Wallet tracking proved that traders want to see what other traders do. Most of what widely followed traders do never touches a chain, but they post it.
Centaur · Research
October 29, 2026 · 11 min read
Trading apps made wallet tracking a core feature around a simple behavior. Users find traders who seem to have an edge, then come back, often daily, to see what those traders did next. That habit has underwritten some of the most used products in crypto, but it rests on an assumption few people stop to examine, that the traders worth watching leave their footprints on-chain.
Many of them do not, or at least not in any form a wallet tracker can read. A large share of widely followed traders run perpetuals on centralized venues, buy equities through brokers, and hold commodities in accounts no block explorer will ever see, which leaves their on-chain presence sparse, partial, and at times misleading. Their trades still leave a trail, only a less obvious one, running through posts on X and updates in Telegram channels.
Social trade tracking follows that trail systematically. It captures each trade a named trader shares, prices it at the moment of posting, and builds a verifiable record over time. Wallet tracking, for all its limits, proved that traders wanted visibility into what other traders were doing.
Wallet tracking follows specific on-chain addresses, usually ones with a conspicuous trading history, and shows you what they buy and sell as the trades settle. Fast-trading apps spotted the appeal early. Photon and BullX were among the first to tread that path, letting users follow wallets and watch their trades land. Cielo pushed the idea further, pairing each wallet with its win rate, ROI, and the markets where it has shown the most edge.
What followed was less a strategy than a natural progression of appetite.
Follow a serious discretionary trader for long enough, and the on-chain trail starts to fade, then eventually disappears. Perpetual futures on centralized exchanges, equities, commodities, and most leveraged positions trade beyond the reach of on-chain tools. What remains is an anonymous address with no stated rationale. It may show you a buy, but not who made it, why they made it, or whether the trade meant anything at all.
The gap is wider than it looks. In the 90 days to 28 October 2026, 71.3% of positions opened by X accounts that Centaur tracks sat outside crypto, in stocks, indices and commodities. Although wallet trackers can follow some of that exposure through tokenized stocks and commodities, most of it runs through brokers and futures exchanges, where no transaction reaches a chain.
Even inside crypto, a trader can open a leveraged long on a centralized exchange, add to it, trim it, and close it without a single transaction reaching a block explorer, leaving a wallet tracker with a quiet, lopsided address while the trader runs a full book elsewhere.
The wallet is only the tip of the iceberg. Very little of that activity is hidden, though, because many of these traders describe it, trade by trade, to anyone who cares to read.
That public account of trading is precisely what wallet tracking was never built to capture. On X, the trades that matter, a long or short opened, a position added to, a trim, and a full exit, sit inside a topsy-turvy feed of price charts, memes, arguments in the replies, and running commentary on the macro picture, all of it overwhelming in sheer volume.
Telegram channels tend to be more orderly, sometimes almost clerical, with sizing, scaling plans, and exit updates posted while the trade is still open. Put the two together, and you get a public, timestamped record of trading.
The difficulty is how thinly that record is spread. In the 90 days to 28 October 2026, Centaur processed 1,603,898 messages from tracked traders and extracted 16,057 trade events, about one in every hundred posts.
That small fraction is where the value sits. We call these posts social trade signals, posts that disclose a trader’s position. A bullish take on a stock, for instance, is pure sentiment, but sentiment, however loudly expressed, cannot be scored. A disclosed long at a stated price is different. It shows a trader’s conviction and invites the only question that matters, whether they were right.
Social trade tracking begins where the ordinary social feed leaves off. It finds the posts that reveal an actual position, turns them into structured trade records, and follows those records over time to measure what happened next. Each record captures six pieces of the trade.
Among those six, lifecycle is where social trade tracking pulls furthest ahead of single-call tracking. One trader in Centaur’s ZEC data opened a long at $787 on 26 August 2026 and added at $770 the same day, posted a trim at $1,020 on 4 September, about 30% above the first entry, then added again at $1,167 on 11 September as the rally kept going. A call tracker would record that sequence as one bullish call and move on. A lifecycle record keeps the whole story, the entry, the profit-taking, and the leap of faith required to add back higher, each with its own price and timestamp.
Whether a record like that deserves trust comes down to two methodological choices, technical in name but consequential in practice. The first is traceability, with every event linked to its source post so anyone can audit the record. The second is the scoring window.
Some platforms mark every call against the latest price, which means a call never quite settles and keeps re-scoring itself as the market moves, often flattering a trader whose positions later turned against them. Fixed windows at one, seven, and 30 days close the books on each result. It is less generous to the trader, but far more useful to everyone else.
This is the record Centaur builds for every tracked trade, linked back to the original post.
| Wallet tracking | Sentiment tracking | Social trade tracking | |
|---|---|---|---|
| What it follows | On-chain addresses | Volume and tone of posts | Trades disclosed by named traders |
| Identity | Anonymous address | Topic or creator | Named trader with source post |
| Performance measured | On-chain P&L per wallet | None per creator | Return, win rate, and lifecycle per trader |
| Coverage | On-chain assets | Market mood topics | All meaningfully publicly discussed assets |
| What it proves | What a wallet did | What people are saying | What a named trader disclosed, and how it performed |
The same logic is now gaining ground across retail trading, where the spotlight has shifted toward verified social features. The largest platforms have arrived there by different routes, though some of the similarities are hard to miss.
Robinhood opened the beta of Robinhood Social on 18 March 2026 with a feed showing live trades, one-year and daily P&L, profit rate, and trade history. Every profile is KYC-verified, with coverage spanning stocks, options, crypto, and prediction markets.
eToro set out on that path years earlier, building social and copy trading around a visible record next to each trade, drawn from traders on its own platform. Binance took a looser approach with Binance Square, a social feed inside its exchange app where users post market views, while FOMO shows what its own users hold alongside a written thesis for each position.
Set side by side, these products reveal the same tension. Verified social features are becoming standard, yet each one covers only the traders active on that platform. Many widely followed traders still share their trades on X and Telegram, well outside those walls.
The design also raises a regulatory question. As Fortune's coverage of the launch noted, accounts endorsing stocks can start to resemble advice, leaving each platform to decide how those records should be labelled.
Reaching the traders beyond any single platform means sourcing data from elsewhere. At that point, the question stops being about features and starts becoming one of infrastructure.
That infrastructure is the social trading signal layer, bridging the gap between public social networks and trading products while doing the unglamorous work neither side is eager to handle itself. It reads posts from X and Telegram, extracts social trade signals, verifies and scores them, then delivers structured records through an API or MCP server that AI agents can call directly. The same data can then feed different products, giving each platform room to build its own experience on top of it.
Every team eventually weighs building the pipeline in-house against licensing the layer, a trade-off our analysis of exchange activation sets out. Either way, the layer is only as useful as the data flowing through it, which is why evaluation deserves as much attention as integration.
Most vendor demos look convincing, which is rather the point of a demo, so the real test is whether the data holds water once it is in front of the traders. These eight checks are a sensible place to start.
Today, these checks protect a product feed. As copy strategies and automation move closer to the mainstream, the same checks will sit much closer to the money.
That shift is already under way. Social trade tracking is following in the footsteps of wallet tracking, with visibility and alerts already live, while copy strategies and automation are waiting further down the road. Each step brings a harder question with it. If a platform is going to let users put money behind a strategy built on public traders, the record needs to show what would have happened in practice, with an equity curve and every assumption around sizing, entry, and exit laid out in the open.
That curve becomes the strategy's fact sheet. AI agents reading social trade signals will need the same kind of paper trail. Centaur is working toward both, starting with trader-level backtests.
Build on verified social trade data. Explore the API docs or Get Access
