same output for identical inputs regardless of user circumstances
• Recommendations may not be suitable for your situation
• No human review for suitability

---

7. What Uptick Does NOT Do

What Uptick Does NOT Provide

Uptick explicitly does NOT:

• Provide investment advice - Recommendations are algorithmic and informational only
• Manage your money - Does NOT execute trades or manage portfolios
• Assess suitability - Does NOT determine if recommendations match your situation
• Provide financial planning - Does NOT address taxes, retirement, estate planning, insurance, debt
• Guarantee returns - Does NOT guarantee performance, profitability, or any specific results
• Protect against losses - Does NOT prevent losses or limit downside risk
• Access your accounts - Does NOT connect to brokerage accounts or have account access
• Provide tax advice - Does NOT consider your tax situation
• Provide legal advice - Does NOT address legal or regulatory matters
• Act as a fiduciary - Does NOT have legal duty to act in your best interest

What You Must Do Yourself

YOU are responsible for:

• Determining suitability of recommendations for your situation
• Conducting independent due diligence and research
• Executing trades in your own brokerage account
• Monitoring your portfolio performance
• Assessing whether recommendations align with your goals
• Consulting with qualified professionals (financial advisers, tax advisers, attorneys)
• Managing risk and understanding investment risks
• Making final investment decisions

---

8. How to Use Uptick Effectively

Best Practices

1. Use as One Input Among Many

• Do NOT rely solely on Uptick recommendations
• Consider multiple sources of information and analysis
• Cross-reference recommendations with other research
• Get second opinions on major decisions

2. Conduct Due Diligence

• Research recommended stocks independently
• Understand why you are buying each stock
• Review corporate filings yourself
• Understand the company's business and strategy

3. Assess Suitability

• Determine whether recommendations fit your situation
• Consider your risk tolerance and time horizon
• Evaluate whether recommendations align with your goals
• Ensure recommendations match your financial circumstances

4. Consult Professionals

• Talk to a financial adviser before implementing major recommendations
• Get tax advice before making significant trades
• Consult an attorney about legal or estate planning implications
• Get second opinions on major financial decisions

5. Monitor and Adjust

• Track how recommendations perform for you
• Rebalance as needed based on your situation
• Adjust recommendations if circumstances change
• Review recommendations regularly

6. Manage Risk

• Don't over-concentrate in any single stock
• Maintain appropriate diversification
• Only invest amounts you can afford to lose
• Keep emergency funds separate from investments
• Manage portfolio volatility based on your risk tolerance

---

9. Important Disclaimers

Investment Risk Disclaimer

Investing in securities involves significant risk of loss. All investments are subject to market risk and may decline in value. You could lose most or all of your investment.

Not Investment Advice

Uptick does NOT provide investment advice. Our recommendations are algorithmic outputs for informational purposes only. They do not constitute personalized investment advice tailored to your situation.

Not an Investment Adviser

Uptick is NOT a registered investment adviser with the SEC or any state regulator. We do not provide investment advisory services and have no fiduciary duty to act in your best interest.

Not Suitable for Everyone

Uptick is NOT appropriate for:

• Beginning investors without market knowledge
• Risk-averse investors who cannot handle volatility
• Investors nearing retirement who cannot recover losses
• Investors without sufficient emergency funds
• Investors who cannot afford to lose capital
• Anyone who cannot understand and accept investment risks

Consult Professionals

Before making any investment decisions based on Uptick's recommendations, you should consult with:

• A qualified investment adviser or financial professional
• A tax professional
• An attorney (if relevant)

No Guarantee

There is no guarantee that Uptick's recommendations will be accurate, suitable, profitable, or will achieve any particular returns.

---

10. Support and Additional Resources

Getting Help

Questions about How Uptick Works?

• Email: contact@uptick.ai
• Response time: Within 30 days

Additional Information

• Terms of Service: https://uptick.ai/terms-of-service

  • Legal terms, liability limitations, important disclaimers

• Privacy Policy: https://uptick.ai/privacy-policy

  • Data collection and privacy practices

• Risk Disclosure Statement: https://uptick.ai/risk-disclosures

  • Detailed explanation of all risks

• Investment Adviser FAQ: https://uptick.ai/investment-adviser-faq

  • Answers to common questions about Uptick and investment advice

---

11. Contact Information

For questions or feedback about this page:

Email: contact@uptick.ai

Mailing Address: Uptick, Inc. Attn: Legal Team 1120 Avenue of the Americas Fourth Floor New York, NY 10036 United States

Response Time: We will respond to inquiries within 30 days.

---

END OF HOW UPTICK WORKS

---