Methodology
How Dotstronic reads the chain.
Every figure in the app comes from a definition on this page. Version dotstronic-m1: the version is stamped on every result, and it changes whenever a rule, a denominator or a coverage definition changes.
Principles
Dotstronic is a research tool for public Solana mainnet data. It does not promise profitable trades, does not predict prices, does not identify insiders and does not declare a token safe or unsafe. It reads, it counts, and it says where each number came from.
- Every signal has a trail. A result names its source, its unit, when it was observed, when it was fetched, what it covers and what stands behind it.
- Missing is not zero. A value that could not be read is shown as missing, with the reason. Zero is only ever shown for something that was measured as zero.
- Chain readings and provider figures stay apart. What Dotstronic reads from an RPC endpoint is never merged with what a market provider reports.
- Nothing is filled in. When a source fails, the panel says so. No random number, cached guess or example value takes its place.
Data statuses
Each panel carries one of six statuses. They are shown with an icon and a word, never by colour alone.
| Status | Meaning |
|---|---|
| Available | Read in full for the stated coverage. |
| Partial | Read, but part of the picture is missing. The coverage line says which part. |
| Stale | Read earlier than the freshness limit for this kind of data. |
| Unavailable | The configured sources could not provide this. Nothing is estimated in its place. |
| Unsupported | Dotstronic does not interpret this. |
| Error | A request failed. Nothing was concluded from the missing answer. |
A reading turns stale when it is older than its freshness limit: 5 minutes for market data, 10 minutes for decoded trades, 10 minutes for chain reads and 30 minutes for the token index.
Mint state
The mint account and its metadata account are read in one request, at one slot, with getMultipleAccounts. From the mint account: decimals, raw supply, mint authority, freeze authority, the owning token program and, for Token-2022, the extensions Dotstronic interprets.
- Exact amounts. Raw supply and every balance are handled as integers in base units from the RPC response to the screen. Formatting with decimals is the last step and never passes through floating point.
- Authorities have three states. None (no address holds the control), active (an address holds it, shown with whether it is program-derived), or the read failed. A failed read is never shown as “none”.
- Capability, not verdict. An authority says what an address can do. It does not show intent and is not a safety rating in either direction.
- Unsupported extensions are named and marked as not interpreted. An extension that is not interpreted is not the same as one that is harmless.
- Names are untrusted. Name, symbol and metadata link are labels chosen by whoever controls the metadata. They are shown as plain text; the link is displayed and never opened. Only the mint address identifies a token.
Holders
What can be said about holders depends on what the RPC endpoint will answer, so the panel has three modes and always says which one it is in.
| Mode | How it is read | What it may claim |
|---|---|---|
| Census | One filtered program scan of every token account of the mint, attempted when the index reports at most 25,000 holders or the answer stays under a size limit. | Unique owners, token accounts, and concentration among all owners. |
| Largest token accounts | getTokenLargestAccounts: the 20 largest accounts, with their owners resolved. | Shares of supply for those accounts only. Not a holder count, and accounts are not wallets. |
| Unavailable | Neither read worked on the endpoint. | Nothing. The provider’s indexed figures are shown beside it, labelled as provider figures. |
- Accounts and owners are different things. Token accounts are summed by the wallet or program that owns them. An owner with five accounts counts once.
- Denominator. Every share is of total supply read from the mint account. No balance is removed from the denominator.
- Two concentration readings. “Every owner counted” includes pools, vaults, programs and burn addresses. “Wallets only” leaves those out of the numerator and keeps total supply as the denominator.
- Labels are verifiable. An owner is labelled from the chain (which program owns the address, whether the address is program-derived) and from the pool lists of the market providers. Anything else stays “not classified”.
Price and liquidity
- One provider at a time. Pools come from GeckoTerminal; DexScreener is used only when GeckoTerminal cannot answer. The two are never added together, because both list the same pools.
- Selected pool. The listed pool with the largest reserve. The headline price and liquidity are that pool’s own figures, not an average across pools.
