Secrets to Short-Term Trading Ch. 1: Why a Losing Streak Feels Like Proof, Even When It Isn't

Investment Background

Larry Williams built one of the most public track records in trading — turning $10,000 into more than $1.1 million in a single year at the 1987 World Cup Trading Championship — trading systems whose win rate was frequently below 50%. He spent decades afterward making the same point in different forms: the number people obsess over when judging a trading system — how often it wins — is close to useless on its own, and worse, it actively misleads people about what they should expect to feel while running it.

The library already has the arithmetic for this. way-of-the-turtle Chapter 1 derives the expectancy formula and shows, with a worked example, why a 36%-win-rate system can out-earn a 90%-win-rate system. That math is correct and this book will not re-derive it. What Williams's own career adds is not a better formula — it is decades of first-person testimony about what it is actually like to sit inside a statistically sound losing streak, and why almost nobody can do it on the strength of the formula alone.

The Wall Street Translation

The Formula Was Never the Hard Part

Here is the gap this chapter exists to close: knowing the expectancy math and being able to act on it under a real losing streak are two entirely different skills, and the second one is what actually determines whether a trader survives.

Consider a system with a genuine, back-tested 38% win rate and a payout ratio that makes it solidly profitable in expectancy — the same shape of system way-of-the-turtle ch01 shows can outperform a 90%-win-rate system. Run the arithmetic forward and a losing streak of ten, twelve, even fifteen consecutive losses is not a warning sign. It is an expected, unremarkable event that will occur multiple times over the life of the system.

Knowing that in advance changes nothing about how it feels in month four of a drawdown, watching account equity fall while every instinct argues that a system that keeps losing is a system that has stopped working.

Two Separate Judgments Getting Fused Into One

The specific error Williams's career keeps illustrating: traders fuse two separate judgments — "is my edge still valid" and "have I had a run of bad luck" — into a single feeling, and the feeling always points toward abandonment during a drawdown, regardless of which judgment is actually true.

A system's edge can degrade for real reasons: the market regime that produced the edge has genuinely changed, competition has arbitraged it away, or the original backtest was flawed. Those are legitimate reasons to stop. A losing streak that falls within the range the system's own historical statistics predicted is not one of them — but it produces an identical emotional signal to the legitimate case, which is exactly why the two get confused.

Division of Labor With the Rest of the Library

Book	Owns
way-of-the-turtle ch01	The expectancy formula itself — why win rate alone is the wrong number, proven with a worked calculation
man-who-solved-the-market-simons ch02	Trusting a statistically validated signal with zero narrative support — the general psychology of holding a position you cannot explain
This book	The specific experience of surviving a predictable, statistically normal losing streak without mistaking it for edge decay — a narrower, more operational problem than either book above addresses

The distinction from the Simons book matters: that chapter is about tolerating a position without a story for why it works. This chapter is about tolerating a losing streak your own system's history told you to expect — the story exists, the data exists, and the difficulty is purely emotional endurance, not epistemic uncertainty.

Executable Trading Rules

1. Before running any system live, write down its historical maximum losing streak — not its win rate. If the system has produced fourteen consecutive losses in backtest, a live losing streak of fourteen is not new information. It is the system behaving exactly as documented.

2. Separate your review calendar from your emotional state. Decide in advance — weekly, monthly, quarterly — when you will actually evaluate whether the edge has changed, using data (regime shifts, structural changes in the market, a live track record diverging from backtest by more than a stated threshold). Never let a bad week trigger an off-schedule review; that is exactly when judgment is least reliable.

3. Track streaks, not just returns. Most trading journals record profit and loss. Add a running count of consecutive losses next to your system's documented historical maximum, so a live streak of nine against a historical worst of fourteen reads as "within range," not "alarming."

4. Expect the hardest moment to arrive right before the system works again, not at the start of a drawdown. The temptation to abandon a sound system compounds with each additional loss, which means statistically the strongest urge to quit tends to cluster near the end of a normal losing streak — not the beginning.

Relevance to a Retirement Portfolio

This chapter's lesson transfers to retirement investing more directly than its short-term trading origin suggests. A globally diversified equity core will, with certainty, produce multi-year stretches of underperformance relative to some other visible benchmark, and multi-year stretches of nominal losses during a bear market. Both are statistically normal and both produce the identical emotional signal — "this isn't working" — that a losing streak produces for a trading system.

The retirement-specific version of Rule 1 is: know your portfolio's historical maximum drawdown and worst multi-year stretch before you are living through one, so that when it arrives, it reads as expected behavior rather than proof that the plan has failed. This is not an argument for abandoning diversification during a downturn — it is the opposite: the ability to tell a normal drawdown from a genuine change in the plan's soundness is what keeps a sound retirement strategy intact through the exact periods that end undisciplined ones.

Chapter 2 turns to a different kind of misread signal: institutional positioning data that is real and free, and routinely misunderstood in the opposite direction — read as a timing tool when it isn't one.