Doing Some Backtesting by Hand for the first time, have a few ?s

zghorner

zghorner

upload_2023-10-24_17-14-44.png


This is trading CL futures front month only, 15m candles, between certain daily hours + volatility level as condition for entry, both entry and exit signals are extremely simple and easy to spot.

My initial system of this style looked good to the eye and did well in trending markets but got absolutely obliterated in chop...came back with a PF < 1.0. SL far too loose.

I've been adjusting it this week and the above picture is my latest result. kept the same entry as it was clear as day/reliable and modified exit signal as well as added one other condition for entry. Its super simple and I am confident it will continue to perform after adding additional rows of data...but:

1) What sort of PF do you look for when testing? In the above I have added 5 points against me for each exit to account for slippage, fast markets, etc... and feel like 1.77 is conservative.
2) Where do you see obvious improvement that can be made? To me and my limited knowledge...MAE:MFE ratio looks solid. ETD looks like shit. My entry/exit are reliable and easy to trigger but I will continue to explore improvement. Need to add another metric for points moved in favorable direction before entering to give me a better idea on timing the move. ETD giving back an average of 50 points per trade (from MFE) before exiting just sounds horrible. recommendations?
3) What is your opinion about Tradingview backtesting? I am liking doing it by hand as I feel I can trust the results but it is time consuming...Im not trying to test this decades back but at least several months up to a few years would be nice.

thanks ET you the best BB gurl.
 
zghorner said:
View attachment 326001

This is trading CL futures front month only, 15m candles, between certain daily hours + volatility level as condition for entry, both entry and exit signals are extremely simple and easy to spot.

My initial system of this style looked good to the eye and did well in trending markets but got absolutely obliterated in chop...came back with a PF < 1.0. SL far too loose.

I've been adjusting it this week and the above picture is my latest result. kept the same entry as it was clear as day/reliable and modified exit signal as well as added one other condition for entry. Its super simple and I am confident it will continue to perform after adding additional rows of data...but:

1) What sort of PF do you look for when testing? In the above I have added 5 points against me for each exit to account for slippage, fast markets, etc... and feel like 1.77 is conservative.
2) Where do you see obvious improvement that can be made? To me and my limited knowledge...MAE:MFE ratio looks solid. ETD looks like shit. My entry/exit are reliable and easy to trigger but I will continue to explore improvement. Need to add another metric for points moved in favorable direction before entering to give me a better idea on timing the move. ETD giving back an average of 50 points per trade (from MFE) before exiting just sounds horrible. recommendations?
3) What is your opinion about Tradingview backtesting? I am liking doing it by hand as I feel I can trust the results but it is time consuming...Im not trying to test this decades back but at least several months up to a few years would be nice.

thanks ET you the best BB gurl.
More...

What do you mean backtesting manually by hand? Like looking at 15m candles and determine visually where you trade off that?
 
hilmy83 said:
What do you mean backtesting manually by hand? Like looking at 15m candles and determine visually where you trade off that?
More...
Ya.

Taking care to make entry/exit points conservative to allow for stuff. CL is pretty liquid during my trading hours which also helps with the gap between test and live.
 
1.10 is too low and 2.00 is suspiciously high so 1.25-1.75 is realistic

probably in real world yours will be more closer to 1.25-1.35 with commissions etc.
 
upload_2023-10-24_23-6-50.png


added second trade for today and double checked my numbers, added premove ticks+time and see some value there. tomorrow will add more.
 
Are you trying people to feel compassionate about you so they can help with the backtest? Good technique, show how painful it is, to do that by hand, and someone will step in.

Post the dataset in a CSV format and tell us what you are trying to test. I might be able to get it done quickly.
 
Drawdown Addict said:
Are you trying people to feel compassionate about you so they can help with the backtest? Good technique, show how painful it is, to do that by hand, and someone will step in.

Post the dataset in a CSV format and tell us what you are trying to test. I might be able to get it done quickly.
More...
just looking for advice. useful metrics to add, strengths/weaknesses of the system, where to look for improvement based on the numbers, etc...
 
You must log in or register to reply here.
Back
Top