CL trading system, long-only Very simple entry & exit rules, 2 entry filters (price-action based), "catastrophic" stop of 250-ticks (hit about 20 times in backtest). 10.5 years of backtesting (from July 2003, as far back as I have tick-data for CL) ... 130 trades / year on average. Backtesting using 1-sec TF for order execution, 1-tick systematic slippage on entry & exit, $5 comms.
A couple iterations to refine the entries. Still very simple rules. The average yearly DD in MC sim (using that system's trade distribution) comes at 6k, w/ 2.5k stdev ... On the 10 years, 2 are just past 1 stdev (2012 & 2010), 4 years are ~at the mean, 3 years are about 1/2 stdev less than the mean, and one odd year - 2013 - is almost 2 stdev better than the mean (1.7k max DD in the entire 2013). This last aspect is a little concerning, as of course the next average or larger than average DD is just around the corner.
The dark side of CL ... Short only. Similar principles, entry/exit rules very simple. Kinda disappointing there isn't more opportunities for this setup (~60 / year on average). The max. historical drawdown is quite steep (about 1 year of P&L), but logically situated end of 2007 to mid 2008 - in the raging CL bubble.
Dom can you tell me where the slippage total is calculated on Ninja. Can't see slippage total on your screenshot?
On the right-hand side in the section Historical Fill Processing, the Slippage setting is #ticks of slippage per MKT / STP order ... for these backtest, I use 1-tick per order (all are either MKT or STP), and the FillType "BetterThanDefault" guarantees it is indeed 1-tick (the Ninja default FillType restricts the fill to the fill-bar H/L range, so in some cases you wouldn't have any slippage - which is why I created this "BetterThanDefault" fill-type).
Long & Short now in the same system. I fixed a couple minor bugs, and refined the short setup. Nice to see a lot of the Short side drawdowns "erased" by the Long side. The largest drawdown in 2010 barely exceeds the MonteCarlo mean max.DD for 10 years using that backtest trade distribution ... larger DD in the future are to be expected.
Sorry? 1 sec bar signals? And a 250 tick stop loss....?? I'm no expert, but thats insane. Why not test with much tighter stops....10 ticks, 20 ticks, 30 ticks.... etc. If you need a 250 tick "buffer" to maintain a positive expectancy system, then you've got one of those inverted risk-to-reward scalping systems that blow up over the long term. Thats what it sounds like.
Thanks for the feedback. I certainly agree that 250-ticks initial stop (about the average daily range for CL) is a lot. Average winning trade is 70-ticks (long) to 98-ticks (short) ... this is not a scalping system. All trades are 1 contract, no add-on. Trade decisions are made on 1-min TF, but trade execution is on 1-sec TF (for better accuracy of backtesting). As for stops ... I have yet to produce a CL trading system with a meaningful positive expectancy at 10-ticks initial stop. My experience is that getting out on stop is the worse exit strategy ... I prefer using the initial stop as "catastrophic" stop, and actively manage exits. "Catastrophic" might have different meaning for different people. I use 5 to 10% of the max. allowable drawdown for a system to define that level. For this one, I am using 24k as the max. allowable drawdown (roughly twice the max. historical DD, although this is just a confirmation datapoint, I use MC simulations to decide on the max. allowable drawdown), so 250-ticks is just past that 10% mark (v06 actually uses a 225-ticks initial stop). To answer your question directly, below is the performance table using various initial stop size (0 means no-stop). As you can see, reducing the initial stop size essentially reduces the overall win% and P&L, without having much impact on the max. historical drawdown.
Dom, is the system intraday only or is it more of a swing system. Do you think your entry has any positive expectancy considering it requires such large stops? Does this not mean your entry could be somewhat random and the system expectancy is due to exits? Do you not find it frustrating holding CL with large stops like this especially with the whipsaws and volatility.