As someone who has performed many thousands of backtests and optimizations on nt7, I began to develop a short list of 'wants', little changes that I felt could've made the process as a whole more efficient, or 'telling', or useful. .
One of these is in regards to the statistical display after an optimization/backtest run. Formerly in nt7 this was under 'graphs', currently it is under 'analysis', where one can break down the data according to DAY OF or HOUR OF (or half hour of) entry, or exit, and see how well the trades were faring over narrower subsets of time.
However, all of the options within nt7, and this seems to hold true with nt8, only show you the total net profit/loss over these time frames. A *far* more useful statistic, in my opinion, might be things like the average trade profit, or profit factor, or sharpe, or any number of other categories, of the trades over these specific subsets of time.
Solely using net profit becomes deceiving, and I'll give just one of many examples as to why. . if I'm looking to gauge how well a strategy does over its entire historical period over different days of the week, and I bring up the 'days of week' graph and am confronted with numerous green/red bars, I may see that sunday has a much smaller green bar than those green bars for the other days of the week, and therefore I could walk away thinking that these trades performed worse on sunday.
HOWEVER, if I then look at a breakdown of how many trades were taken by this example strat, I'll notice that sunday took just 200 historical trades, compared to nearly ten times as many trades taken on all other days. . this means that sunday actually has a higher ROI for each individual trade relative to all other days, meaning the earlier graph was quite deceiving. . .
Just a thought, but I'd love to see one or two additional statistical categories in the 'graph' dropdown menu (avg trade profit would be the most informative, in my opinion).
I'm going to use this thread to continue pointing out changes, big and small, that I think could greatly improve the backtesting and optimization structure within NT8. . please let me know if this is either wasted effort (if the time for improvements has passed), or if this isnt the proper forum to do so.
Thanks
One of these is in regards to the statistical display after an optimization/backtest run. Formerly in nt7 this was under 'graphs', currently it is under 'analysis', where one can break down the data according to DAY OF or HOUR OF (or half hour of) entry, or exit, and see how well the trades were faring over narrower subsets of time.
However, all of the options within nt7, and this seems to hold true with nt8, only show you the total net profit/loss over these time frames. A *far* more useful statistic, in my opinion, might be things like the average trade profit, or profit factor, or sharpe, or any number of other categories, of the trades over these specific subsets of time.
Solely using net profit becomes deceiving, and I'll give just one of many examples as to why. . if I'm looking to gauge how well a strategy does over its entire historical period over different days of the week, and I bring up the 'days of week' graph and am confronted with numerous green/red bars, I may see that sunday has a much smaller green bar than those green bars for the other days of the week, and therefore I could walk away thinking that these trades performed worse on sunday.
HOWEVER, if I then look at a breakdown of how many trades were taken by this example strat, I'll notice that sunday took just 200 historical trades, compared to nearly ten times as many trades taken on all other days. . this means that sunday actually has a higher ROI for each individual trade relative to all other days, meaning the earlier graph was quite deceiving. . .
Just a thought, but I'd love to see one or two additional statistical categories in the 'graph' dropdown menu (avg trade profit would be the most informative, in my opinion).
I'm going to use this thread to continue pointing out changes, big and small, that I think could greatly improve the backtesting and optimization structure within NT8. . please let me know if this is either wasted effort (if the time for improvements has passed), or if this isnt the proper forum to do so.
Thanks
Comment