Using Streak Analysis to Understand Your Bot's Patterns

A three-trade losing streak feels significant. A five-trade winning streak feels like confirmation that something is working. But whether those sequences are actually meaningful depends entirely on context, how long your streaks typically run, what conditions were present during them, and whether the pattern points toward a configuration opportunity or is simply normal variance.

Relay's Streak Analysis examines the winning and losing streaks in your bot's trade history and surfaces what those patterns actually mean. This guide covers how to run the analysis, how to interpret the results, and how to use streak data as a practical input for configuration decisions.

What streak analysis examines

Relay's streak analysis looks at your closed trade history and identifies:

• Current streak — whether you're currently on a winning or losing run and how long it's been
• Historical streak distribution — how long winning and losing streaks typically run for your specific bot
• Context around streak transitions — what conditions were present when a streak ended
• Patterns in streak timing — whether streaks correlate with time of day, day of week, or volatility environment

The analysis compares your current streak against your bot's historical streak distribution to give you perspective on whether what you're seeing is unusual or within normal range.

Triggering streak analysis

From the capability chip or grid: Click "Streak Analysis" from the Relay welcome screen chips or select it from the full capability menu.

Through conversation:

"Run a streak analysis on my QQQ UT bot." "What's my current streak on SPY BEAR ODTE?" "How long do losing streaks typically run for my bots?" "Am I on a winning or losing streak right now, and how does it compare historically?"

You can run the analysis for a single bot or across your portfolio. Per-bot analysis is usually more informative because different strategies naturally produce different streak characteristics.

Reading the output

Your current streak Relay reports what streak you're currently on and for how long. This is the most immediately actionable piece of information, it tells you whether you're in a run, and whether that run is unusual given the bot's history.

Historical distribution The analysis shows how losing and winning streaks have typically distributed across your bot's full trade history:

• Average winning streak length
• Average losing streak length
• Maximum winning streak
• Maximum losing streak
• How frequently streaks of different lengths have occurred

This context is what makes the current streak reading useful. A 4-trade losing streak means something different for a bot whose average losing streak is 2 trades versus one whose average losing streak is 5 trades.

Streak context Relay surfaces what conditions are typically present during streaks, VIX levels, time of day, day of week, to help you understand whether the current streak appears connected to identifiable market conditions.

What streaks do and don't tell you

This is the most important thing to understand before acting on streak data: a streak is not inherently a signal.

In any probabilistic system, and trading is probabilistic, runs of wins and runs of losses occur naturally, even in a well-functioning strategy. A three-trade losing streak on a bot with a 75% win rate is expected to happen regularly. It doesn't mean the bot is broken.

What streak analysis helps you do is distinguish between:

Normal variance — streaks that fall within the expected distribution for your bot, based on its historical win rate and the characteristics of its trade history. These don't require action.

Unusual patterns — streaks that are significantly longer than your bot's historical norm, or patterns where losing streaks consistently cluster around specific conditions. These warrant investigation.

The key question isn't "is this a losing streak?" It's "is this losing streak unusual for this bot, and if so, why?"

When a losing streak warrants deeper review

Use these signals to decide whether to investigate further:

The streak length exceeds your bot's historical maximum If your bot has never had a losing streak longer than 4 trades and you're now at 7, that's outside the historical range and worth examining.

The streak is coinciding with a VIX environment your bot struggles with Check VIX Correlation to see whether the current conditions are in a range where your bot historically underperforms. If so, the streak may be contextual rather than a configuration problem.

The streak started after a configuration change If you modified the bot recently and a losing streak began shortly after, there's a potential connection worth evaluating using What-If Simulation.

Win rate has materially declined alongside the streak A streak of losses is more concerning when accompanied by a broader decline in win rate over the same period, rather than occurring within an otherwise normal win rate pattern.

When a losing streak probably doesn't require action

• The streak is within the normal distribution for your bot's history
• The streak coincides with clearly unfavorable conditions (high VIX for a low-VIX bot, unusual market events)
• The streak is short in absolute terms and your win rate over the broader period is intact
• You haven't changed anything recently and the bot has had similar streaks before that resolved on their own

Making configuration changes in response to normal variance is one of the most common ways traders degrade bot performance. The streak analysis is designed to help you distinguish between the two cases before deciding to act.

Cooldown settings and streaks

One configuration parameter that interacts directly with streak behavior is the cooldown window, the minimum time between trades. If you're finding that your bot often chains consecutive losses in quick succession (short losing streaks at high frequency), a longer cooldown may reduce that pattern by filtering out trades that trigger immediately after a loss.

Use streak analysis to surface the timing of consecutive losses, then use What-If Simulation to model whether a longer cooldown would have reduced streak depth without significantly reducing overall trade count or win rate.

Winning streaks: the other side

Winning streaks also deserve attention, though they're often overlooked. A bot on a long winning streak may simply be in a favorable environment, which means its recent performance may not be representative of its typical behavior.

If the weekly review shows strong recent results, check whether those results coincide with an unusually long winning streak. If the streak is well above the historical average, maintain healthy perspective on the recent numbers and continue monitoring rather than sizing up based on a hot streak alone.