Some Background Information

Within the last few years, the advent of zero-commission trading and unprecedented accessibility to market knowledge through social media transformed the world of trading and investing. 2020-2021 were particularly unique and unprecedented years for investors, both new and old. While mature investors experienced volatility beyond their wildest dreams, newer investors were presented the buying opportunity of a lifetime following the S&P 500’s catastrophic 20% decline (in real dollars) caused by COVID crash from December to March. And they bought. In droves.

In fact, Schwab’s Generation Investor Study reveals that 15% of all U.S. retail traders began in 2020. Complementing Schwab’s analysis, Citadel Securities claims that up to 20% of all trading volume was attributed to retail on average. With 88% of these new investors self-reporting “average” or “lower-than-average” investment knowledge, this caused some interesting aberrations in the stock market that we’ll look to explore today.

We preface our analysis by introducing stock baskets. Customarily, stock baskets are securities that share a similar theme: the most notable amongst retail traders (we’ll use investors and traders interchangeably -- there is a distinction) dubbed the “meme basket” substantiated by GME, AMC, BB, etc.). The autocorrelation and subsequent uncoupling are well studied and understood. Each basket has some underlying mechanism that ties them together: in the aforementioned basket, aggressive buying pressure due to the influence of social media. However, the relationship between tickers that are effectively typos of one another (e.g. TSLA and TLSA) are not very well understood, at least in the retail space. In this analysis, we’ll investigate and summarize the following:

1. What is “typo trading” and why does it happen?
2. Does good or bad news have more impact on a sample of Levenshtein Basket stocks?
   • Tesla, Inc. (TSLA), Tiziana Life Sciences PLC - ADR (TLSA), Talaris Therapeutics (TALS)
   • Gamestop Corp. (GME) and Globus Medical, Inc. (GMED)
3. What is the correlation in price between similarly named stock tickers?
4. How long does it take for the "typo’d” stock to begin rallying and declining? How long does the mispricing window of opportunity last?
5. What is an example strategy to play this type of market distortion?

The mandatory legalese at the beginning of every financial article so I don’t get sued

Bread Cat Bounce IS NOT a financial advisor, tax professional, broker, or legal advisor. Bread Cat Bounce is just a cat (albeit a sentient and humorous one) that happens to be a researcher of financial opinions and debatably questionable educational content. All information, data, strategies, reports, and articles are provided for humorous and educational purposes only and should not be considered or inferred as personalized investment advice and is not intended to be, nor shall constitute, an offer to sell or solicit any offer to buy any security. Unfortunately, cats are also bound to certain US regulations that prohibit us from giving personalized investment advice or other advice whatsoever on a personal basis. Bread Cat Bounce does not accept any form of compensation whatsoever from companies or assets that they may research and write about, except Discord (which they aren’t allowed to do anyways). Bread Cat Bounce does not recommend or endorse any brokers, dealers, or investment advisors (especially those pesky fuckers). Bread Cat Bounce's reports, writings and other media releases are based on their own opinions, current news & events, interviews, corporate news & reports, SEC, SEDAR, other regulatory filings, and any other information learned from sources and experiences. Research may contain errors, and you should not make any financial decision based solely on what you read in Bread Cat Bounce’s reports and writings.

It’s your money, it’s your responsibility to perform your own due diligence, and you must make your own decisions. Be advised and aware that buying and selling financial instruments involves risk. Bread Cat Bounce accepts no liability whatsoever for any direct or consequential loss arising from any use of their writings, products, services, or other content. You are responsible for your own investment research and decisions. You should seek the advice of a qualified investment advisor and fully understand any and all risks before investing. Historical results of Bread Cat Bounce’s analysis are no guarantee of future results. All content and references to third-party sources is provided solely for convenience. This information may be inaccurate, use at your own risk.

TL;DR

What is “typo trading” and why does it happen?

• The Levenshtein Basket groups together stocks whose tickers or company names are commonly mistaken for each other, primarily by retail traders
• Once there is news released that pushes a retail stock higher, this buying pressure leaks into stocks within the Levenshtein basket that are typically and otherwise uncorrelated
• Smartphone trading encourages risky behavior which extends into unintentionally trading different securities (e.g. relatively big fingers on a small keyboard)
• ADRs, meant to simplify foreign investing for Americans, are oftentimes another layer of intricacy for entry-level investors
• Asian investors may also have difficulty differentiating between tickers, as their markets rely on numbers instead of the US market’s custom of letters for their stock tickers

Are the trades more reliable on the long or short side?

