Instant Graduation Filter: A First Skip Rule for AI Agents
Build an instant graduation filter for pump.fun agents: define the skip rule, handle unknown labels, read the snapshot, and test on paper before wallet access.
On this page
Disclosure: QuantBase is made by the team behind memecoin.buzz.
Risk: Most meme coins go to zero. Not financial advice.
An instant graduation filter gives an AI trading agent a clear instruction: skip a pump.fun coin if a single buyer took its whole launch curve within 2 seconds. That is more precise than asking the agent to find healthy launches or avoid suspicious activity. Those phrases leave too much room for interpretation.
The filter does not predict what a coin will do. It excludes a particular launch pattern from the coins your rule can consider. The useful work is defining that pattern, deciding what happens when the label is missing, and checking that the agent applies the rule at the right time.
What instant graduation actually means
A bonding curve is the launch-stage mechanism that prices buys and sells. Graduation marks the curve selling out and the coin transitioning toward open-market trading. For the broader explanation, see what graduation means on pump.fun.
In QuantBase’s classification, an instant graduate is bought out within 2 seconds of launch, with a single buyer taking the whole curve. An organic graduate fills through many buys over more than 2 seconds. These describe how the curve filled—not ratings of the coin or its creator.
The distinction matters because the same milestone can come from different activity. A completed curve might reflect many purchases, or it might reflect a single buyer taking the available launch inventory. Treating both as equivalent evidence of broad interest loses that distinction.
There is also a technical boundary to keep separate. The pump.fun public documentation describes a period after the final curve buy when curve trading is unavailable until migration happens. A graduation label is therefore not a promise that an agent can immediately execute an open-market trade.
Read the snapshot without turning it into a forecast
For the last 24 hours, as of 25 Oct 2026, 13:40 UTC, QuantBase’s recorder reported 825 graduations: 332 instant and 493 organic. Instant graduations accounted for 40.2% of graduates in that window.
The denominator matters: that share describes graduates, not all new launches. It also describes the stated recording window, not a permanent feature of pump.fun. The chart is a snapshot, not a continuously updating feed.
This split gives a concrete reason to distinguish the categories when writing a rule. It does not tell you what happens to either group afterward. Nothing in these counts establishes future price behavior or proves that excluding a category improves a trading strategy.
Our instant versus organic graduation guide explains the categories in more detail. Here, the task is narrower: turn the category into an instruction an agent can apply consistently.
Write the instant graduation filter as a hard gate
A hard gate is a condition that blocks further action instead of merely lowering a score. If your intention is to exclude instant graduates, a flattering summary, lively chat, or appealing coin name should not override that exclusion.
Use the following as a proposed rule specification, not as a claim about any product’s command syntax:
- Instant label: exclude the coin from entry consideration.
- Organic label: allow the coin to proceed to the remaining checks.
- Missing, pending, or conflicting label: do not authorize entry under this graduation-based rule.
- Other checks fail: exclude the coin even if its graduation is organic.
That last condition prevents a common mistake: treating a coin that passes this check as approved to buy. An organic label answers a launch-history question. It does not answer every question about the coin.
A plain-language draft could read:
Skip coins labeled instant. Consider only confirmed organic graduates, then apply the remaining entry checks. If graduation type is unknown or conflicting, do not enter. Passing this filter alone must never authorize a purchase.
The unknown case is important. A coin that has not graduated is not automatically organic. Nor does simply waiting beyond the instant threshold establish that many buyers filled its curve. The rule needs an observed classification, not a guess based on elapsed time.
This specification deliberately considers entries only after graduation. It is not a filter that can identify a future organic graduate while the curve is still filling.
For the broader workflow around this gate, use the pillar guide to screening pump.fun launches with AI-agent rules.
Why a single buyer taking the curve matters
A single buyer completing the curve gives you different evidence from a sequence of purchases. It demonstrates that the curve was filled, but not that many independent participants chose to buy during the launch.
That is the central reason for the filter. If your research question concerns launches whose curves filled through many buys, instant graduates do not fit that definition. Excluding them makes the rule match the question.
Keep the inference narrow, though. The label alone does not establish the buyer’s identity, intent, or current holdings. It does not prove wrongdoing. It also does not tell you whether the buyer later transferred or sold tokens.
