Weekly Algorithm Review: 11/11/2023 to 11/17/2023
Performance Rankings
Overall Market: +1.01%
Long Term Portfolio: +0.94%
Base Algorithm: +0.84%
Variable Market Neutral: +0.31%
Variable Sector Neutral: +0.31%
Market Neutral: +0.03%
Sector Neutral: -0.08%
This was a weaker week for our portfolio. Its under-performance of the market is pretty slight, but this was a strong week for smaller cap stocks - so we expect better out of it. Compared to an equal-weighted ETF, the portfolio under-performed by a larger margin. Unfortunate, but it’s just one week of a fairly long-term portfolio strategy. We expect weeks like this to happen to it.
The algorithm under-performed the portfolio by 10 bps, which is essentially in the black. This isn’t really a bad week for it, but after last week we said it would need to do fairly well to justify a full release. It’s not bad performance, but it definitely isn’t the “fairly good” we’re looking for.
The distribution of returns by ticker this week is much closer to what we want to see. The distribution of return by ticker is pretty flat relative to the portfolio. We’d obviously like a larger portion to be green, but this is a step in the right direction.
In terms of sector allocations, we don’t see a lot of big moves from the algorithm this week - part of the reason its final performance is so close to that of the portfolio. The algorithm has, naturally, made some large allocations this week, but in terms of overall sector distributions, it ended things pretty close to our overall portfolio.
All in all, this a worse week than what we were hoping to see. The portfolio finished roughly in the black, but following the major losses we saw last week, neutral performance isn’t enough to impress us. It’s a drastic improvement, and still definitely worth running the beta for another week or two. But we’re largely in the same place we were at the end of last week. If we some strong performance from here, the algorithm may be worth a full release. If not, we’ll likely follow through on our tentative plan to replace it with a reworked intra-day algorithm. Speaking of:
What’s In The Pipeline?
Work has been focused on getting the new intra-day bot online for its internal beta. Tentatively, things look promising. We’ve identified a few problems with our previous intra-day beta, which we think will lead to a stronger showing here. The first of these is an issue with our previous data provider. The current intra-day bot we have released to users doesn’t require 5-minute candles to be streamed during the day - just live prices. All of our new models, however, require live 5 minute candles. This requires us to use a different data source than usual. Previously, we’ve used Yahoo Finance for this, but it hasn’t been particularly reliable. We’ve seen it give inaccurate data a few times (or at least, data that doesn’t line up with data shown on Tasty), leading to it giving false signals. We’ve switched to Polygon - and so far, haven’t seen any inaccurate signals.
We’ve also done some work to speed up calculations. We’re expecting buy/sell signals to come out within 2 minutes of the end of each 5 minute candle. We’re looking to cut this in half, but for the sake of our internal beta, we’re happy with this level of latency.
We’ll give updates on this as we get them, but currently it’s looking promising!
Misc. Data For The Week
