Is there a bug in FPL() function when plotting in strategies vs studies?

[dudemanguy]

I'm trying to use FPL in order to determine how many shares to order in strategies. But for some reason when FPL is plotted in strategies it returns a different number than when plotted as a study, and the slippage makes the backtesting inaccurate. Either that or I'm doing something wrong with the code. Here's an example for a simple MA200 strategy:

input length = 200;
input price = close;
input aType = AverageType.SIMPLE;

plot MA = MovingAverage(aType, price, length);
ma.SetDefaultColor(Color.GREEN);

def buy = price > MA and price[1] < MA[1];
def sell = price < MA and price[1] > MA[1];

plot FPL = FPL();
input initial = 10000;
def portfolio = initial + FPL();

input kelly = 1.0;
def orderShares = RoundDown(kelly * portfolio/open[-1],0);

AddOrder(OrderType.BUY_TO_OPEN, buy, tradesize = orderShares, tickColor = Color.GREEN, arrowColor = Color.GREEN, name = "");

AddOrder(OrderType.SELL_TO_CLOSE, sell, tickColor = Color.RED, arrowColor = Color.RED, name = "");

I'ved tested it on different tickers, ETFs, timeframes, and there is always some slippage. Any insight appreciated!

Just to clarify, it seems that if you use

def portfolio = initial + FPL();

to determine bet sizing it seems to start its own FPL calculation as opposed to using the internal FPL() calculation. Changing tickers seems to fix the problem, but if the study parameters are updated then the behavior starts happening again.

 

[Ramisegal]

I'm trying to use FPL in order to determine how many shares to order in strategies. But for some reason when FPL is plotted in strategies it returns a different number than when plotted as a study, and the slippage makes the backtesting inaccurate. Either that or I'm doing something wrong with the code. Here's an example for a simple MA200 strategy:

input length = 200;
input price = close;
input aType = AverageType.SIMPLE;

plot MA = MovingAverage(aType, price, length);
ma.SetDefaultColor(Color.GREEN);

def buy = price > MA and price[1] < MA[1];
def sell = price < MA and price[1] > MA[1];

plot FPL = FPL();
input initial = 10000;
def portfolio = initial + FPL();

input kelly = 1.0;
def orderShares = RoundDown(kelly * portfolio/open[-1],0);

AddOrder(OrderType.BUY_TO_OPEN, buy, tradesize = orderShares, tickColor = Color.GREEN, arrowColor = Color.GREEN, name = "");

AddOrder(OrderType.SELL_TO_CLOSE, sell, tickColor = Color.RED, arrowColor = Color.RED, name = "");

I'ved tested it on different tickers, ETFs, timeframes, and there is always some slippage. Any insight appreciated!

Just to clarify, it seems that if you use

def portfolio = initial + FPL();

to determine bet sizing it seems to start its own FPL calculation as opposed to using the internal FPL() calculation. Changing tickers seems to fix the problem, but if the study parameters are updated then the behavior starts happening again.

I had similar experiences...
Check out this post it may assist you troubleshoot.
Thread 'GandalfProjectResearchSystem Dashboard For ThinkOrSwim' https://usethinkscript.com/threads/gandalfprojectresearchsystem-dashboard-for-thinkorswim.17429/

 

[dudemanguy]

I had similar experiences...
Check out this post it may assist you troubleshoot.
Thread 'GandalfProjectResearchSystem Dashboard For ThinkOrSwim' https://usethinkscript.com/threads/gandalfprojectresearchsystem-dashboard-for-thinkorswim.17429/

I found it was related to memory. If I have too many charts open and make calculations over a really long time period, it will tend to get lazy and not calculate correctly. The easy fix is to just go to options, turn on extended hours, update, then turn off extended hours, update. It will force the correct calculations.