I didn't do any walk forward testing on this 8 futures bucket, but I did live trade a 4 future bucket that included CL, GC, NQ, and ZN all with $1,000 stops from July - November, 2017. I was in a sense doing some walk forward on 3 instruments and the 4 variants of my studies. During this time on the first 4 futures bucket, my P & L reached +$35K. At that time I doubled my positions. The account then reached +$42K, but hit a draw period of -$7,500 per single position bucket, so as I was trading double it resulted in a -$15,000 draw down on the account. Not out of line for past history for that strategy, but the larger P & L swing was a more difficult occurrence to mentally trade through. Caused some "sweaty bum time" as Sir Alex Ferguson would say. I ended up doubting the my 4 future bucket systematic approach at times after that -$15K draw, so I began making discretionary reversals of trades where i "felt" I should reverse trades against the systematic long - short decision tree. I did this on 10 trades over 4 weeks. 2 of those worked in my favor, 8 of those worked against me were winning systematic trades that I turned into losers. 80% of my discretionary changes were wrong. The account ended up dropping another -$17K based on changing by those trades on my impulses. Lesson learned. If you commit to a systematic approach, don't outsmart yourself if you are not a discretionary trader. In the long run that slip up may turn out as a positive because it led to changes. I dropped ZN as it wasn't performing well enough under any of my 4 strategy variants. I studied a lot more additional futures and commodities to see if there were better options than ZN. I explored in more depth different trade stop loss levels for each instrument - $1000, $500, and $250. I explored the idea that adding additional instruments in an attempt to smooth out the P & L curve and/or make it more advantageous by showing similar draw down levels combined with better up run periods. So that is the back story to how I started live trading this 8 futures bucket.
I can't find your stats (was Sharpe Ratio in there?), but 5 months is not statistically significant as a testing interval unless, so it's possible that your approach works but you pulled out too early.
Yes, that could be true. It also surely makes it hard to tell if a strategy will work when you reverse the calls turning what should be winners into losers!
Think of it this way - if you are running a strategy with a Sharpe of 1, you have a real chance of having a down year and still be within the strategy parameters. That’s even without messing up or overriding anything
Week Jan 8th – Jan 12th Basically flat this week. Monday started with a winning day that was the 10th in a row going back to December 22nd. That tied the record for most consecutive winning days for back tested results beginning in 2016. Tuesday was a losing day, so it failed to set a new record. Consider it lucky to begin trading this new basket during an outlier, positive run. 3 wining days, 2 losing days. 19 winning trades, 21 losing trades. 15 of the losing trades were full stop losses. NAV: +$148.25 Adherence to outlined trading strategy: 100% - Sort of. I have been a little unnerved at the rapid price movements in livestock futures LH and LE after the 8:30AM opening print. I intentionally changed by entry time for these instruments to 8:35 AM this week and set the $500 stop loss exits from the 8:35 print. My actual trades for those this week resulted in a loss, -$817.60. The “system” trades from the 8:30 AM open for these instruments showed a gain of +$670.00. So moving the trade opening to 8:35AM resulted in a net NAV loss of -$1,487.60 from my expected strategy result. This weekend I will run all the variants using an 8:35 AM open for LH and LE to give me proper comparisons to the 8:30 AM opening results. Then I will let math decide which open to use. If one of the variants at the 8:35AM open sees results within 10% of the 8:30 AM results, I will switch to 8:35 AM opens. Otherwise I will revert back to the 8:30 AM opening trade entry and learn to live opening bell volatility in LH and LE.
That is why the "sort of" is posted along with it. I made the decision to live trade the 8:35 AM entry for livestock instruments last weekend, so I could experience live fire difference between the two entry points. I executed the intentional change consistently all 5 trading days. Then posted details of the change, genesis for trying it, results, and next steps in the journal log. I would consider a failure making an unintentional misstep like misinterpreting a trade signal, placing a wrong order entry trade or misplacing a stop trade resulting in missing expected system trade results. See the Week Jan 2nd – Jan 5th week log. It shows an unintentional mistake in strategy adherence along with an explanation.
Great work MidwesternTrader and thanks for posting. Few questions please. 1. Is each instrument you trade uncorrelated? 2. What do you mean by variants? 3. So, you run a total different system on each instrument? 4. Are your systems running continuously regardless of market conditions, or do you stop the system when you suspect or analyze the system and market condition does not match? I could imagine the former is challenging to do. On the other hand, day trading market conditions are unique and always changing. I am interested in your response here. Thanks,
No problem Simple. 1. The purpose of the 8 future basket was to trade a mix of products to provide some degree of uncorrelated trading. There are some exceptions - CL and HO are almost 100% correlated. When CL goes up or down, HO almost always follows because it takes crude oil to make heating oil. Gasoline is also nearly 100% correlated to oil for the same reason. Here is a website that tracks correlations. http://www.mrci.com/special/correl.php The Oil based futures are all highly correlated for obvious reasons. But this web page also shows in the past 180 days that the Oil products are highly correlated to the equity markets. In this case, the correlation exists because Oil and Equities have both been in a bull market the last 180 days. Probably more of a coincidence than a root correlation, IMO. 2. Variants are slight wrinkles in my core trading system. It is a trend continuation predicting, reversal signaling strategy. One of the things that can be adjusted is the sensitivity for signaling a trend reversal. Each market moves to its own rhythm. The exact same method that works with one instrument may not be appropriate for the another based on how the underlyings traditionally trend over time. Take a look at Gold (GC). Trends tend to be long lasting as GC is driven primarily by longer term FX moves in the US Dollar. Now let's look at Natural Gas (NG) over the same time period and see why it is called the widow maker. I use a variant that has a quicker trend reversal signal on Natural Gas. I use a different variant on Gold that requires a longer, more sustained move to trigger a reversal signal. 3. The variants are not different systems per se, just have different sensitivities to trigger trend reversal calls. 4. I do not stop making trades if one of the instruments or one variant is performing poorly for a stretch of trades. If you stop trading after a down run in a particular market, then you will be miss the trades that get things back on track. I don't want to miss those trades, so I grind through bad stretches. As I track all 4 variants on each instrument, if a different variant begins outperforming the one I am using over several months, then I may switch to that variant. But I am generally more concerned that I am using the right stop loss levels. I track different stop levels with each variant to give me a feel for how the market is moving each day. For instance, CL price volatility has reduced in 2017, so CL trades using a .50 stop loss are now performing about the same as trades using a $1.00 stop loss. In 2015 and 2016 my system showed much better performance using the wider $1.00 stop loss as CL's daily price movements where much stronger and two-sided. In 2017 that price action has tightened up and a .50 stop loss would is performing just as well as the $1.00 for the first time in several years.
@MidwesternTrader Great insight in your thinking process and how you trade. You do a lot of analysis. I guess that you have automated (most of) that? Otherwise you would get crazy from calculating all the numbers each day for all the instruments you follow.