Algorithm Performance: 08/11/2022
Performance Rankings
GoodVibrations: +0.55%
Kokomo: +0.41%
Trading Bot: -0.01%
FeelGoodInc: -0.26%
Long Term Portfolio: -0.32%
The Neural Net: -0.34%
Base Algorithm: -0.67%
The Market: -0.71%
What Happened And Why?
Here’s the good news: the hedgers we published did great today. Both of them beat the market by more than a percent. It sucks that the Base Algo lost money today, but looking at the long term portfolio as well as sector returns, I can understand why it happened.
The only sectors that were net positive intraday are Energies and (only by a little) Financials. We try to minimize energy-based tickers in the algorithm, since they’re so news-dependent. Since our algorithms can’t account for news in their recommendations, it would be a major liability for us. Unfortunately, that leaves virtually no way for our non-hedging algorithms to turn a profit today. No matter which sector they leaned into, they were looking at more negative options than positive ones.
The piece that’s out of place here is FeelGoodInc. As a hedging system, it had plenty of opportunity to turn a profit today. Its losses along with the Base suggest to me a modeling failure today. Upon further investigation . . . that’s exactly what happened today. I reviewed our trading signals and found that the signals GV and Kokomo used to turn a profit were the same ones that led FGI and the Base to take losses.
Tentatively, this seems to be a fluke. I haven’t finished a full backtest of each trading signal using transformed vs raw price information, but it’s looking like modeling with raw price wins most of the time. It seems like days of this kind happen sometimes (where modeling with raw price information performs worse than when it’s transformed), but the opposite occurs more often.
So in summary, why didn’t the Base and FGI do well today? Because the trading signals they use broke mostly established patterns. There isn’t much we can do to prevent this; any machine-learning system depends on historical patterns remaining useful.
I do want to note that, while FGI came in negative today, it still beat both the market and the long-term portfolio we feed it. I still consider today to be, all in all, another win for it.
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:
That’s all for tonight. Thank you all for reading. Congrats to everyone who made money today, best of luck tomorrow.
-Asher
