How to read a trading track record, and how to spot a fake one

The five numbers

CAGR

Compound annual growth rate: what the account made per year, compounded. Ours: 41.8% in backtest, net of costs and funding. Meaningless without the next four.

Max drawdown

The deepest fall from a peak to the following trough. Ours: −18.8%. This is the number a copier actually lives through, and the one leaderboards hide.

Volatility

How much the account swings, annualised. Ours: 24.3%. A 40% return at 24% volatility and a 40% return at 80% volatility are different products.

Sharpe ratio

Return per unit of volatility. Ours: 1.56. Above 1 is good for a liquid strategy over years; above 3 over years is usually a measurement error or a fraud.

Calmar ratio

CAGR divided by max drawdown. Ours: 2.23. It answers "how much return per unit of pain", which is the copier's question.

Sortino (return per unit of downside volatility, ours 1.97) refines Sharpe for strategies with lopsided returns, which trend following has. Win rate, profit factor and average win are mostly noise: a trend strategy can have a 40% win rate and excellent returns, because the wins are large and the losses small.

Backtest, demo, paper, live

A backtest is a simulation on past data. It proves the rules would have worked on that data, computed with full knowledge of what happened. It says nothing about whether the rules were chosen because they fit that data. Treat every backtest, including ours, as an upper bound.

Paper trading simulates fills at live prices. Demo trading sends real orders to an exchange's test environment with simulated capital, so order handling, rejections and fills are real. Live is real capital. Only live records prove execution, and only long ones prove anything about returns. A record should say which it is on every page; ours labels every account DEMO, PAPER or LIVE.

Why short records mean nothing

In our 300-week backtest, 52% of individual weeks lost money and 42% of four-week periods. Over twelve months, 1% of periods did. A strategy that compounds at 40% a year still looks broken one week in two. So a two-week record, good or bad, carries almost no information, and a leaderboard sorted by 7-day or 30-day ROI is sorted by luck.

Red flags

• No losing trades visible. Every real record has them. A profile showing only winners is either curated or has closed losers under a different account.
• Restarted records. A "since inception" date that keeps moving forward means the drawdowns were deleted with the old account.
• ROI on margin. Some leaderboards compute return on the margin posted, not on equity. A 10× position that gains 5% shows as "+50% ROI". Ask which denominator.
• Returns before funding. Funding can cost a leveraged long book 8–9% a year in normal conditions and far more in a bull market. Gross figures flatter.
• Screenshots instead of data. A record you cannot check against the exchange is a claim, not a record.
• Sharpe above 3 over years, or no drawdown above 5% in a leveraged crypto strategy. Either the period is short, the measurement is wrong, or the losses are hidden.
• Leverage advertised as a feature. "125×" on a profile is a warning about the trader, not a description of the strategy.

What a credible record shows

A date it started that has never changed; every position and fill, including losses; the label of what capital it is; returns net of fees and funding; a drawdown figure next to every return figure; and, once live, a link to the exchange's own page for the account so the numbers can be checked against the source. That is the standard this site is built to, and it is the standard to hold any lead trader to before copying them.