How to read a backtest equity curve

How the line is built

TickRun starts strategy equity at the entered initial capital. Each session’s net strategy return changes the previous value:

equity[t] = equity[t−1] × (1 + strategy_return[t])

When the strategy is long, it earns the close-to-close asset return using the position held from the previous session. When it is flat, its market return is zero. A transaction cost is subtracted when position changes. The curve therefore includes compounding, open positions, and modeled costs.

The Buy & Hold line begins with the same capital and compounds the selected ticker’s daily adjusted-close return. It stays exposed throughout the loaded interval. Both lines use the same vertical unit, making direct value comparison possible.

Read shape before endpoint

A steadily rising curve is visually appealing, but ask what produced the smoothness. A strategy may remain flat for years and then make one successful trade. A stale or incorrectly filled price series can also create artificial smoothness. Conversely, a valid trend strategy may show many small setbacks before a few strong periods.

Break the line into segments. Where did most growth occur? Are gains spread across several market environments? Does performance vanish after the first half? A final value can hide structural change.

Look below previous peaks

A drawdown starts when equity falls below its running high. It ends only after that high is recovered. The deepest percentage decline is maximum drawdown, but the chart also reveals duration and repeated smaller declines.

Two strategies can share a −20% maximum drawdown. One may recover in a month; the other may stay below its peak for three years. The percentage alone does not show that difference. Trace the peak, trough, and recovery visually, then use the drawdown guide for the exact calculation.

Understand flat sections

A flat strategy line usually means the position is zero and TickRun is assigning zero return to cash. Flatness is not automatically safety or failure. It shows that the entry rule did not hold market exposure during that interval.

Compare the benchmark during the same section. If Buy & Hold falls while strategy equity stays flat, the rule avoided a decline. If the benchmark rises, the rule missed that gain. Over a high-interest-rate period, real cash might have earned something, but TickRun’s current curve does not credit it.

Investigate sudden jumps

A sharp upward step can be a genuine large move, but it deserves inspection. Find the date on the price chart, check whether the strategy was already long, and examine corporate actions or missing sessions. A signal created at the end of a session should not earn the return into that same close.

A split in raw data can create a false collapse or jump. TickRun uses saved adjusted OHLCV, but manually replaced files must keep all fields consistent. Very large daily changes should be reconciled rather than accepted because they improve results.

Compare with Buy & Hold carefully

If strategy equity finishes above the benchmark, ask how. Did it avoid a major decline, participate in gains, or simply take a different amount of risk? If it finishes below, ask whether drawdown was meaningfully reduced. Return without exposure and risk context gives an incomplete comparison.

TickRun’s benchmark has no modeled entry or exit cost, while active position changes pay the selected cost. It is continuously invested, while the strategy can be flat. Those are known differences, not calculation errors. The benchmark guide explains alternative comparisons.

Linear and logarithmic views

TickRun’s displayed portfolio chart uses a conventional value scale. On a linear scale, the same vertical distance represents the same amount of money. As wealth grows, later percentage moves look larger: a 10% rise on 20,000 is twice the money change of a 10% rise on 10,000.

A logarithmic scale instead gives equal vertical distance to equal percentage changes. It is useful for long histories with large growth. TickRun does not currently offer a log toggle, so calculate percentage changes when comparing early and late slopes rather than relying only on visual angle.

Check concentration

A curve whose profit comes from one brief period is less broadly supported than one showing repeated contributions. Remove the best trade as a thought experiment. Does the overall conclusion survive? Check the best year, largest gap, and strongest trend.

This is not permission to delete inconvenient observations. The purpose is sensitivity analysis: understanding what the result depends on. If one event controls the answer, say so and seek evidence on later untouched data.

Common warning signs

• An immediate jump on the entry signal’s own closing return.
• A split-like 50% move with no economic explanation.
• A perfectly smooth line from illiquid or missing prices.
• A huge late gain that dominates ten years.
• Strong gross results that disappear with modest costs.
• A strategy line beginning later than the benchmark without a clear warm-up convention.
• A final open trade that is invisible in completed-trade statistics.
• A chosen curve that was the best of thousands but has no out-of-sample test.

A five-minute curve review

1. Confirm both lines start with the same capital and dates.
2. Locate the biggest rises and falls on the price chart.
3. Mark the worst drawdown’s peak, trough, and recovery.
4. Identify flat periods and compare them with the benchmark.
5. Check whether profits are concentrated in one trade or regime.
6. Review transaction costs and the final open position.
7. Compare default, nearby, and optimized settings.
8. Reserve judgment until the rule is tested on later data.

What the curve can and cannot say

An equity curve is the clearest summary of the simulated wealth path under stated rules. It can reveal timing, drawdowns, concentration, and long periods of inactivity. It cannot prove that historical fills were available, that parameters were not overfit, or that future returns will follow the same path. Read it as evidence to investigate, not a promise.