I developed an algorithm and I need a little help interpreting the results.
I tested/optimized the algorithm with 1-min data on the ES from 2007-2010. There are only 2 or 3 variables that are optimized (targets and stop losses). I then forward tested on the data from 2010 thru to the present.
I get the following results based on the last 3 years with of backtested trades:
Basically it makes lots of intraday momentum plays on the ES/NQ. 50% work out. But the winners are about twice as profitable as the losers. The returns are pretty consistent on a monthly basis over the last 6 years. The equity curve looks relatively stable even though the drops in 2008/9 and 2011.
But my sharpe is very low. And that's causing me some grief.
I've double checked that I don't have any data-snooping bias. I've included IB's commision rates. But no slippage since its ES. But at the same time, I ensure that the tick has crossed my limit price before I close the trade.
I initially tested this in Multicharts, but then wrote some code in java to also test it, and I get the same numbers.
So I've got a decent P/L, drawdowns I can live with, but a really shitty sharpe ratio, I'm guessing because the stdev of the earnings is large and 50% are losers.
I'm not done playing with the algorithm, but can anyone give me some insight into the sharpe ratio, and the other performance ratios? I've not developed algorithms professionally, so I don't have a basis to compare these results to - other than the sharpe should be much higher.
I tested/optimized the algorithm with 1-min data on the ES from 2007-2010. There are only 2 or 3 variables that are optimized (targets and stop losses). I then forward tested on the data from 2010 thru to the present.
I get the following results based on the last 3 years with of backtested trades:
- Sortino = 9+
- Sharpe = 0.7
- PF = 1.5
- Win/Loss = 1.8
- % Winning Trades ~50%
- Max Drawdown = 8%
- Annual RoR = 350%
Basically it makes lots of intraday momentum plays on the ES/NQ. 50% work out. But the winners are about twice as profitable as the losers. The returns are pretty consistent on a monthly basis over the last 6 years. The equity curve looks relatively stable even though the drops in 2008/9 and 2011.
But my sharpe is very low. And that's causing me some grief.
I've double checked that I don't have any data-snooping bias. I've included IB's commision rates. But no slippage since its ES. But at the same time, I ensure that the tick has crossed my limit price before I close the trade.
I initially tested this in Multicharts, but then wrote some code in java to also test it, and I get the same numbers.
So I've got a decent P/L, drawdowns I can live with, but a really shitty sharpe ratio, I'm guessing because the stdev of the earnings is large and 50% are losers.
I'm not done playing with the algorithm, but can anyone give me some insight into the sharpe ratio, and the other performance ratios? I've not developed algorithms professionally, so I don't have a basis to compare these results to - other than the sharpe should be much higher.