Glossary for strategy testing

The terms used in the library, explained without jargon. Everything is a basis for your own studies — not investment advice.

Calmar

Annual return divided by the largest drawdown. A high Calmar means good return relative to risk. We calculate after financing cost and always compare with simply owning the index.

Profit factor

The sum of all profits divided by the sum of all losses. Above 1 means the profits are larger than the losses. The figure says nothing about how large individual losses are.

Largest drawdown

The largest fall in capital from a peak to a trough. We measure on running value, including unrealised loss in an open trade, the way ProRealTime does.

Time without a new peak (stagnation)

The longest time the strategy has gone without reaching a new high. The drawdown shows how deep it fell, the stagnation how long it took to recover.

Straightness (R²)

How straight the equity curve is. A value near 1 means steady growth, near 0 means the profit came in lumps or from a single period.

Buy & hold

Simply buying the index and holding it. All strategies are compared with it, over the same period and without leverage, because a strategy that is in the market only part of the time otherwise looks better than it is.

Financing

The interest cost charged when a CFD or spread betting position is held overnight. It affects strategies that stay long in the market and is deducted in our figures.

Spread

The difference between the buy and sell price. In our code files the spread is given in the header and should be set in ProBacktest, not put in the code.

PRT verified

The strategy's trades have been compared with a real ProBacktest run on the same market and timeframe. The label says the code does what we say, not that the strategy is profitable.

Quality gate

A test on the history: enough trades, a profit factor above the limit, a smooth curve and profitability in at least three of four sub-periods. Passing the gate is not a guarantee, only a reasonable starting point.

ProfitFeed data and IG data

ProfitFeed provides long minute data that we adjust to IG and use before the first bar of the IG series. IG data is the real data. The ProfitFeed part is indicative and reported separately.

Monte Carlo

The trades are reshuffled in random order many times to see how large a drawdown one could have had with bad luck. It shows the spread around the actual result.

Out-of-sample

Data that was not used when the strategy was chosen. Without such a part it is impossible to tell a robust idea from one that merely fits the past.