Coding's betting panel made money, but Polymarket is really not a good place for "arbitrage"

A couple of days ago, I shared a piece titled "I Built My Own Investment Dashboard with AI", where I talked about several tools I coded: a cross-market asset panel, an investment map, a personal content operations station, and a Polymarket betting monitoring panel that I've been using frequently recently.

Over the past two and a half weeks or so, I started with a principal of about $1,600 and achieved a return of over 30%. The real-time statistics from the panel basically matched the final actual net profit, with only a difference of around 6U, which can be attributed to minor discrepancies like pending orders/liquidity rewards.

But what I really want to say in this article is not "Polymarket is a great way to make money," nor do I want to package it as some kind of arbitrage tutorial.

On the contrary, after completing this round, I increasingly feel that Polymarket is not a place suitable for charging in with an "arbitrage" mindset.

I. Let's Talk About What This Panel Actually Is

I probably started building this panel around May 21st.

The initial need was simple: I didn't want to open a dozen betting pages every time just to toggle back and forth on yes/no price changes, and I didn't want to manually record things in Excel anymore.

That's right. Before this, I had been using Excel to track buys/sells, floating P&L, settlement nodes, and event types. A simple method for a simple person.

But as anyone who has actually played with it knows, many bets on Polymarket can easily spiral out of control precisely because manual record-keeping is highly inefficient. For instance, you might initially want to make a small purchase, but then the odds move and you want to add more, lacking an intuitive overview. Or, a certain betting event suddenly experiences a sharp movement, and if you fail to update your spreadsheet data in time, you might miss the window to stop loss or add to your position. The list goes on.

Ultimately, the whole process is too fragmented. Without a system, people easily place orders based on emotion.

So, from the very beginning, I built this panel to place every single bet back into a unified framework, turning this feeling into a relatively visual and comparable presentation of information.

After several iterations, I split it into two tabs: "Portfolio Dashboard" + "Opportunity Monitor."

The "Portfolio Dashboard," the core of the entire panel, is a dynamic system that can fetch Polymarket real-time data and recalculate. It's divided into several functional areas (refer to the first image in the article):

• Overview Bar: Total capital (planned, actual reference is limited), capital invested, portfolio value, portfolio floating P&L, total floating P&L (including closed positions). Get a clear picture of the entire account at a glance.
• Tier Allocation Ratio: This is the core risk management module of the panel, also what I consider the most counterintuitive and important area. I'll keep this a secret for now and explain in detail in the next chapter.
• Theme Cluster Exposure: I assigned a "theme cluster" label to every bet, dividing them into East Asia, Middle East, Crypto, US Stocks, Pre-IPO (customizable for adding/deleting). The panel automatically aggregates the proportion of each cluster and sets a single cluster upper limit threshold of 12%. Why design it this way? Mainly to combat Polymarket's most insidious trap – false diversification. More details in the next chapter.
• Individual Position Details: Tier, direction, entry price, settlement price, shares, P&L, entry date, settlement date, notes, etc. Everything is clear at a glance in one row, with the ability to sort ascending/descending and filter by labels.

The "Opportunity Monitor" is a watchlist. I place markets I'm interested in but haven't bet on yet into it.

For each market, I record several key fields, including the event name (with a direct link to the trading page), T1/T2/T3 tier judgment, current yes/no price, returns, annualized returns, unusual movements (customizable threshold, e.g., if the movement exceeds 20% within 24 hours, a popup reminder appears as long as the webpage is open), my designated observation nodes, and the countdown to betting expiry.

There are two small design features I'm particularly satisfied with here: First, I found a suitable Polymarket API endpoint. By directly parsing the betting event's webpage link, it automatically prompts the yes/no options, corresponding prices, and the classification of different options under the same event, significantly reducing manual entry effort. Second, the Tier classification of the same bet automatically re-adjusts based on the remaining days.

Shortly before Anthropic released Mython, my watchlist displayed a clear price anomaly, which could basically be judged as a high-probability deterministic event. Entering at that time allowed me to capture a return of about 10 points. Without a watchlist, it's very difficult to consistently capture such opportunities.

