Weekly Algorithm Review: 10/14/2023 to 10/20/2023
Performance Rankings
Variable Sector Neutral: +0.56%
Sector Neutral: +0.41%
Market Neutral: -0.09%
Variable Market Neutral: -0.13%
Long Term Portfolio: -1.64%
Overall Market: -1.84%
Base Algorithm: -2.09%
This was a weaker 2nd week for the beta. Our beta-1 portfolio is continuing to outperform, defeating both the S&P 500 and an equal-weighted ETF this week. We’re very happy to see this. The algorithm on the other hand, under-performed the portfolio by 45 bps. This isn’t a huge loss, but it’s certainly not a great sign. Regardless, this kind of performance is safely within our expectations of the algorithm.
We intend to maintain the beta release for at least another week or two before deciding if we want to move into a full release, or scrap this for an alternate version of the algorithm.
Sector allocations immediately reveal a reason for the algorithm’s under-performance this week. It leaned into Tech and Consumer Discretionary, at the expense of Financials, Healthcare, and Consumer Staples. Leaning out of Healthcare and Financials were perfectly fine moves this week, but its more bullish choices came back to bite it: Technologies and Consumer Discretionary both did poorly, while Consumer Staples was one of the best sectors to hold this week. In short, it looks like the reason the algorithm under-performed was that it forecasted a more bullish week. As it turned out, this was a better weak to lean defensive in your holdings.
Looking at the ticker distribution, there’s nothing too notable. The distribution is flatter than that of the long term portfolio, which we want to see, but it lacks a lot of the bigger winners from our portfolio this week. This obviously isn’t good to see, but it’s somewhat reassuring. There isn’t 1 losing ticker the algorithm kept doubling down on, or a huge number of winners it completely missed. Rather, it looks like a general under-allocation into winning stocks, backing up my take that the algorithm failed this week primarily because it leaned bullish - and the market very much did not.
What’s In The Pipeline
Development on the daily algorithm will mostly be pausing for the near future. I’ve laid some ground work for more effective, and efficient research on intra-day models to replace our existing bot. We’ve wanted to upgrade this for some time, and this will be a big step in our ability to do so. As discussed previously, in the long run, we expect more consistency from a system that trades five minute candles versus one that trades daily candles. In any case, we intend to continue the beta for at least two more weeks, unless this coming one is particularly noteworthy. The results of the beta will aid development of the daily algorithm, so there’s some additional incentive for us to pivot to other research until then.
We intend to continue the beta in the meantime. This week’s performance was unimpressive, but safely within our expectations. Should negative performance continue, we’ll be more inclined to scrap this algorithm and move to another version instead. As is, one week in the black and one in the red isn’t enough that we want to pull this algorithm.
Misc. Data For The Week
