Back to Community
High Sharpe Weekly Strategy

The attached back-test combines the weekly EIA Crude Oil and Natural Gas Inventory reports.

Each Position is opened under different parameters of both long and short

  • Long ETF: when reported inventory is calculated as "bullish"
  • Short ETF: when reported inventory is calculated as "bearish"

Close position with a trailing stop loss

Both the stop and profit targets are determined based upon proprietary volatility models and are treated unique to each week

This model is applied towards derivatives and futures to maximize returns while simultaneously diminishing risk

Clone Algorithm
Backtest from to with initial capital
Total Returns
Max Drawdown
Benchmark Returns
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 571421e723815d0f4d957be0
There was a runtime error.
7 responses

Thank you for sharing. How do you define bullish and bearish inventories? Can you post the ultra.csv? Looking forward to learning more. I have a commodities trading friend that has been trading reports similar to this for 10+ yrs. I never considered it w/ quantopian. But I think it's a good idea.

pretty awesome risk profile. How can we mimic the CVS?

Agreed it looks great but in the end hard to judge if the real input isn't visible.

I would also like to see the csv to make sure you aren't introducing any lookahead bugs.. I unfortunately assume that is the case as the curve looks too good to be true for simple inventory system.

Hey John,

Looks good. Wondering how it performed at the different ratios for profit taking and stop loss that you tested?



The input trading signal is simply the EIA's weekly crude oil and natural gas inventory reports on Wednesdays and Thursdays, respectively.

No forward looking bias was instituted; it's important to not tailor strategies to backtests - the above mathematical levels and profit taking ratios are determined beforehand and thus are appropriated for all trades.

The interesting thing is this is just a visualization of a backtest using Quantopian. The real treat is the system that can exit at a the best statistically relevant Gaussian level. Remember, speed is the most important thing here, you can have the best model but without the technology to support it, it's no use.

Multiple things are important here. A: We still have no clue what you do in your .csv and how you get to values in it. B: You state speed is important...question is how important? Around the numbers markets are super jumpy and the whole world is eying for trades. So if you need the very best entry yes destined to be failing. But if you have some room for entry could be fine.