- Liquidity is the dollar value of both sides of a pool, as the provider reports it. “Listed pools” totals add the pools of the one active provider.
- Price history is the candle series of the selected pool. Other pools can trade at a different price.
- No slot. Market data is off-chain provider data. It carries a fetch time and no slot, because none was read.
Trading flows
Flows are built from swap events a provider decoded in the token’s largest listed pools (2 pools, up to 300 events each from the last 24 hours). The unit of counting is the economic trade.
- One transaction, one trade. All swap events of a transaction are merged and the transaction contributes its net movement of the token. A route split across pools, or a multi-hop route, is counted once.
- Round trips. A transaction whose net movement of the token is zero only passed through it. It is counted separately and appears in neither the buy nor the sell column.
- Transfers are not trades. A transfer into a pool or an exchange is not a sale. Only decoded swaps are counted.
- Failed transactions moved nothing and are never counted.
- Dollar values are the provider’s valuation at the time of each trade. Today’s price is never applied to an earlier trade. A trade with no valuation is counted in token quantity and reported as unpriced.
- Equal windows. When one pool hits the provider’s event limit, the sample is cut to the span every read pool covers, so the early part never holds one pool’s trades alone.
- Coverage is stated. The panel lists the pools read, the pools not read, events merged, events set aside and trades flagged by the provider.
When Dotstronic decodes a transaction itself (launch scans, wallet pages, the Decode button), each balance change is classified by one written rule:
| Classified as | Rule |
|---|---|
| Decoded swap (buy) | The token balance of a signer went up and a quote asset left the same signer, in a transaction that invoked a program known to exchange assets. |
| Decoded swap (buy) | The token balance of this address went up in a transaction that invoked a trading program, while a different signer paid a quote asset. |
| Decoded swap (sell) | The token balance of a signer went down and a quote asset reached the same signer, in a transaction that invoked a program known to exchange assets. |
| Venue side of a trade | The balance of a non-signing account moved opposite to a signer in a transaction that invoked a trading program. Treated as the venue’s side of the trade. |
| Liquidity-like movement | The token and a quote asset moved in the same direction for a signer in a transaction that invoked a trading program. Consistent with adding or removing liquidity; not classified as a trade. |
| Transfer in | The token moved between owners in a transaction that invoked no known trading program. |
| Mint | A mint instruction for this token credited an account of this owner. |
| Burn | A burn instruction for this token debited an account of this owner. |
| Unknown purpose | The balance changed in a way none of the documented rules explains. |
Launch activity
“Launch” can mean four different events. They are reported separately and never substituted for one another.
| Starting point | Definition | Source |
|---|---|---|
| Mint creation | The transaction containing the instruction that initialized the mint account. | Launch scan (chain) |
| First observed transfer | The earliest movement of the token between two owners inside the examined window. | Launch scan (chain) |
| First supported pool | When the token index first saw a pool for the token. | Token index (provider) |
| First observed swap | The earliest purchase the swap rule recognises inside the examined window. | Launch scan (chain) |
- Finding the start. With an indexing provider configured, the oldest transactions are requested first. Without one, signatures are paged back from the newest transaction within a budget (12 pages of 1,000 on the key-free endpoint, 40 on a configured one).
- If the start is not reached, no early buyers are listed at all. A window that does not start at the beginning says nothing about who was first.
- Window. The first 60 successful transactions on the key-free endpoint, 200 on a configured one. Buyers who arrived later are outside it.
- “Earliest observable” is not “first ever” unless the creation transaction was observed in the window.
Launch clusters and the wallet graph
The Launch Cluster Scanner surfaces candidate relationships for investigation. A relationship is an observation about public transactions. It is not proof of common ownership, and a cluster is not a confirmed transaction bundle: no bundle provider is queried, so the term “confirmed bundle” is never used.
| Rule | What it requires and how confidence is set |
|---|---|
| co-signed | Two early buyers signed the same transaction. Confidence: high. |
| direct-funding | One early buyer sent SOL directly to another before its first purchase. Confidence: moderate; high when their first purchases are also within two slots of each other. |
| direct-token-transfer | One early buyer transferred the token directly to another inside the examined window. Confidence: moderate. |
| shared-funder | Two or more early buyers received SOL from the same address before their first purchase. Low when the source is a high-activity address or the funding is more than seven days older than the purchase; moderate when the transfers fall within one hour of each other; high when, in addition, the source is the first funder of every wallet or their first purchases are within two slots. |
- Confidence is a word (low, moderate, high) produced by the rules above, shown with its reasons. There is no percentage, because nothing here measures a probability.
