I’m not sure that what Donald says is different from what Dustin and Science Trader said. The point is this. Suppose that n(i) is the number of units on trade i, and that p(i) is the profit per contract on trade i, and that N is the number of traders. Then the current P/L per unit is computed as
PL1 = Sum{p(i) ; i = 1, …, N} / N
The other formula is
PL2 = Sum{n(i)p(i) ; i = 1,…, N} / Sum{n(i) ; i == 1,…, N}
I’ve argued for months that the second formula is better. This is what Dustin and Science Trader suggest as well. It implies a weighting of the trades since you can also write it as
PL2 = Sum{w(i)p(i); i = 1,…,N}
where the weights are w(i) = n(i) / Sum{n(i); i=1,…,N}.
The fact that weights are used does not invalidate it as a method to detect scalping systems.
Ah, that is indeed different, and I would not do this in the P/L per unit statistic. It’s just another statistic.
The same statistic quoted in percentage terms can also be provided for comparison. Some statistics like the Sharpe Ratio, % Wins and Return on Starting Equity are better expressed in percentage terms exclusively. Some other statistics that can be quoted in both dollar and percentage terms are:
Largest Winning Trade ($), Largest Winnning Trade (%),
Average Winning Trade ($), Average Winning Trade (%),
Largest Losing Trade ($), Largest Losing Trade (%),
Average Losing Trade ($), Average Losing Trade (%),
Average Trade ($), Average Trade (%),
Std Dev of Ave Trade ($), Std Dev of Ave Trade (%),
Win/Loss Ratio ($/$), Win/Loss Ratio (%/%),
Average Drawdown ($), Average Drawdown (%),
Worst Case Drawdown ($), Worst Case Drawdown (%).
