c**t
发帖数: 9197 | 1 a system has win rate 80%.
Another has 90%.
which one to use?
Well it depends. sometime the 90% system generate a lot less signals.
so for the same time period, you might end up with more money with the 80%
system.
Note that a 100% win rate system will never generate any signal.
Since it does not exist.
I think there is some kind of curve here. and 80% seems a sweet spot. |
T*********s
发帖数: 17839 | 2 we need know the average and sigma |
E**O
发帖数: 1980 | 3 那种backtest，acount大起大落的winning rate高也不能要。
【在 c**t 的大作中提到】
: a system has win rate 80%.
: Another has 90%.
: which one to use?
: Well it depends. sometime the 90% system generate a lot less signals.
: so for the same time period, you might end up with more money with the 80%
: system.
: Note that a 100% win rate system will never generate any signal.
: Since it does not exist.
: I think there is some kind of curve here. and 80% seems a sweet spot.
|
w******s
发帖数: 16209 | 4 i think a simple win rate won't be helpful.
if you know when you win you could make 100 points gain and the other is 2
points gain,...
there are some standard to evaluate the system..
【在 c**t 的大作中提到】
: a system has win rate 80%.
: Another has 90%.
: which one to use?
: Well it depends. sometime the 90% system generate a lot less signals.
: so for the same time period, you might end up with more money with the 80%
: system.
: Note that a 100% win rate system will never generate any signal.
: Since it does not exist.
: I think there is some kind of curve here. and 80% seems a sweet spot.
|
C****a
发帖数: 1639 | 5 Use sharpe ratio or sortino ratio to measure the performance.
Yes, you need a trade-off between sensitivity and specificity.
【在 c**t 的大作中提到】
: a system has win rate 80%.
: Another has 90%.
: which one to use?
: Well it depends. sometime the 90% system generate a lot less signals.
: so for the same time period, you might end up with more money with the 80%
: system.
: Note that a 100% win rate system will never generate any signal.
: Since it does not exist.
: I think there is some kind of curve here. and 80% seems a sweet spot.
|
y*****l
发帖数: 5997 | 6 System A:
Percent profitable: 67.90%
Risk adjusted ann. return: 49.65%
Avg. trade (win & loss): 401.45
Ratio avg win/avg loss: 1.7
Profit factor: 3.6
System B:
Percent profitable: 97.10%
Risk adjusted ann. return: 38.57%
Avg. trade (win & loss): 2484.66
Ratio avg win/avg loss: 5.69
Profit factor: 190.66
Which one is better?
【在 c**t 的大作中提到】
: a system has win rate 80%.
: Another has 90%.
: which one to use?
: Well it depends. sometime the 90% system generate a lot less signals.
: so for the same time period, you might end up with more money with the 80%
: system.
: Note that a 100% win rate system will never generate any signal.
: Since it does not exist.
: I think there is some kind of curve here. and 80% seems a sweet spot.
|