- Same-slot activity alone is weak. Popular launches attract many independent buyers in the first slots. It is reported as a pattern with that caveat and creates no group by itself.
- A shared exchange withdrawal is not a link. A funding source with a full page of transactions inside three days is treated as high-activity and lowers confidence. Funding is “close in time” within 60 minutes and “old” after 7 days.
- Every edge exposes its type, supporting transactions, time window, method, confidence reasons and known ambiguity.
- The graph is bounded and laid out once. Nodes are sized by the number of relationships they take part in. A table view carries the same content.
Holder positioning and cost basis
Cost basis is an estimate from observed transactions, shown only for wallets inside a launch scan’s window and always labelled with its coverage. There is no token-wide profitability figure: that would need every holder’s full history.
- Method: weighted average cost, kept in separate pools per wallet for tokens bought with SOL, tokens bought with a USD stablecoin, and tokens of unknown cost.
- Unknown cost stays unknown. Tokens that arrived by transfer, mint, airdrop or a purchase with no observed price go to the unknown pool. They are never assumed to have cost zero.
- Outflows take proportionally from all pools. Only the part of a sale that comes from a known-cost pool and was paid in that pool’s own unit produces a realized result.
- Gross proceeds are not profit. Proceeds from unknown-cost tokens are reported as gross proceeds. Tokens bought in one unit and sold for another are reported as cross-unit and are not netted.
- Unobserved inventory. If more left a wallet than was seen arriving, it held tokens from before the window, and the result is marked partial.
- Fees. SOL amounts are what the swap transaction took from the wallet, including tips and account rent paid in the same transaction and excluding the network fee.
- Realized and unrealized are kept apart. Only realized results on matched sales are computed; no unrealized figure is shown.
Evidence rules
The evidence feed is produced by deterministic rules. Each finding stores its rule id and version, the token, the observation window, the inputs, the threshold, the coverage, references, the time the data describes and the time the finding was generated. No language model writes or scores any of it.
| Rule | Measures | Threshold | Cannot show |
|---|---|---|---|
| Buy and sell volume out of balanceflow-imbalance v1 · window | Dollar volume of decoded buys and of decoded sells over the provider’s trailing 24 hours. | The larger side is at least 1.5× the smaller side, and gross volume is at least $100,000. | Covers the venues the provider indexes. Volume is not intent: market makers and arbitrage add to both sides. |
| Pool liquidity movedliquidity-change v1 · window | The provider’s own 24-hour change in the token’s pooled liquidity, in percent. | The change is at least 20% in either direction, and liquidity is at least $50,000. | A dollar figure moves with price as well as with deposits and withdrawals. It does not say who moved liquidity or whether it is locked. |
| Indexed holder count movedholder-change v1 · window | The provider’s own 24-hour change in the number of holders it indexes, in percent. | The change is at least 5% in either direction, for tokens with at least 1,000 indexed holders. | A holder count is an index figure. One person can hold many wallets, and one wallet can be a pool or a vault. |
| Supply held by the largest holdersconcentration-level v1 · window | The share of supply the provider attributes to the largest holders. | The share is at least 60% of supply. | The provider’s figure does not separate pools, vaults and custodians from ordinary wallets. Open the Holders tab for the labelled list. |
| Top-holder share rose between two snapshotssnapshot-concentration v1 · snapshot | The provider’s top-holder share in two stored snapshots. | The share rose by at least 3 percentage points. | Two readings say the share differs; they do not say which wallets changed. |
| Liquidity fell between two snapshotssnapshot-liquidity v1 · snapshot | Pooled liquidity in dollars in two stored snapshots. | Liquidity fell by at least 20%. | A dollar figure falls with price as well as with withdrawals. |
| Total supply changed between two snapshotssnapshot-supply v1 · snapshot | The raw supply recorded on the mint account in two stored snapshots, compared as exact integers. | Any difference. | Says that supply changed, not why: mints, burns and scheduled emissions all change it. |
| A mint control changed between two snapshotssnapshot-authority v1 · snapshot | The mint authority and the freeze authority recorded on the mint account in two stored snapshots. | Either one differs. | A control describes what is possible. It does not show intent. |
- Window rules read the provider’s own trailing 24-hour statistics. A token whose first pool is younger than 24 hours has no “24 hours ago”, so its change figures are left out.
