Position sizing discrepancies between calculation, backtest and realtime

Hi all,

I wonder if anyone can help me understand how PRT deals with position size calculation and execution. I recently posted a strategy (https://www.prorealcode.com/topic/is-this-system-too-good-to-be-true/) that uses Nicolas’ dynamic position sizing function to calculate the position size based on the distance to the stop loss and the percent of equity at risk.

Here’s a snippet of the position size function

// money management
ONCE Capital = 5000
ONCE Risk = 0.01
equity = Capital + StrategyProfit
maxrisk = round(equity*Risk)
PositionSizeL = abs(round((maxrisk/StopLossL)/PointValue)*pipsize)
PositionSizeS = abs(round((maxrisk/StopLossS)/PointValue)*pipsize)

// open a long position and implement stop loss
BUY PositionSizeL PERPOINT AT MARKET
StopLossL = abs(close - LOWEST[10](low)) + sp
SET STOP LOSS StopLossL

In the attachment you find a screenshot of the strategy implemented in realtime trading and also the backtest overlaid with the PositionSizeL variable graphed. You can see that:

1. PositionSizeL calculated by the above function = 3
2. The position size implemented in the backtest = 10
3. The position size implemented in realtime = 10
4. The initial equity was £5000 and the PositionSizeL function only permits positions to be opened with a max risk of 1% of the account equity, which in this example is £50. You can see from the screenshot that the amount that would be lost of the stop is hit is £176, more than double the max risk.

Can anyone explain the discrepancy between the graphed value of PositionSizeL = 3 and the position size implemented in the backtest and realtime trading? Does it have something to do with the ‘BUY … PERPOINT’ command? How does PRT pass the PositionSizeL number to the engine that actually implements the executed position size?

Screenshot-2019-08-28-at-09.39.20.png