Algorithm Performance: 2022-08-04
Performance Rankings
Trading Bot: +0.05%
The Market: -0.05%
Kokomo: -0.24%
Long Term Portfolio: -0.32%
GoodVibrations: -0.42%
The Neural Net: -0.48%
Base Algorithm: -0.61%
What Happened And Why?
I’ll be blunt, today was a bad day for us. The market was negative, and almost everything we have failed to out-perform it. Let’s dig in here, and see why.
Our returns by ticker tell a pretty clear story. The Base Algo put 11.2% its portfolio in $WMT (more than any other ticker), and with that going down almost 4% during the trading day, it was a major reason we lost. We lost similarly on $COST, allocating 10.34% of our portfolio (the 2nd most) but only went down 0.68% intraday. So far, the answer seems clear: we leaned hard into Consumer Staples/Consumer Defensive, and because that sector took a hit today, we took a bigger hit than the market.
The Neural Net lost for many of the same reasons. It’s losses by sector aren’t as bad, but this chart will look pretty familiar:
So why did we lean so hard into Consumer Defensive? What signals did our models find so appealing? I can’t dig too deep here (if I tell everyone exactly how our proprietary models work, Allen will kill me), but let’s speak generally for a second. Of the 32 trading signals our models see, 14 of them work well enough on $COST backtests, and gave buy signals last night. Of these 14 signals, 9 of them strongly interact with moving averages. With $WMT down after a string of bad news days, our algorithms have mistaken a company with bad news for a company that’s just oversold.
If we look at a quick chart of $WMT over the last month, we can see why the algorithms think this. The last time it spiked down this hard, it came back up for some clean profits. Our algorithms aren’t able to consult the news, and so they don’t think this is any different from last time. They’re expecting a similar rebound. Who knows, maybe they’re right and over the course of a week this report will age like milk. In the meantime, I’d be hesitant to hold these, as I expect a lot of our algorithms to keep putting some money into them.
But why didn’t Kokomo or GoodVibrations avoid that drop? They didn’t do well either today. The reason for this is that there isn’t a great way to hedge against ticker-specific news. Kokomo hedges its stocks by trading $SPY. $WMT has a daily market beta of about 0.586. Today however, $WMT didn’t move because of the market as a whole, it moved because of bad news exclusive to itself. Our market positions couldn’t have done anything to offset this.
GoodVibrations had the same problem. We mostly hedge $WMT by trading $XLP, with a beta of 1.143. Today, the Consumer Defensive sector barely moved (it was down about 0.44% intraday). Our hedge offset our losses here, but only barely. Kokomo can hedge against market movements, and GoodVibrations can hedge against sector-wide movements. But when 1 stock in particular does poorly, we don’t have any way to hedge against that. If we look at intraday sector values, Consumer Staples didn’t do that poorly. A negative day for sure, but not substantial.
Unfortunately, there isn’t much we can do here. We aren’t able to incorporate news analysis into our algorithms right now, and can’t just make it ignore technical signals that have done well in the past. We’re going to consider holding less $WMT than the algorithms recommend, and advise all of our users to check recent news on all tickers our systems recommend. Never make a trade without due diligence.
Tomorrow’s Outlook
The full algorithm reports will be published tomorrow morning, once Allen has had a chance to vet its recommendations. In the meantime, here are our tentative exposures for the trading day tomorrow:
No significant developments here, as per usual. We’ll highlight it if any of our algorithms substantially change their distributions.
That’s all for tonight. Let’s all hope for better results tomorrow. Thanks for reading, and good luck trading everybody.
-Asher