sharpe ratio

[trend2009]

33% losing years, and Sharpe of 2.5, something doesn't seem right.

the losing year is very small. that is why 33% losing years still have 2.5 sharpe. This strategy lost in 5 bull years, but a small loss. as below:

year stat:200300 net gain:-4037.7
year stat:200400 net gain:36832.4
year stat:200500 net gain:-5290.0
year stat:200600 net gain:-2185
year stat:200700 net gain:38319.8
year stat:200800 net gain:48684.3
year stat:200900 net gain:91822.8
year stat:201000 net gain:-24124.
year stat:201100 net gain:168564
year stat:201200 net gain:-40061.
year stat:201300 net gain:20314.9
year stat:201400 net gain:56301.5
year stat:201500 net gain:54321
year stat:201600 net gain:77223.8
year stat:201700 net gain:3754.74
year stat:201800 net gain:144918
year stat:201900 net gain:45620
year stat:202000 net gain:200510
year stat:202100 net gain:54295.4
year stat:202200 net gain:153242

 

[spy]

33% losing years, and sharpe of 2.5, something doesn't feel right.

Yeah, those years could have had small losses I suppose. Or, could be a over-fitted backtest. Over-fitting will show huge numbers. It's just one of many metrics and if your account isn't exceptionally large I don't think 2.5 is unreasonable.

Of course the best metric is with a decent sized account using real money. Even then, I think things are more fragile than we like to believe and so I wouldn't harbor expectations. The chance of getting hit by a bus, no matter how small, is still there. That's just life.

 

[Businessman]

the losing year is very small. that is why 33% losing years still have 2.5 sharpe. This strategy lost in 5 bull years, but a small loss. as below:

year stat:200300 net gain:-4037.7
year stat:200400 net gain:36832.4
year stat:200500 net gain:-5290.0
year stat:200600 net gain:-2185
year stat:200700 net gain:38319.8
year stat:200800 net gain:48684.3
year stat:200900 net gain:91822.8
year stat:201000 net gain:-24124.
year stat:201100 net gain:168564
year stat:201200 net gain:-40061.
year stat:201300 net gain:20314.9
year stat:201400 net gain:56301.5
year stat:201500 net gain:54321
year stat:201600 net gain:77223.8
year stat:201700 net gain:3754.74
year stat:201800 net gain:144918
year stat:201900 net gain:45620
year stat:202000 net gain:200510
year stat:202100 net gain:54295.4
year stat:202200 net gain:153242

That is not sharpe of 2.5.

I am seeing average 0.85 sharpe, average about 55000 stdev about 66000

 

[Businessman]

Perhaps you are calculating sortino ratio?

 

[trend2009]

That is not sharpe of 2.5.

I am seeing average 0.85 sharpe, average about 55000 stdev about 66000

I calculated sharpe based on daily return of the last 20 years, then annualized it, that gives me 2.5.

 

[trend2009]

Yeah, those years could have had small losses I suppose. Or, could be a over-fitted backtest. Over-fitting will show huge numbers. It's just one of many metrics and if your account isn't exceptionally large I don't think 2.5 is unreasonable.

Of course the best metric is with a decent sized account using real money. Even then, I think things are more fragile than we like to believe and so I wouldn't harbor expectations. The chance of getting hit by a bus, no matter how small, is still there. That's just life.

I only have 1-2 parameters, with 200k trades. That is not curve fitting

 

[Businessman]

I calculated sharpe based on days of the last 20 years, then annualized it, that gives me 2.5.

You have enough yearly data, to use the years.

Annualised daily data gives an estimate for when you dont have enough years.

 

[spy]

I only have 1-2 parameters, with 200k trades. That is not curve fitting

No need to be defensive. I didn't accuse you of anything, just commenting on possibilities. How am I supposed to know what you did, or didn't, do?

 

[trend2009]

You have enough yearly data, to use the years.

Annualised daily data gives and estimate for when you dont have enough years.

So what would Wall Street think about this strategy?

 

[trend2009]

No need to be defensive. I didn't accuse you of anything, just commenting on possibilities. How am I supposed to know what you did, or didn't, do?

I know. I just clarify.