- Snapshot rules compare two readings Dotstronic stored itself. They need at least 30 minutes between snapshots, the same methodology version, and a newer snapshot no older than 7 days. Without history storage they are disabled, not simulated.
- Feed order. Snapshot findings first, then rules take turns, each contributing the finding furthest past its own threshold.
Data coverage
| Source | Used for | Coverage and limits |
|---|---|---|
| Solana RPC | Mint state, holders, launch scans, wallets, transaction decoding | Chain data at a stated slot. The key-free public endpoint allows about one transaction lookup per second, refuses largest-account queries and limits program scans. |
| Jupiter Tokens API v2 | Explore lists, search, indexed holder count, 24-hour statistics, evidence feed inputs | Tokens Jupiter indexes. Lists hold at most 100 tokens. Holder counts and “top holders” are the provider’s own figures. |
| GeckoTerminal API | Pools, price history, decoded trades | The first page of a token’s pools (at most 20). Up to 300 trades per pool from the last 24 hours. About 30 requests a minute, so answers are cached. |
| DexScreener API | Pool labels for holders, fallback pool list | Pools DexScreener lists. Used alone, never added to GeckoTerminal. |
| Helius (optional) | Oldest-first transaction history | Only when a key is configured. Without it, launch scans page back within a budget. |
| PostgreSQL (optional) | Snapshot history for snapshot rules | One snapshot per token per 30 minutes, kept 90 days. Without it, snapshot rules are disabled. |
Which of these this deployment can reach right now is measured, not assumed: see service status.
Known limits
- A complete holder list for a large token needs an RPC provider that serves filtered program scans or largest-account queries. The key-free endpoint does not.
- Decoded trades cover the two largest listed pools and at most the last 24 hours. Trades in other pools, and older trades, are not in the sample.
- A route that crosses one read pool and one unread pool is counted on the read leg only.
- Launch scans find only transactions that reference the mint address. Transfers made with the older Transfer instruction, and swaps on venues whose instructions do not name the mint, are missing from the window.
- Bundle membership cannot be confirmed. A Jito tip is reported when one is visible in a transaction; that is all.
- Nothing here says whether liquidity is locked, who can withdraw it, or what a program other than the token program can do.
- Token-wide cost basis and profitability are not computed.
- Rate limits are per server instance and are a courtesy bound; provider quotas are the real limit.
Planned and not yet available: MCP server for agent clients; Telegram alert delivery; Background alerts after the page is closed; Full-history holder cost basis.
Privacy
- No account, no wallet connection. Dotstronic is read-only. It never asks for a signature and cannot move funds.
- Stored in your browser: your watchlist, thresholds, alert history, recent lookups and display preferences, under one local-storage key. You can export or delete it in Settings.
- Sent to this site’s server: the addresses, signatures and search text you look up, so the server can read them from the chain and from its data providers. The providers see the server’s requests, not your browser.
- Not built in: analytics, advertising or tracking scripts. The page loads no third-party scripts, fonts or images; token images are re-encoded and served from this site.
- Operator logs. Whoever hosts a deployment can see ordinary request logs (address, path, time), as with any website. Dotstronic itself keeps no per-visitor record, and its rate limiter holds addresses in memory only.