• It is likely that good news has more of an influence on the Levenshtein Basket than bad news does
• This doesn’t mean that bad news does not matter; adjacent stocks may still get hit due to the impact on the market overall which could trigger a multi-sector selloff

Is there a correlation in price between similarly named stock tickers? How long does the mispricing window of opportunity last?

For TSLA/TLSA:

• The win rate historically is roughly ~49% which basically means that unfortunately this strategy is actually than a coin flip over the long term
• The peak correlation on most days was found within the first half-hour of the trading day, which where we expect the greatest amount of volume traded to occur
• This correlation is likely to be the strongest due to pre-market action -- retail buyers will typically only trade during regular market hours; when they see TSLA extremely positive, they are likely to chase these gains (and likely to buy TLSA as well)
• The correlation drops heavily after the five minute window, implying that the majority of traders close out on this trade once they figure out their mistake

For GME/GMED:

• The win rate historically is roughly 74% with an average peak correlation of 62% (79% for winners) which means this is a potentially viable strategy
• The peak correlation (excluding days with zero correlation) is pretty consistent across all times of the trading day, but is more likely to happen towards the end of the day
• One would think as meme stocks lose relevance in social media and the news, the frequency of Typo Trading would go down; however, the occurrence is actually quite regular, which further lends itself to being a viable strategy in one’s trading arsenal

What is an example strategy to play this type of market distortion?

• Shares are your friend due to the inherently outsized volatility that is characteristic of low-liquidity stocks
• For TSLA/TLSA: If you are, for some reason, still dead set on trading this despite all of the warning signs, here is your expected payout structure:
• 30% -- TLSA will drop due to outside selling pressure; if this occurs despite TSLA increasing in price, you should look to exit immediately.
• 25% -- TLSA will remain flat and your loss will be modest/close to zero, primarily due to slippage fees and liquidity premium (since the secondary ticker is so illiquid).
• 45% -- The other (less than) half of the time, TLSA will pop up with TSLA. Here, there are a number of strategies you can employ to exit the position depending on your risk tolerance and trading goals!
• For GME/GMED: A possible (but certainly not optimal) strategy is to buy 30-delta calls at the nearest expiry in order to capture the most value through gamma ramping if GME explodes in value. Similar to TSLA, you have less than 5 minutes to enter this trade.

“Typo Trading” a.k.a. the Levenshtein Basket

Since “Typo Trading” doesn’t have the same convoluted and sesquipedalian characteristics as most finance jargon does for whatever reason (see what I did there), I found it apt to wholeheartedly embrace finance culture and name this phenomenon the “Levenshtein Basket” after the Levenshtein distance. Snarky jokes aside, the Levenshtein Basket aims to group together stocks whose tickers or company names are commonly mistaken for each other, primarily by retail traders. Without the risk management tools of today’s professional firms, trading neophytes are prone to purchasing stocks that are similar, such as mistaking an OTC network security company for the boomer’s version of Discord. There are a quite a number of instances where this happens on a news-worthy basis (courtesy of Business Insider and CNBC):

• October 2013: Tweeter Home Entertainment (TWTRQ) was halted multiple times after the stock rose ~1500% prior to Twitter's (TWTR) IPO
• February 2017: Snap Interactive "STVI" spikes 164% prior to Snapchat's IPO (SNAP)
• April 2019: Confusion over Zoom Technologies (ZOOM) and Zoom Video (ZM) after ZM goes IPO
• March 2020: Trading of ZOOM is paused for a week after ZM sees a parabolic share price increase due to COVID-19
• January 2021: Tweet from Elon Musk about "Signal App" pumped SIGL (Signal Advance)

From my (admittedly brief) investigation of news articles, this phenomenon caught the eyes of financial news outlets even all the way back to 2013. What’s particularly fascinating is that the reporting frequency of the Levenshtein Basket effect increased once retail trading became truly mainstream circa 2019 thanks to Robinhood. So, at least anecdotally, we can verify the existence of this phenomenon.

