Research
Trading System Performance Metrics

Updated: 2014/09/15


Trading system performance metrics are used to compare and evaluate trading system viability. Back test reports and system optimizations will generate many different types of quantified results. Results should fall within the acceptable range for each metric, but their interpretation is very dependant on type of system they represent as well as the psychology of the trader.

Metric

Description

Preferred Value

Net Profit

Final Capital less Initial Capital.

Final Capital - Initial Capital

Higher Net Profit is preferred.

Good
Bad

More than 0
Less than 0

Net Profit %

Net Profit divided by Initial Capital.

(Net Profit / Initial Capital) * 100

Higher Net Profit % is better.

Good
Bad

More than 0
Less than 0

Exposure %

Sum of the number of exposed bars divided by the total number of bars. Single bar exposure is the value of open positions divided by portfolio equity.

Lower Exposure % is preferred.

Considered per system a low exposure % is preferred, although it may also be an indication that capital is under-employed.

Net Risk
Adjusted Return %

Net Profit % divided by Exposure %.

Net Profit % / Exposure %

Higher Net Risk Adjusted Return % is better.

Compounded Annual Return %
CAR %

The percentage rate of return which, if compounded over the time period of the trading test, would yield the cumulative gain or loss identical to the trading system.
Since trading income does not normally grow at a constant rate, the CAR% calculation smooths the return in order to make in simpler to compare with that of other investment types.

Correct formula:
100% * ( (Final Capital / Initial Capital) ^ (365 / Days in Test) - 1 )

Simplified formula:
Net Profit % * (365 / Days in Test)

Higher CAR % is better.

Good
Neutral
Bad

More than 10%
From 0% to 10%
Less than 10%

Risk
Adjusted Return %
RAR %

RAR % includes the system Exposure % as a risk measurement in the calculation of return.
This makes RAR % more useful when comparing the returns of systems or portfolios with different risk levels.

 CAR % / Exposure %

Higher RAR % is better.

Good
Bad

More than 10/Exposure
Less than 0

Average Profit/Loss $
Expectancy $

Expectancy $ is the currency or points size of the average profit compared to the size of the average loss per trade.
This represents expected currency or points gain or loss per trade.

(Profit of Winners + Loss of Losers) / Number of Trades

Higher Expectancy $ is preferred. Must be greater than 1.

Good
Bad

More than 0
Less than 0

Average Profit/Loss %
Expectancy %

Expectancy % is the percentage size of the average profit compared to the size of the average loss per trade.
This represents expected percentage gain or loss per trade.

(% Profit of Winners + % Loss of Losers) / Number of Trades

Higher Expectancy % is generally better. Must be above 1.

Good
Bad

More than 0
Less than 0

Average Bars Held

The number of bars or candles the average trade is in the market.

Sum of Number of Bars in Trades / Number of Trades

Longer is often considered better. Length should relate to the type of trading system.

Maximum Trade Drawdown $

The largest peak to valley currency or points decline experienced in any single trade.

The lower the better.

Maximum Trade Drawdown %

The largest peak to valley percentage decline experienced in any single trade.

The lower the better.

Maximum System Drawdown $
MaxSysDD $

The largest peak to valley currency or points decline experienced in portfolio equity over the back test period.

The less the better.

Maximum System Drawdown %
MaxSysDD %

The largest peak to valley percentage decline experienced in portfolio equity over the back test period.

The lower the better.

Good
Neutral
Bad

0% to 10%
10% to 30%
More than 30%

Recovery Factor

An indication of how rapidly a system recovers from drawdowns.

 Net Profit / Maximum System Drawdown 

Higher is better. Must be above 1.

Good
Neutral
Bad

More than 2
From 1 to 2
Less than 1

CAR/MaxDD

Rates the Compounded Annual Return of the system by the largest drawdown required to achieve the gain.
This is the default optimization target for Amibroker and a good starting point for choosing your system or parameter set.

Compounded Annual Return % / Maximum System Drawdown %

Higher is better.

Good
Neutral
Bad

More than 2
From 1 to 2
Less than 1

RAR/MaxDD

Similar to CAR/MaxDD, but using the Risk Adjust Return % instead of the CAR %. This method includes an element of the system Exposure % in the resulting ratio.

Risk Adjusted Annual Return % / Maximum System Drawdown %

Higher is better.

Good
Neutral
Bad

More than 2
From 1 to 2
Less than 1

Profit Factor

Profit of the winning trades divided by the loss of the losing trades.

Higher is better.

Excellent
Good
Neutral
Bad

More than 3
More than 2

From 1 to 2
Less than 1

Payoff Ratio

The ratio of the Average Winning trade to the Average Losing trade.

Average Win / Average Loss 

Higher is better.

Good
Bad

More than 1
Less than 1

Standard Error

Standard Error measures the volatility of the Portfolio Equity Line. A smoother equity line is preffered.

Lower is better.

Risk-Reward Ratio

RRR measures the relation between the risk inherent in the trading system and its potential gain.

Slope of Equity Line / Standard Error

Higher is better.

Ulcer Index

UI measures investment risk and includes the depth and duration of drawdowns in the calculation.
It focuses purely on losses.

Square Root of the Sum of Squared Drawdowns / Number of Bars

Less is better.

Ulcer Performance Index

UPI compares performance to a risk free investment using UI to include its associated downside risk.
The risk free rate is set as 5.4%.

(Annualized Return – Risk Free Rate) / UI

Higher is better.

Sharpe Ratio

Sharpe Index is a widely used metric that rates the performance of a system by adjusting for risk. Sharpe is often criticized for penalizing large profits as well as losses.

Calculation: Investopedia

Higher is better.

Excellent
Good

More than 2
More than 1

K Ratio

K detects inconsistency in returns. The higher K Ratio is the more consistent the return you may expect from the system.

Calculation: Investopedia

Higher is better.

Good

More than 1

Win %

Percentage of trades resulting in a Net Profit.

Higher is always better.

Mean Reversion systems tend to have a much higher Win % than Trend Following systems.
Some succesful systems may have a very low Win %.
A higher Win % has a positive impact on trading psychology.

Trades #

The number of trades made to achieve the result.

Lower is usually better.

Fewer trades made results in less slippage and less trading costs.