The organic label has limits too. Many buys are not proof that every buyer is independent. Do not translate a launch classification into a broader claim that the underlying data does not support.
A useful decision record would identify instant graduation as the reason for exclusion. It should not add an unsupported story about manipulation or future selling. The agent needs a reason it can verify, not a detective novel with a coin ticker.
Keep the source classification separate from your interpretation. That makes it easier to review whether the agent followed the filter without debating an invented account of what happened.
Check timing before testing the rule
The most important testing question is simple: when was the graduation label available?
If a test uses the eventual organic label to approve a purchase that supposedly happened before graduation, it is using information from the future. This is look-ahead bias: letting a past decision benefit from facts that were not yet known.
For this guide’s graduation-based rule, the clean sequence is to observe graduation, obtain its classification, apply the gate, and only then consider an entry. A strategy that buys during the curve needs different decision-time conditions; the eventual graduation category cannot serve as an already-known approval.
Review these cases before treating the rule as ready for paper trading:
- Confirmed instant: the exclusion fires, regardless of other favorable checks.
- Confirmed organic: the coin advances only to the remaining checks.
- Unclassified: the rule withholds entry rather than inventing a label.
- Late information: a label received afterward does not retroactively justify an earlier decision.
These are proposed checks for rule behavior, not results from a completed experiment. They help distinguish an agent that follows the instruction from an agent that merely repeats it convincingly.
For each decision, review the label available at that moment and the action the rule allowed. A later classification can help explain the record, but it cannot rewrite what the agent knew earlier.
Keep the graduation filter unchanged while reviewing this behavior. Changing several conditions together makes it harder to understand which instruction caused a particular exclusion.
Put the rule on paper before wallet access
QuantBase is a meme-coin trading app for Solana pump.fun coins. It accepts plain-language instructions about what to buy, skip, and sell, then turns them into a readable, editable rule. Its graduate labels let rules exclude instant graduations.
Its stated process tests a rule against launches from its own recorder, with the sample size shown, before running it on paper money against live launches. Today, QuantBase runs rules on paper; live trading is coming soon. Paper trading uses no real funds and cannot touch a wallet.
You can test an instant-graduation skip idea on paper with QuantBase before considering wallet access.
Read the resulting rule rather than assuming your wording was interpreted exactly. Check that instant means exclude, unknown does not mean approve, and organic still requires the remaining entry conditions. Keep exits explicit too: take profit, stop loss, and a time limit belong in the rule rather than in an agent’s improvisation.
For any agent that could touch a wallet, start on paper money, set hard spending and position limits, and never share a seed phrase, recovery phrase, or private key. Wallet permissions do not substitute for a correct filter, and a correct filter does not substitute for wallet limits.
The QuantBase traders’ Telegram includes a free bot for coin checks covering developer dumps and daily pump.fun stats. Community discussion can help clarify terminology, but it should not silently change the rule you are testing.
FAQ
What is an instant graduation filter?
It is an exclusion rule for coins classified as instant graduates. Under QuantBase’s definition, that means a single buyer took the whole launch curve within 2 seconds. The filter describes an entry restriction, not a price forecast.
Does organic graduation mean a coin is approved to buy?
No. Organic graduation describes a curve filling through many buys rather than an instant buyout. Passing that filter should only move the coin to the remaining checks, not authorize an entry by itself.
What should the agent do when the label is missing?
For the graduation-based rule described here, it should withhold entry. A missing label is not evidence of organic graduation. Make that behavior explicit so the agent does not fill the gap with an assumption.
Can I use the eventual label for a purchase during the curve?
Not as information that was already known at that earlier moment. Doing so introduces look-ahead bias into the test. A curve-stage strategy needs conditions available when its entry decision is made.
Can QuantBase run this rule with real funds today?
QuantBase currently runs rules on paper, and live trading is coming soon. Its paper runs cannot touch a wallet. Any wallet-connected setup should begin with paper testing, hard limits, and no sharing of seed phrases or private keys.
Sources
Written with AI assistance and checked by automatic gates: every number must trace to a source above, and quotes are short. Spot an error? Tell us in Telegram.
