Watching A Graduation Live With Public Data Only

Watch the curve account directly and treat everything else as commentary. The curve state is a public account whose contents can be read on demand, completion is a flag in that account, and the migration is a transaction you can open in full. Every progress bar and every chart you might watch instead is derived from those same facts, later and with the derivation unexplained.

What you are actually watching

During the curve phase there is one thing that matters: a program account holding the curve state for a mint. Inside it are the position, the accumulated reserve and a flag saying whether the curve is complete. Those few values determine the price, the distance remaining and the moment of the handover.

This is unusually tractable. Most market observation involves inferring an unobservable state from noisy prints. Here the state is written down. If you can read the account, you know the situation exactly, with no interpretation and no lag beyond the network's own.

After completion, the thing to watch changes. The curve account becomes a historical record and the interesting object is first a transaction and then a pool. Both are equally readable, but the pool introduces the first genuinely unknowable quantity of the whole exercise: which third-party surfaces have indexed it, which is a question about other people's infrastructure rather than about the chain.

The curve account, read directly

Reading the account directly means asking a node for its current contents rather than asking a website what it thinks. Both the request methods and the subscription methods are part of the public interface described in the Solana RPC documentation. Two access patterns are available and they suit different purposes.

Polling asks for the account at an interval you choose. It is simple, it is easy to reason about, and its resolution is exactly the interval you picked. For watching one token across an evening it is entirely adequate and it fails in an obvious way: if it stops, you notice, because readings stop appearing.

Subscription asks to be notified when the account changes. It is lower latency and better suited to anything automated. Its failure mode is worse, because a dropped subscription looks exactly like an account that stopped changing, and the two are indistinguishable without a heartbeat. Anything built on subscriptions needs a periodic poll alongside it purely to prove the pipe is alive.

Either way, record what you read with a timestamp. The series is the valuable artefact, not any single reading, and nobody keeps this history for you.

Four sources and what each is good for

The ordering is deliberate. Anything higher in the table can be used to check anything lower, and the reverse is never true. When two of them disagree, the one closer to the state is right, and the disagreement itself is usually the most interesting thing you have seen all evening.

The observation procedure

1. Fix the mint. Get it from a source you trust and record it in full. Every later step derives from this, so an error here contaminates everything.
2. Locate the curve account. Find the launchpad program account holding curve state for that mint, and confirm it references the mint you expect rather than assuming the lookup was right.
3. Establish the condition. Read from the launchpad what marks a curve complete, and identify which field in the account state corresponds to it. Do this before you need it.
4. Start reading. Poll or subscribe, and log each reading with a timestamp. Keep the raw values rather than a percentage you computed, because you may want to recompute later.
5. Set a distance alert, not a time alert. Alert when the remaining distance to the condition falls below a level you choose. Nobody can alert you on a time, because the time is unknowable.
6. Treat the flag as the event. Completion is the curve account being marked complete. Not a chart, not a post, not a screener badge.
7. Find the migration transaction. Look at or shortly after the completion boundary for the transaction that creates the pool, and open it in full in Solana Explorer rather than skimming a summary.
8. Record the pool. Address, owning program, both deposits, and what happened to the LP position. This is the payload of the entire exercise.

What to record, and when

• The mint, in full, before anything else.
• Each curve reading with its timestamp, kept as raw values rather than as a computed percentage.
• The completion slot and block time, which are the exact boundary between the two series.
• The migration signature, which is the primary evidence for every later claim about this graduation.
• Both deposit amounts, from which the opening price and the pool constant follow directly.
• The pool address and the program that owns it, which together identify the venue without inference.
• The state of the LP mint immediately after migration, which settles whether the seeded position was burned, locked or left spendable.
• Depth readings at intervals over the following hour, because depth history is the series nobody stores and everybody later wants.

Illustrative arithmetic on remaining distance

Things people watch that carry no information

Some of the most-watched indicators around a graduation contain nothing, and knowing which ones frees attention for the ones that do.

