Hi folks, welcome back to another Market Mood *bonus letter*:

And if you’re new here, try out a few of our recent posts to get acquainted:

• Fed Says Boo 👻 | Market News Roundup (9/26)

• All Quiet on the Western Front 🌲| Market News Roundup  (9/19)

1. the past 6 months on the market = 🔴

How the market has performed:

Through the past six months, the US stock market (and economy at large) has had one of its roughest stretches of the past half-century — perhaps the worst of my 24 years around the Sun. During these last six months, the S&P 500 has fallen almost exactly -20% into and out of and back into official “bear market” territory, inflation has peaked at a 40-year high above 9% YoY growth, and the US economy has entered a “technical” recession with two straight quarters (& counting) of negative GDP growth, not to mention the international supply chain struggles and rising geopolitical strife — by most definitions, a truly historic time to be alive, so congrats for making it this far. During times like these, the natural question(s) become: how do we navigate? what does everyone think about the markets? what comes next? For us, we think the answer lies in analyzing the news en masse:

Reviewing news sentiment

Over these six months, we’ve been on a personal mission to reliably figure out which news is moving the markets. If you’re unfamiliar with the format of our weekly Market Mood roundup, here’s a quick outline of how it works: we scrape a few thousand news articles and posts from a variety of publishers. From here, each week we score the most bullish and bearish stocks in news coverage based on the portion of optimism vs. pessimism expressed about across each post (more on that here) then review and rate one bullish & one bearish stock to watch. Below we’ll analyze the subsequent performance of stocks we’ve rated.

2. our sentiment rating performance ✅

First, some updates to our sentiment analysis pipeline

Before we get into the performance and returns of the stocks we’ve reviewed recently, it feels important to say that we are still learning how to best quantify the mood of news coverage; we are not perfect (yet) — though we have come quite a ways. It also feels important to mention a few changes we’ve made to our news sentiment pipeline since we last wrote one of these retrospective reviews back in April (shown below). Most notably, we’ve begun adding social media posts from Twitter and Reddit into our span of news coverage. Second, we’ve also tweaked the way we detect mentions of stocks and how we score the “relevance” of articles to ultimately improve the accuracy of our sentiment rating system — visit www.babbl.dev to see for yourself.

Now, our rating performance & results:

With the table set, let’s dive into the pièce de résistance: how well our picks have done vs. the market. Looking back over the past 5+ months since early April, we’ve reviewed and assigned weekly sentiment ratings to 44 stocks. Of the 42 picks published before this week, our news sentiment rating has correctly predicted the stock’s direction over the following week exactly half of the time (21 out of 42) — which is a slight dip from our accuracy rating from our last retrospective in Q1 2022. Granted, the market was much choppier over the past 2 months, so as long as we don’t go below 50%, I’ll live with it, though definitely room for improvement.

Where we did improve from our previous review was in the magnitude of hits & misses — this time around, we tended to miss a bit smaller and win a bit bigger across the board (with the exception of two big misses that ended up skewing our overall performance a bit). Generally, it was a few fewer bogeys and a few more birdies, and on the whole, I’d call the picks a success: the stocks we reviewed returned a net +18% over the respective first week following our rating, shown below. Of each all our ratings, a few of our best & worst performances were:

Best picks:

• Coinbase (COIN) 8/15: bearish rating🔴, -22% the following week🔴

• Peloton (PTON) 9/19: bearish rating🔴, -17% the following week🔴

Worst picks:

• Snapchat (SNAP) 7/18: bullish rating🟢, -29% the following week🔴

• GameStop (GME) 8/1: bearish rating🔴, +25% the following week🟢

Full list of picks:

Similar to our previous review in April, the lion’s share of our outperformance vs. the market came from bearish picks, which makes sense given the bear market landscape. And with the benefit of hindsight, if we exclude our biggest miss (Snapchat bullish rating on 7/18), then our 1-week net return becomes +47% overall, which is a lot more respectable.

What if you were to invest solely based on our ratings — say, buy naked calls or puts on a stock according to our bullish / bearish ratings and hold them through today, just for giggles? While we would absolutely NOT recommend doing this (both because our sentiment algorithms are still in beta, and because the news and its corresponding sentiment tends to mostly be relevant through the following 7-14 days), our absolute net return across all 42 picks would be 153% through today. At a $100 investment into each pick on the Monday morning of each rating ($4200 total capital deployed), holding through today would equate to a total return of $6,426 — the data is included in this Google Sheet for reference

3. the road ahead 🛣️

Of course, there are always a few caveats here. First, as we’ve said, the markets have been especially horrible year-to-date, so the fact that roughly half of our picks have been bearish (betting on downward movement), you’d expect a solid return on these; though on the other hand, the magnitude of return is still notable. Second, there is always some degree of chance and luck in the markets (even a broken clock is right twice a day), so it’s still a bit early to write home and call it a day.

What is reassuring is that our ratings have managed to beat the markets over a second retrospective review session, and perhaps quite handsomely at that, with a considerable increase in sample size. But before we get ahead of ourselves, it’s important to remember the old saying:

Going forward, there is still plenty for us to improve on our sentiment framework to make our scoring of news coverage truly representative and actionable. A few things on our to-do list over the next few months (identified with some help from readers like you!):

1. adding weights for “credibility” or “importance” of individual posts from different sources / authors (after all, not all opinions are created equal) — we mentioned this last time, and sooner or later this will need to be a top priority

2. better accounting for complex language nuances like negation, sarcasm, and Twitter-speak (ie. using emojis and incomplete sentences; if you’ve ever seen a Cam Newton Instagram post, you know what we mean — a fun example of this here)

3. a higher quantity of explicit sentiment ratings over time. Simply put, we need more data, and people like you need more sentiment ratings for this to be useful. We’re working on automating this currently, but could take a while. The portion of correct picks is where we need to improve; I’m less worried about the magnitude of “correctness”.

These are just a few ideas, but there’s still a long road ahead for us. Do you have any ideas for how we could make this sentiment more reliable or useful for your day-to-day trading ritual? Let us know by commenting below or emailing us at [email protected] — talking with our readers might literally be my favorite thing in the world! Anyways, that’s all for today folks, thanks for reading. 🤗

Lastly, if you’ve made it this far, we’re still giving a \way a few more 100% FREE FOR LIFE discounts to our real-time stock market news sentiment tracking website, currently in Beta at www.babbl.dev. Just visit the site, create an account, and enter the promo code “FIRST250” to claim your full discount — it’s truly a win-win; you get a better bird’s eye view of market-moving news, and we get feedback to fix bugs and make the site better!