Is Binance evil? The data tells an unexpected story

Author: ltrd Compiler: Block unicorn

Anyone who has consistently profited over the long term knows that making decisions without bias or emotion is key to a sustainable trading career. You must think outside the box and constantly reassess risk-reward and the probability of adverse choices. This is why a well-structured research process is crucial for every successful trader.

But why I'm speaking to you this way—and why I'm calling this article "Is Binance Evil?"

is simple. Over the past few weeks, I've witnessed strong sentiment surrounding Binance and other exchanges. Some of the arguments against exchanges, particularly Binance, are certainly valid, but I kept seeing biased reasoning and conclusions, so I decided to conduct a simple, transparent study of one hypothesis: H₀: "Binance is evil and a negative for projects listed on the exchange." The first thing that inspired me to conduct this research was a post by Scott Phillips (I actually love your posts and your way of thinking—nothing personal here, I hope you'll forgive me). He posted a beautiful chart showing the average price performance of all cryptocurrencies in the first 300 days after Binance's listing. There's nothing wrong with the chart itself—I like the analysis—but one thing that bothered me was the statement: "Binance is a cancer to the industry." I just couldn't see the connection between the data in the chart and the conclusion. Imagine you walked into my office (as many people do every day) and told me, "Tom, look at this chart—Binance is the cancer of this industry." You better have backed up everything on your work laptop, because you're never going to touch it again. This post isn't really about Binance—it's about testing a hypothesis and figuring it out. It's about methodological integrity and convincing people that your hypothesis is valid. Before we begin, I'd like you to critique my arguments with your own analysis. That's exactly what we do in research sessions. I'm not going to get angry—I'm so used to constructive criticism that I don't even care; I just want to make sure my analysis is correct so I can learn from it. Your only goal is to go through it and point out every possible flaw in my reasoning. I'm not here to prove that Binance isn't evil. I'm just trying to verify that this hypothesis is true. When I see this type of chart, I always think: There's a random correction missing here. What does this mean? It means I want to look at random listings from other similar exchanges and then subtract those results from the Binance dataset. That's how you remove the bias. In our case, it's not actually random because we can easily calculate all the factors associated with listings on other exchanges. Typically, in high-frequency trading, you can't "calculate everything," so I call it a random correction. When you conduct research, you need to clearly state your hypothesis: I selected all products listed on Binance (spot market) starting from January 1, 2022. Why this date? Because I didn't want to introduce confirmation bias by using data from 2020-2021, as I already knew the results would be significantly skewed towards the positive and not representative of the current market. I only included USDT trading pairs. I only included products that have been traded for more than 90 days. I excluded the first day (which is why all charts start at 0). Why? This is because exchanges handle market openings differently. Some exchanges will "artificially" create a first order well below fair value, simply to make the chart appear to surge upon listing—a completely spurious phenomenon. Some exchanges announce listings long before or immediately upon listing, making it impossible to effectively isolate announcement effects. Removing the first day makes the analysis clearer and more comparable. Of course, you can also come up with your own treatment. After completing my analysis, I obtained the following results: This is the cumulative return of all tokens that met my criteria over the first 90 days after listing on Binance's spot market. What do we see? From the outset, there was immense—absolutely immense—selling pressure. After a few days, things stabilized somewhat, and then we entered a steady downtrend. Why was this? Part of the reason is due to the overall cryptocurrency market trend. On average, tokens tend to fluctuate downward after listing. Furthermore, I selected all tokens listed after January 1, 2022, right after the bull run, so the overall environment wasn't very favorable. Now, let's address my biggest concern—the lack of random corrections. To me, without random corrections, there's no real research. Even if you showed me your last 100 runs, with an average of 10.50, I couldn't judge unless I saw it compared to the overall market. Without a benchmark, there's no judgment. In this case, "overall market" should refer to other comparable exchanges—such as Coinbase and Bybit. So, in order to do this correctly, we need to do the exact same calculations for Bybit and Coinbase (under the same conditions). Let's take a look at the chart below.