• A percentage bar with no stated methodology. If it does not say what quantity it shows or when it refreshed, it cannot be checked, and an uncheckable number is not evidence.
• Transaction count without distinct signers. Many transactions from few addresses and few transactions from many addresses look identical in a count and mean opposite things.
• Chat volume during the final minutes. It tells you people are watching, which you already knew because you are one of them.
• Comparisons to other tokens that graduated. Each curve is its own account with its own history, and similarity in one number does not transfer any other property.
• Anyone announcing imminence with confidence. The only people who could know are the ones about to buy, and their knowledge is about their own intention rather than about the market.
• The first candle after the pool opens. It is dominated by arbitrage and routed legs, and it describes plumbing rather than sentiment.

Latency, and why you are always slightly late

Every observer is behind the chain by some amount, and the amount differs by observer. A node has to receive and process the block. Your request has to travel. Your poll interval adds up to its own length. A subscription is faster and still not instant. Everything downstream of that, including every website, adds more.

The practical implication is not that observation is futile but that it should not be confused with participation. Watching a curve complete is a way of knowing accurately what happened, shortly after it happened. It is not a way of being first, and treating it as one produces frustration and bad decisions in equal measure.

It also explains a phenomenon that reads as unfairness. People who trade the first slots of a new pool were not tipped off; they were running infrastructure closer to the chain, watching the same public state as everybody else, with less distance between reading and acting. The asymmetry is real and it is structural rather than secret.

Teams that maintain activity across a launch face the same latency question from the other side. Anything that has to act at the moment the venue changes needs to resolve the new venue itself rather than wait for a person to notice; that is the practical difference between a script and a Solana trading volume bot built with the transition in mind, and it matters for exactly the minutes described here.

The first minutes after completion

Once the pool exists, several unrelated processes run at once and they are easy to mistake for one another. Arbitrage connects the new pool to the rest of the market, and it is mechanical rather than discretionary. Routed swaps land as multiple legs, so both transaction counts and turnover run higher than the number of human decisions behind them.

Liquidity providers may arrive to capture fees during the busy window, which raises depth. Some of them leave later, which lowers it again. Meanwhile each third-party surface indexes the pool at its own moment, so quoted prices differ across front ends for a while before converging.

The correct posture in these minutes is to record rather than conclude. Take depth readings, which an explorer such as Solscan will give you from the pool token accounts directly. Note when each surface you use starts showing the pool. Write down the first few prices with their sources. Conclusions drawn during this window are almost always about plumbing that has since resolved, and the notes are what let you tell the difference an hour later.

Two observers, one event

It is worth naming the two distinct reasons people watch a graduation, because they want different things and mixing them produces a procedure that serves neither.

The first observer is operational. They are on the team, they need the pool address as soon as it exists, and their success condition is that the right address reaches the right people quickly and the wrong one reaches nobody. For them the interesting part of the evening begins at completion, everything before it is waiting, and the whole value of watching is being able to read the migration transaction rather than wait for a third party to describe it.

The second observer is analytical. They are building an understanding of how these events behave and their success condition is a clean record: a timestamped series, two deposit amounts, a depth history, and an honest separation of what was read from what was inferred. For them the migration is the middle of the exercise rather than the end, and the hour afterwards carries most of the information.

Both use the same sources and the same procedure. What differs is what they keep. The operational observer needs one address, verified, immediately. The analytical observer needs everything, and needs it recorded as it happens because none of it can be recovered later. Deciding which of the two you are before you start is the difference between a useful evening and a set of screenshots that answer no question you will actually have.

The discipline that makes this worth doing

The value of watching directly is not speed. It is that you end the evening with facts you read yourself instead of impressions assembled from other people's displays. When somebody later claims a graduation went a particular way, you have a signature, two deposit amounts and a timestamped series, and they have a screenshot.

The discipline is small and consistent. Read state rather than derivations. Keep raw values with timestamps. Separate what you observed from what you inferred, on the line where you write it. And accept that the one quantity everybody wants, the time of completion, is not available to anybody at any price, because it depends on decisions that have not been made yet.

Questions the desk is sent