If you did a backtest with continuous futures contracts, it is really necessary to roll when your data provider rolls. Otherwise, your backtest results will be different than your live results. I use Tradestation, and I have things set up so that I know when they roll the continuous contract, so I can too. Because of different continuous rollover dates with different data providers, I have seen great strategies in Tradestation fall apart in Multicharts, and vice versa. Of course, both have the same code, so the root cause is the data.
You are correct, the continuous contract data is really dangerous to use. I performed all back testing on the actual contract I would have traded at the time to try to prevent the continuous contract erroneous data point errors.
Week Jan 29th – Feb 2nd Back to red and “The Grind” continues. Seems like this week was a “So close, no cigar” type of week. Several trades just clipping stop loss levels to close out losers, then reversing in the direction my trades needed. This includes a NQ stop on the big drop Friday. If the same trade was made in the ES or YM, the short signal trade at the exact same time holds without getting stopped and end ups a nice winner. But that is all randomization, there will be also weeks where several trades barely miss getting stopped out and push the weekly P & L green instead of red. 2 wining days, 3 losing days. 17 winning trades, 21 losing trades with 17 full stop losses. Whereas I attributed using strict stop losses as a primary driver for last week’s profits, they showed their flip side this week by getting me out of some trades just a little too early before continuation moves occurred. Everything has positive and negatives. You take the bad with the good. NAV: -$2,556.49 Adherence to outlined trading strategy: 100%. I’ve made two early adjustments to the system since live trading started in mid-December (changing HE & LE entry to 8:35AM and using the -3.0 stop loss on KC). In my analysis post this week I’ll follow up with some additional thoughts on looking for discretionary tweaks that can be put on to improve performance. Maybe some other traders can provide their experiences on making tweaks and adjustments as they trade certain methods / strategies.
Making Discretionary Tweaks to Systematic Strategies My trading of this strategy so far has been mostly rigid in terms of sticking to the model’s trading calls and the stop levels used. The two changes I have made (changing HE & LE entry to 8:35AM and using the -3.0 stop loss on KC) were confirmed by math over two years (net impact to expected P & L performance). But I would like to think that as one gains experience, knowledge and expertise, it will be feasible to make slight discretionary adjustments to improve the odds of success, while still keeping the same core trade premise. That being said, here are a few discretionary “tweaks” that could be made as I get more experience interpreting the markets I’m trading. Adjusting Stop Loss Levels to Match Existing Volatility I track a series of stop loss levels with each instrument against the 4 main trading signal variants I have developed. They basically are: · 1 contract, $1,000 stop loss · 1 contract, $500 stop loss · 2 contracts, $300 stop loss per contract with $600 total stop loss. For NQ these different stop losses translate to: · 1 contract, -50 points · 1 contract, -25 points · 2 contracts, -15 points each, -30 total points I’m currently trading the 1 contract, -25 point stop approach as it has a slightly better positive expectancy over the past 2 years. But for different time periods over that 2 year stretch, the 1 contract, -50 point stop has traded better (higher volatility periods) and the 2 contract, -15 point stop each has traded better (low volatility periods). Recently the 2 contract, 15 point stop approach outperformed from Jan 2nd – 26th during the market melt up. Unsurprisingly as volatility bumped up this week, the 1 contract -50 point stop approach performed best as its wider stop traded the volatility better. I would think tweaking the stop approach to fit market conditions would be something that can translate into better profits. Another example regarding CL. In 2015 using a 1 contract -$1000 stop loss was by far the best performer. In 2016, the 1 contract, $1,000 stop and a 2 contract $500 stop each contract performed about even. In 2017 to present, the 2 contract $500 stop each contract performed the best. My guess this is attributable to the declining volatility in CL over the past three years. Swapping Instruments in the Basket Secondly, @southall replied to my consideration for changing the KC stop that I might consider changing products to trade due to KC increased slippage. I’ve back tested other “softs” and the grains with decent back tested returns in SB, CC, and ZW. Along with additional slippage in KC, it has been in an uninspiring trading range between 120.00 and 128.00 since October. So the question begs should I replace KC with something else trading with more volatility right now? Here again I think gaining experience with these products every day will enable me to pick and choose which instruments to trade in a given sector. Other softs and grains that have higher volatility than KC right now are performing better. Anyone want to share their thoughts with making minor discretionary tweaks to their trading systems / strategies for taking single day trades or 1 – 3 day swing trades? How long did it take for you to be “in tune” with an individual market to make stop loss level tweaks or swap instruments you were trading to something else, but still trading your core premise / strategy?
A fixed dollar stop across contracts with wildly different notional values and volatilities seems inaccurate to me. You might try using a percentage of of each contracts margin requirement which encompasses both values and has the added benefit of being already calculated by the exchanges.
Then to take this a step further, use a percentage of each contract as you suggest and then apply a volatility factor for each contract in some fashion to help determine the stop loss level. Is anyone out there doing this? If so, how are you determining the volatility factor and applying it in the calculation? I assume in your suggestion to use the margin requirement that exchanges already have a volatility factor included.
You can determine volatility factor by looking into recent historic vol (say 20 day), ATM option vol or even ATR (average true range)
Switching from a fixed distance to a volatility based measure for stop distance is not a minor tweak, this will have a major effect on your back testing results. You will also want to test more than two years as you will want to see how it worked during previous periods of high volatility.