Read monthly and yearly EA performance

Define the monthly result

For a closed-results table, group net position outcomes by the stated closing date and timezone. Reconcile profit, commission, swap and any other report charges once; adding already included costs again understates results. A position opened in one month and closed in another belongs to the closing month in this method. Open losses may therefore be absent from that table.

Name the return denominator

EUR 300 profit is 3% of EUR 10,000 and 1% of EUR 30,000. Both ratios are arithmetic, but they describe different capital assumptions. Sum cash P/L for yearly cash totals. Compound successive period returns only when they consistently represent the same account and deposits or withdrawals have been handled; do not blindly add percentages with changing denominators.

Look for concentration and shared losses

Compare the worst month, negative-month count, trade count and the share contributed by a few exceptional months. For three EAs, align the same covered months before summing results or computing dependence. Keep a verified inactive month as zero, a missing month as unavailable and the current year as year-to-date. A monthly cash chart cannot establish intraday portfolio equity drawdown.

Checks to make before deciding

• Group closed net outcomes by exit month in a declared timezone. If using marked-to-market returns, label that different method and reconcile open positions across boundaries.
• Show every month in the usable common history. Distinguish verified no-trade months from missing data; display missing months with an unavailable marker, not a green zero.
• Report yearly net P/L, trade count, equity drawdown and exposure. Label partial years and avoid annualizing a short favorable period without explaining the assumption.
• Measure the share of positive months and years together with losing-month magnitude. A high positive-month fraction alone says little about the size of a rare negative month.
• Align all series to the common verified period and timezone. Preserve zero-activity periods but exclude genuinely unavailable data; document how many paired months remain.
• Compute correlations from comparable periodic returns, not cumulative balance curves. Declare denominator, return convention, frequency and Pearson or alternative measure.
• Show the full matrix plus rolling or subperiod estimates where the sample permits. A single long-period correlation can conceal periods when all strategies lost together.
• Check shared currency, direction, session, volatility and broker exposure. Different EA names do not ensure different economic risk factors.