II. Polymarket's Expected Value Trap and the "T1, T2, T3" Design Principle

The above is a brief introduction. What I really want to talk about is a thought I had after my practical testing.

That is, in binary markets like Polymarket, there is a significant structural trap. It's very unfriendly to players who like "single large positions," but relatively suitable for diversified allocators who prefer "opening a supermarket and buying a lot of different things."

I'll try to explain my thinking clearly. If there are any errors or omissions, please just ignore them:

Suppose the yes price c for a certain betting event is 0.80, meaning the market believes this event has an 80% probability. If I personally judge that the true probability q of this event is 0.90, then the expected return for this bet can be roughly calculated as:

EV = q / c - 1 = 0.90/0.80 - 1 = 12.5%

This looks good, but Polymarket isn't a bond. Hidden behind this 12.5% is a very sharp tail risk: if your judgment is wrong, the loss isn't 12.5% but 100%.

So, in my panel, I don't just look at the "expected return." I also monitor two things simultaneously:

• One is the gap between my own probability judgment and the market price, i.e., q - c (I have set an automatic take-profit alert target value, which is the midpoint between the entry price and 100). This is the core of whether an edge truly exists.
• The other is the impact on the total account if this single bet goes to zero if I'm wrong.

The second reason is also the origin of the T1, T2, T3 stratification I mentioned in the first chapter.

Simply put, I divided them into three categories:

• T1 High Conviction: For me, my comfort zone lies in events related to East Asia and certain geopolitical matters. I believe there is an information asymmetry between the East and West. I add them after repeated verification.
• T2 Relatively Stable: Events where I feel the current implied probability is significantly lower than the actual yes or no pricing.
• T3 Pure Speculation: Those with very high odds. These shouldn't be held for long. The best strategy is to bet on the reversal, take advantage of short-term mean reversion to capture short-term profits.

However, it's important to note that T1 comes with hidden costs, especially for long-term targets. For instance, a T1 level bet might have a static return of 18%, but if settlement takes 180 days, the annualized IRR might only be 3-4%, which is worse than just leaving the money idle. During this time, your principal is locked up, causing you to miss out on later high-IRR opportunities.

So, internally for T1, I further break it down into several time buckets (this part is purely personal thinking, so I won't share it further). Anyway, short-term T1-A can be played more aggressively, while long-term T1-C needs to be restrained. Allocating too much capital to long-term, low-IRR targets is a hidden efficiency drain.

T2 has an edge, but you need to leave room for being "wrong." The upper limit per bet is 8-10%, meaning even if this bet results in a total loss, the total account loss is controlled within 10%, allowing you to continue participating in subsequent opportunities.

T3 odds are attractive, but you should use minimal position sizes for observation. Don't rely on it for big profits. Instead, focus on betting on reversals and short-term mean reversion – this allows you to continuously track high-odds events and build a feel for this type of market.

Overall, position size limits essentially create a tolerable space for the possibility that "my judgment might be wrong."

Here's a very counterintuitive but extremely important point: high conviction does not equal a large position. Even if you think an event has a 95% chance of happening, as long as there's a 5% chance of going to zero, the position size must be limited.

To give an extreme example, suppose you make 10 consecutive bets that you believe have a 95% win rate. Each one sounds very safe. But if they are independent of each other, the probability of getting at least one wrong is approximately 1 - 0.95^10 ≈ 40%.

If you do it enough times, you will eventually encounter that wrong one.

And this is just for independent events. In reality, many Polymarket markets are not independent. They often have correlations. For example, "Will the US-Iran dialogue reach an agreement?", "Will the Strait of Hormuz reopen?", "Will the Middle East situation escalate within the month." These three bets look like three independent markets, but the underlying variable is almost the same – the direction of Middle East geopolitical policy. If you get this direction wrong, all three bleed simultaneously.

This is also the biggest help this panel provides me: not improving my win rate, but limiting my ability to make big mistakes. In short, the core value of this panel isn't return statistics; it's risk management.