As you can see, the Coinbase chart looks much worse than the Binance chart. About 20 days after listing, the expected return drops to around -25% (and the upper confidence interval is still around -20%!). After that, we see the same pattern again—a brief period of stability followed by a slow decline, just like on Binance. The situation with Bybit is slightly different. After 90 days, the expected return still drops significantly, but the initial selling pressure isn't as intense. Based on both the data and my intuition, I believe Coinbase is much more comparable to Binance than Bybit. Now, let's actually compare these exchanges to Binance. To correct for randomness, simply subtract the above results from the main analysis for Binance. The chart below shows this. Intuitively, we now have the net impact of Binance when benchmarked against each exchange (Bybit/Coinbase). You can clearly see—especially in the case of Coinbase—that Binance's impact is positive, not negative. The selling pressure on Coinbase is much greater than on Binance. Of course, once the confidence interval is taken into account, this difference is not statistically significant at the 95% confidence level—but the conclusion is still quite clear: Binance listings outperform Coinbase listings. For Bybit, we can see that it performed significantly better in the first few days after listing. However, the difference quickly increased, and while we can say that Bybit outperformed Binance in the short term, the effect was not particularly significant. After correcting for randomness, we absolutely cannot conclude that Binance is "evil" compared to other exchanges (especially Coinbase), as projects listed on Coinbase perform significantly worse. Now, let's talk about something important—one we don't discuss enough. The Curse of Being the Ultimate Goal Imagine you're communicating with a project team that hasn't yet launched. What would you expect to hear from them? The answer is almost always something like: "Our ultimate goal is to be listed on Binance (or Coinbase, or Upbit)." This statement is crucial when we talk about the impact of a Binance listing on a project. Everyone is waiting for this moment. If you're a major investor or project founder and you truly believe you'll eventually list on Binance, Coinbase, or Upbit, what incentive would you have to sell your tokens after the listing on Bybit? I'd say almost zero—except for some operating expenses that force you to sell a small portion of your tokens. This is why you see significant selling pressure on Binance and Coinbase, and almost none on Bybit (and possibly none on Bitget, KuCoin, or Gate). However, according to our methodology, even after removing the impact of the announcement date, Binance's listing performance outperforms Coinbase's. Now, the question I'd definitely ask you is: "What percentage of tokens do you estimate an average large investor or founder would want to sell after the ultimate goal listing?" We can't answer this question directly—there's no clear data available. But you should at least have an estimate in mind, think about the logic, and come up with a number. As I mentioned before, Upbit is also an "ultimate goal" exchange, and people like listing in South Korea. Unfortunately, we still see strong selling pressure after the listing date. This is almost always a terminal blow for projects – perhaps not as severe as Binance, but still significant – and you can clearly see this in the data. The chart below shows Upbit’s performance and the difference between Binance and Upbit. After 90 days, Upbit slightly outperformed Binance, but the difference is so significant that we cannot reasonably claim that Upbit is the better listing platform. In both cases, we saw strong selling pressure - which is actually completely logical if you think about it deeply. How to price liquidity? There's one thing almost no one has considered. Following its listing, Binance boasts far greater liquidity than any other exchange. Binance allows founders and investors to partially liquidate their positions as needed, or to increase their holdings more significantly when a buyback is needed (honestly, I wish this were more common). So, how should projects or investors price this significant liquidity boost? This is something that (almost) only Binance can offer—and it's absolutely something every participant in this market should be willing to pay for, directly or indirectly. We all want deep liquidity and the ability to go short or long perpetual contracts (of course, our analysis here is focused on spot exchanges, not perpetual contracts, but it's an important feature worth mentioning). A Simple Way to Test Binance's Liquidity Advantage I've been thinking about a simple way to test whether Binance's liquidity is truly superior to other exchanges without introducing significant bias. Here's what I came up with: Find tokens listed on Bybit and Coinbase. Find tokens listed on Binance, but only after they've been listed on Bybit and Coinbase (ideally, as long as possible). Compare the liquidity of Binance, Bybit, and Coinbase a few days after they've been listed on Binance. In this setup, Bybit and Coinbase have mature markets, while Binance is an emerging market. If Binance's liquidity remains significantly better than other platforms, we can confidently say that the liquidity surplus from listing on Binance is real and substantial. This chart shows the distribution of round-trip costs—the cost of executing a $100,000 market buy and a $100,000 market sell. Higher costs indicate lower liquidity. For LA, a token listed on Binance over a month after Bybit and Coinbase, we find that after five days, the round-trip cost on Binance was 184 basis points lower than on Bybit and 110 basis points lower than on Coinbase.

For ONDO, the round-trip costs between Binance and Coinbase are roughly similar - with a slight advantage in Coinbase (only 1.77 basis points, likely due to the difference in minimum tick sizes). Now let's look at the less liquid product, AXL. Here, the cost differences are enormous. For a $100,000 trade, the cost difference is 309 basis points with Bybit and 207 basis points with Coinbase. For a $20,000 trade, the cost difference is still 41 basis points and 46 basis points, respectively. From the perspective of any current or potential holder, these figures are staggering. This is obviously not the only way to approach this topic – but it is a biased starting point. If we want to delve deeper, here are some open questions (which I won’t answer now — time, as always, is limited):

• How should we incorporate broader market movements and their relationship to listing performance?

• How do we quantify announcement effects and incorporate them into the analysis?

• How should we weigh individual cases? Is ONDO more significant than AXL? If so, by what metric (perhaps market capitalization)?

• Should we make the analysis more robust — for example, by winsorizing outliers?

• Do the results change significantly if we exclude BSC tokens from the Binance data?

We can keep asking questions like these forever — that’s the beauty of research.

There is always room for improvement, but ultimately, creativity and research ethics are more important than any specific model. Conducting nearly unbiased research will get you further than any machine learning method. It's always about your ideas, your data preparation, and your culture of reasoning.

Concluding Remarks

We are not here just to discuss research, we are here to discuss Binance.

Whether you think Binance is "evil" or "a cancer in the industry" is entirely up to you. Please examine yourself critically. Don't let bias and emotion constrain you. Because that's not where the money is.