Algorithm Performance: 08/23/2022
Performance Rankings
Sector Neutral: +0.21%
Experimental Sector Neutral: +0.16%
Experimental Market Neutral: +0.01%
Base Algorithm: -0.06%
The Market: -0.13%
Long Term Portfolio: -0.14%
Market Neutral: -0.28%
What Happened And Why?
For the most part: what happened was exactly what we expected. We’re generally bearish this week, so we aren’t surprised that the market, our long term portfolio, and our bullish system are all down today. Most of our hedgers performed well, in about the order we expect to see (we expect our Experimental Sector Neutral to typically outperform the non-experimental, but as this is by a small margin we aren’t giving it too much weight). The only real surprise here is our non-experimental market neutral system. It was our worst performer, but isn’t it supposed to hedge against market movements? Let’s look into it.
The first reason I can see is that market hedging wasn’t effective today. Sure, the market was down today, but not by much. Sector exposure was much more important today. Even in our base algorithm, this was the case; it lost 8x as much from its Healthcare sector exposure as from its overall market beta.
Given how much worse it did than the experimental market neutral algorithm, I’m tempted to believe another problem was modeling. The modeling problem today however, is more unique. Looking at the Base Market Neutral’s biggest losers: AEP and NRG, Market Neutral didn’t identify many more signals for these than Experimental Market Neutral. Despite this, it allocated 57% and 71% more into each (respectively). This is because Experimental Market Neutral identified more positive signals on other tickers, diluting their share of its portfolio. So today, it isn’t that Experimental Market Neutral was better at avoiding losers. It’s that it was better at finding winners.
In short, the main reasons that Market Neutral fell behind today are:
Its modeling wasn’t as strong as Experimental Market Neutral
Market hedging was ineffective today. Sector performance was highly varied, whereas the market didn’t move much.
Crypto Trading - Update
Initial testing on a crypto algorithm has been promising. We took a list of 20 highly-traded cryptos (and 3 others, used solely for hedging), and backtested a system from August 2021 to August 2022. Over that time period, the crypto portfolio on its own had a Sharpe ratio of -0.297, but our algorithm had a Sharpe of 1.443: an alpha of more than 1.7!
This doesn’t mean we’re ready to roll it out - far from it. There are a few hurdles we still need to overcome:
Slippage and gas costs. Our initial testing does not account for either of these factors - neither do backtests of our existing algorithms. But with cryptocurrency, these are much more significant and volatile. We want to either find a way to minimize these, or find platforms that less us avoid gas costs. While slippage can be approximated using historical quotes, historical gas costs are more complicated. We would need historical records for every crypto in our test portfolio, which brings me to our next point.
Number of cryptos. Our algorithms work best with a large number of tradable products. The more we have, the more the law of large numbers can kick in and keep our returns consistent. In order to run this, we would want at least 20 cryptocurrencies we can be bullish on for the near future. This is excluding stablecoins and any of the cryptos we use for hedging. All in all, this becomes a tall order. If you have any cryptos to suggest for such a portfolio, reach out to the HK team and let us know. We’d love your input on this.
That is to say, an automated crypto system is in development, but won’t be rolled out in the near future.
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. Congrats to everyone who made money today. Let’s keep it going tomorrow.
-Asher