Stepping back a little, why is typo trading even a thing? Is there a particular sub-group of retail investors that are extremely susceptible to typo trading? There are a few theories.

Grammarly, a writing assistant app, found that their mobile device users make up to five times more mistakes as their desktop typers (which is around 42 mistakes every 100 words) with misspelled words representing the lion’s share of errors. Dyslexic traders may also contribute to this phenomenon: according to a study on developmental dyslexia, around 5% of the population have issues reading. Personally, I am particularly guilty of being groggy in the morning and making fat-fingered (or is it fat-pawed?) executions from simply being exhausted from shitposting the night prior. C'est la vie.

Still, there are other sources of typo trading. According to China Renaissance Capital’s head of macro and strategy research, Bruce Pang believes that the discrepancy between how US and Chinese markets function can add confusion. ADRs, meant to simplify foreign investing for Americans, are oftentimes another layer of intricacy for entry-level investors. Likewise, Asian investors may also have difficulty differentiating between tickers, as their markets rely on numbers for their stock tickers.

Assumptions

Before we quantify the impact of the Levenshtein Basket on the market, we need to first draw up some assumptions. This will help consolidate the size of the eventual report that comes out (the original investigation spanned many times the published version you see here), as well as partially formalize the scope of this analysis. We’ll assume a few things:

a) Stocks within the sub-basket (e.g. ZM, ZOOM, ZOM) will typically not trade in lockstep with the exception of major news that heavily impacts the primary stock.

If these stocks are heavily correlated outside of displaying the Levenshtein Basket phenomenon, then there is a high possibility of a different, more influential driver of this correlation. To eliminate confounding variables as much as possible, we’ll want to make sure that these stocks aren’t correlated most of the time. We’ll run a quick correlation analysis to verify this assumption in the later sections.

b) Any stock will have, on average, one pairing that is the most salient and has the strongest correlation compared to others in the same sub-basket. This strength of the pairing is inversely correlated with Levenshtein distance.

We assumed that retail traders have limited funds that are drastically less than their institutional counterparts (true), fat fingers (not always true, some of us are just sleep-trading at 6:30 AM), and/or are trading on their phones (true for about 20% of retail executions). A recent study on smartphone trading behavior suggests that mobile investing also increases risky behavior which we assumed extends into unintentionally trading different securities due to typos.

As an aside, the paired stock will almost always be lower in market cap (less conspicuous of a company) and certainly less liquid (so those massive double- or triple-digit gains that catch retail’s attention are possible).

c) Institutional traders are extremely unlikely to start the Levenshtein Basket effect; the initial volume surge is most likely attributable to retail investors trading primarily retail stocks.

Anecdotally, after talking to folks who actually work in finance, institutional traders and quants are a bit more sophisticated than the average house cat trading on their laptop. Assuming that institutional traders are smart (probably a good assumption), institutions are almost certain to have character matching restrictions built in to avoid such an error from happening by accident. Even with all of these restrictions in place, we still hear about institutional fat finger trades happening all the time. However, all of these reported mistakes come from a mistaken number rather than a mistaken symbol. Regardless, algorithmic trading is more than likely contributing to these squeezes in price once they are kick-started.

d) Stocks that increase in price due to Levenshtein Basket activity will revert to the mean eventually; the likelihood of said reversion is monotonically increasing with regard to time.

In other words, once a stock price has been pumped due to the Levenshtein Basket effect, the chance of a stock returning to its "pre-pump" price increases over time as profit taking occurs and the market begins to correct itself of the anomaly. If a stock price remains sticky at its post-effect levels, it doesn’t necessarily disprove the Levenshtein Basket effect but is no longer the primary driver of price.

Good or Bad News? Long or Short?

In short (haha), this analysis primarily focused on retail stocks that have released positive news, which boosts the price of the primary stock. The buying pressure leaks into other, unrelated stocks within the basket which ends up in price action like this:

It’s very unlikely that negative news will induce significant selling pressure in other stocks. In order to sell stock, you have to own it -- most retail investors only trade notable, popular companies. Even so, there is even more evidence that the short Levenshtein thesis is simply not feasible.