III. My Real View on Polymarket After This Round

After over half a month of in-depth testing, my biggest takeaway is this: Opportunities do exist on Polymarket, but it is by no means the arbitrage paradise many people imagine it to be.

When we did on-chain arbitrage before, the rules were mostly clear, and price discrepancies could be locked in. But Polymarket is different. It truly tests your logical understanding of the different shifts in sentiment regarding a particular bet (this feeling is really hard to express accurately in words).

For example, regarding political and economic dynamics related to East Asia, Chinese-language users might indeed have a certain information asymmetry advantage. This is worth exploring, but it doesn't guarantee you'll win. In the end, Polymarket doesn't settle based on "the reality you understand." It settles according to market rules and designated data sources (and we've seen plenty of manipulation issues with UMA).

Furthermore, just because you think something is a sure thing in the Chinese context doesn't mean the definition is the same under English rules. Especially in the rule settings for each bet, there are often some subtle textual traps.

So, based on my actual experience, Polymarket doesn't have that many arbitrage opportunities. It mainly relies on information asymmetry and position diversification. Even with high conviction, you can encounter black swan events.

Once that happens, you lose your entire principal.

As a friend once said, "In investing, even if there's only a 1% chance of going to zero, you should never gamble on it."

Because, in the long run, such expected value is negative.

So my current understanding of Polymarket has become more conservative:

• First, don't treat it as a stable income tool. Even with highly confident investments, especially after you've won a few times in a row, don't feel like you've found an ATM. The scariest thing about binary markets is that after you win consecutively, they make you mistakenly believe you can predict everything. Then you make a large final bet, and it wipes out all your previous profits.
• Second, don't equate high win rate with a good trade. If an event has a 90% win rate but the market price is already 0.95, it might actually have negative expected value. Conversely, an event with only a 40% win rate might have positive expected value if the market only prices it at 0.20.
• Third, don't ignore tail risk. This is especially important. Many people see a 10% or 20% return and think it's very safe. But if this trade can go to zero, then it's not a low-risk return in the traditional sense (From this perspective, I even believe there are simply no so-called low-risk opportunities on Polymarket. Every single one is high risk).
• Fourth, avoid false diversification. Buying multiple different markets doesn't necessarily mean you are diversified. For example, the three bets mentioned earlier – "Will the US-Iran dialogue reach an agreement?", "Will the Strait of Hormuz reopen?", "Will the Middle East situation escalate within the month." These three bets look like three independent markets, but the underlying variable is almost the same.

So now, I prefer to view Polymarket as a judgment training ground.

It perfectly complements the political, economic, technological, and financial news I, as a homebody, love to browse daily. It helps turn those judgments that usually just stay at the "I think" level into something that can receive positive feedback.

These skills are equally useful outside of Polymarket.

Incidentally, besides this Polymarket betting panel, I also used Codex to create a dynamic monitoring panel for private market valuations. It mainly tracks valuation changes for unicorns yet to go public – Anthropic, OpenAI, Stripe, Kraken, etc. – in the private market, and the relationship between these changes and the corresponding bets on Polymarket.

Polymarket is essentially a market of expectations. Sometimes, the signals in the private market have already shifted, but Polymarket prices haven't moved yet. Other times, Polymarket prices move first, but the actual data hasn't caught up. The discrepancy between the two is worth continuous observation.

Of course, this isn't risk-free arbitrage either. Private market valuations themselves are not entirely transparent, and there might be differences between different data sources. But as an observation framework, it's quite interesting. I'll look for an opportunity to write a dedicated piece on this later.

Summary

What I've wanted to convey from start to finish is never "I made 30% with a panel, and you can too."

What I find more valuable is creating a tool that helps you turn feelings into a framework, and the framework into discipline. Many times, many people make money not because they've found some secret formula, but because their judgment happened to be correct in that particular round.

This distinction is very important.

I also suggest you start exploring Vibe Coding. You don't necessarily have to use Claude Code. You can try Codex, or even Kimi's recently launched Kimi Work to get an experience. If anyone finds subscribing to overseas services inconvenient, I can later share some smooth methods I use myself.