Most individual retail investors are discouraged from shorting for several reasons. First and foremost, the natural skew of the market is positive, which poses significant and potentially infinite risk by shorting (“stonks only go up”). Secondly, if a short position moves against you, the position becomes relatively larger, increasing the risk on the total portfolio. On the contrary, if a long position moves against you, the position becomes a relatively smaller portion of your portfolio. Finally, there are exponentially more hurdles in shorting a stock: margin, unsupported platforms, etc. Going long in most cases simply makes more sense for the average investor.

Understanding these limitations of retail traders, we can conclude that good news has an outsized impact on the Levenshtein Basket than bad news does. This doesn’t mean that bad news does not matter; adjacent stocks may still get hit due to the impact on the market overall (Tesla makes up 1.44% of the S&P 500 at the time of writing; particularly bad news could trigger a market-wide selloff). In the future, we may perform another investigation and write-up on the peculiarities of the short-side Levenshtein Basket, but the effect is definitely more pronounced on the long-side.

Correlations in Price Between Similar Tickers

As a proof of concept, we took popular retail stocks that have exhibited the Levenshtein Basket effect and analyzed their historical price trends. For example, the TSLA sub-basket is made up of six permutations and three stocks. Once we figured out which permutations were real ticker symbols, we ran a price correlation analysis to determine their correlation from January 1st, 2020 to September 2021 -- the higher the correlation, the more likely the stock experiences the Levenshtein Basket effect.

Hourly Corr(TSLA, TLSA): 0.52
Hourly Corr(TSLA, TALS): 0.02

Daily Corr(TSLA, TLSA): 0.08
Daily Corr(TSLA,TALS): 0.04

In the past 20 months (inclusive of any potential Levenshtein basket behavior), there is no relationship between the tickers TSLA and TALS on both an hourly and daily level while there is a low correlation between TSLA and TLSA on the hourly. We’ll use this information to dive deeper into TSLA vs TLSA. If we see a high correlation on specific days where TSLA experiences a large move upwards, it’s likely this is due to the Levenshtein Basket effect. Specifically, we looked for the highest maximum correlation in five minute chunks on days that TSLA had its highest daily percent returns. A sample of the top 10 dates are provided below:

Our findings across the top 100 days (out of roughly 400 total days) are summarized below:

• The win rate historically is roughly ~49% which basically means that unfortunately this strategy is no better than a coin flip
• The peak correlation on most days was found within the first half-hour of the trading day, which where we expect the greatest amount of volume traded to occur
• This correlation is likely to be the strongest due to pre-market action -- retail buyers will typically only trade during regular market hours; when they see TSLA extremely positive, they are likely to chase these gains (and likely to buy TLSA as well)
• The correlation drops heavily after the five minute window, implying that the majority of traders close out on this trade once they figure out their mistake

So…  In the previous section, we drew the conclusion that if and when typo trading does happen in the TSLA/TLSA basket, it’s a freak of nature. And researching for alpha sucks. A lot. So much, in fact, I got sick and took a three day long break to start writing the rest of this research paper.

Let’s try this again on GME/GMED this time, which has a daily correlation of 0.01. We’ve seen this happen during the Great Squeezening of 2020, at least anecdotally, so let’s verify with data! (This time with lower expectations)

The top ten increases between 2020 and 2021 are fairly memorable this time around, since it coincides with when both Wall Street and Wall Street Bets went certifiably apeshit. Like with the exercise of TSLA/TLSA, we’ll also calculate the time of peak correlation between GME and GMED. However, this time around, the results are much more interesting than TSLA/TLSA:

Our findings across the 76 days where GME was green (out of roughly 400 total days) are summarized below:

• The win rate historically is roughly 74% with an average peak correlation of 62% (79% for winners) which means this is a viable strategy!
• The peak correlation (excluding days with zero correlation) actually pretty consistent across all times, but is more likely to happen towards the end of the day
• Aggressive buying volume for GME actually happened towards the end of the day back when meme stocks were most salient
• One would think as meme stocks lose relevance, the regularity of Typo Trading would go down; however, the occurrence is actually quite regular, which further lends itself to being tradable!

Analytical Caveat

It’s worth mentioning that the primary drawback to this type of analysis is that we’re working in candles that are five minutes in length. In reality, these trades take place well under that time (on the scale of one minute or so). However, even though the actual ideal point of entry to take advantage of typo trading is closer to a single minute that time rather than the full five minute window, the correlations in price still hold true across the full five minutes since we’re measuring open-close along that timeframe!

Although it’s unlikely that retail traders are completely failing to realize and correct their mistakes, my hypothesis is that there’s some level of Greater Fool Theory that is driving the prices on the mistaken ticker: the “intrinsic value” of the mistaken stock is simply less important than the inflated demand caused by buying pressure induced by the Levenshtein Basket. This is good news for us as the price doesn’t revert immediately (giving us both time to profit and time for the displaced stock to further displace for even greater profit) and an interesting rabbit hole to dive deeper into in the future (see Looking Forward).

Strategies to Consider

The issue with capitalizing on this phenomenon is that typical strategies aren’t available, resulting in several unique challenges. Many of the stocks that are in the Levenshtein basket aren’t available on major exchanges and live strictly in the no man’s land that are over-the-counter exchanges. OTC options are also exotic options and trade strictly against your counterparty which makes them even more illiquid. Please exercise caution before dipping your toes into these strategies!

TSLA/TLSA:

However, with the narrow range of tickers that are available on major exchanges such as NASDAQ, it is possible to profit… kind of. The options chain for many of these illiquid small- and micro-cap stocks are a sorry state for anyone looking to profit through options. For instance, TLSA options trade on a quarterly expiry, rather than weekly expiries that are available to securities with substantially greater liquidity in the underlying. The spread is also damningly wide: the most liquid of TLSA’s calls (2.5C for March 2022) is currently trading at 0.25/0.35 bid-ask with a volume and open interest of… 1.

It seems that the only realistic way to turn a profit here is to buy shares when the market opens at 9:30 AM if the TSLA pre-market is up meaningfully and sell within five minutes if TLSA hasn’t gained traction. The potential outcomes are listed below:

• 30% -- TLSA will drop due to outside selling pressure; if this occurs despite TSLA increasing in price, you should look to exit immediately.
• 25% -- TLSA will remain flat and your loss will be modest/close to zero, primarily due to slippage fees and liquidity premium (since the secondary ticker is so illiquid).
• 44% -- The other less than half of the time, TLSA will pop up with TSLA. Here, there are a number of strategies you can employ to exit the position depending on your risk tolerance and trading goals!

GME/GMED:

GMED has some notable differences from TLSA. First, it’s market cap is significantly higher than TLSA; in fact, prior to GME’s massive run-up in early 2021, GMED was actually larger than GME, which could add to ticker confusion. Secondly, despite having only options trading on a quarterly expiry, there is actually relatively meaningful OI (in the hundreds, instead of in the single digits), which means we can play options on this for maximum leverage!

A possible (but certainly not optimal) strategy is to buy 30-delta calls at the nearest expiry in order to capture the most value through gamma if GME explodes in value. Similar to TSLA, you have less than 5 minutes to enter this trade.

Looking Forward

Although this exercise was done as a proof-of-concept to understand the mechanics behind “Typo Trading,” in the (hopefully not to distant) future I’d love to better formalize 1) what traits lend a stock to being part of the Levenshtein Basket, 2) test this on other meme stocks like AMC/AMCX, BB/BBBY, ZM/ZOM, 3) if there is a generalizable strategy for ALL Levenshtein Basket stocks, 4) whether the mean reversion (fading the pump caused by good news, so to speak) is viable to play as well, and 5) better understanding the finer details of typo trading in shorter time windows. For simplicity, I had only analyzed five minute candles which obviously lent itself to drawbacks (read above) but I’d love to better understand the buying and selling action within the 5-minute window, which could give us a more granular look into the actual market mechanics of how typo trading works.

-- 🐱 @breadcatbounce

Thank you everyone that has provided their insights & feedback on this, I appreciate you so much:

Unusual Whale Team (Unusual Whale, Lynx in Melbourne, Falcon et al.)

Holo the Wolf of Wall Street