TWO 100X+ RUNNERS IN 3 DAYS. MY SCANNER CAUGHT BOTH.

From September 25 to September 28, we reviewed SUPERCLUSTER alerts and tracked each coin’s market cap during the 24 hours that followed.

The review covered smart wallets, KOLs, clusters and launches from tracked developers. It also gave us the first live results from smart wallet autobuy, which launched on September 27.

The most useful part of the report is the connection between the results and the system behind them: how a wallet gets selected, how its next buy becomes an alert, and how that alert is measured.

What the alerts found

Smart wallets produced 159 alerts. After those alerts:

Clusters had the highest 2x rate in this sample:

Among cluster alerts, 21% reached 10x or more. KOL alerts had a lower overall hit rate, but produced the biggest individual run.

Among cluster alerts, 21% reached 10x or more. KOL alerts had a lower overall hit rate, but produced the biggest individual run.

These figures measure the highest recorded market cap within 24 hours of an alert, divided by the market cap at that alert. A 10x run also counts toward the 2x and 5x thresholds. These are peak moves, not realized trading returns.

How a wallet earns a place in the scanner

Every hour, SUPERCLUSTER looks at coins that reached a market cap of at least $500K and checks their top traders.

It looks for wallets that bought within 90 minutes of launch and sold at 2x or more. Those wallets then go through a second check: their own on-chain history.

The bot reviews the last 1,000 transactions. A smart wallet needs at least 10 closed trades and a net profit to qualify. Known influencer wallets go into the KOL list.

That second check is what makes the selection useful. One early purchase starts the investigation; the wallet’s broader record determines whether it stays on the list.

High-frequency bots and snipers buying 30 or more new coins a day are filtered out automatically. Wallets that keep losing are removed manually and blocked from returning.

The largest recorded runs

These were five of the largest moves in the review:

Market caps are rounded for readability; multiples come from the recorded data.

The spread across wallets matters. These runs came from several tracked buyers, giving us multiple histories to study and compare.

What a cluster adds to an individual buy

A cluster forms when two or more tracked wallets buy the same coin within 30 minutes.

The bot keeps one active cluster message per coin. When another tracked wallet joins, the message is sent again at the bottom of the feed with the complete buyer list.

That makes the sequence visible: who bought first, who followed and how the cluster developed.

The timing matters because a cluster builds over time. The first wallet and the fifth wallet can buy at very different market caps. Reading the individual purchases shows how much of the move happened before each buyer arrived.

Across these three days, clusters produced a 52% rate of reaching 2x and a 21% rate of reaching 10x. Both were higher than the corresponding figures for individual smart wallet alerts.

The wallet record behind the badge

The review also showed why the largest single win should be read alongside the rest of a wallet’s activity.

samsrepx’s best run was 98x. For 76Uw…, it was 120x. r4yancore produced the largest individual move at 128x across a much larger sample of 108 buys.

The sample size gives the percentage context. A record of 12 buys and a record of 108 buys carry different amounts of evidence.

SUPERCLUSTER assigns gold, silver and bronze tiers, recalculated every hour using the last three days of alerts. Gold requires at least 10 alerts and a 2x rate of at least 30%. Smart wallets also need at least $20K in profit and 30 trades.

For wallets with at least 10 alerts, a 2x rate below 15% meets the bronze performance threshold. The rating becomes more useful as the wallet builds a longer record.

How the next purchase reaches the feed

Helius webhooks stream transactions from tracked wallets into the bot within seconds.

The scanner reads which token entered the wallet and how much SOL left it. It ignores dust purchases, and only the wallet’s first purchase of a particular coin triggers a buy alert.

DexScreener supplies coin data. If a token is too new to appear there, the scanner reads its name from the blockchain and calculates market cap from price multiplied by supply.

At the alert, the bot records the market cap. It then checks the coin for 24 hours, every minute initially, followed by intervals of five and 30 minutes, keeping the highest observed value.

A brief spike between checks can be missed. The report therefore describes the peaks the system actually recorded.

Turning the signal into a managed position

Smart wallet autobuy went live on September 27. Its first 45 real-money trades had a 47% profitable-trade rate and an average return of +4.9% per trade. The combined result was approximately flat. We need a longer sample to assess execution.

Each user has a separate bot wallet, with encrypted keys that can be exported.

For coins still on pump.fun, the bot buys directly on the bonding curve using our transaction builder. Smart wallet autobuy can also buy coins that have migrated to an exchange through Jupiter.

Once a position is open, the bot checks its price:

Exit controls include take profit, partial take profit, stop loss and a time-based exit. After a partial take profit, the remaining position has an exit rule triggered by either a 40% drop from its high or a return to the entry price.

The bot also enforces one buy per coin per day, a daily limit and a maximum of three open positions. During rollout, we added migrated-token support, stopped duplicate daily purchases and fixed protection of the remaining position after a partial take profit.

Paper mode uses live prices without placing real trades, giving users a way to test settings.

What underperformed and what changed

Developer launches were the weak point: 8% reached 2x. Since September 22, seven serial launchers had generated 185 of 188 launch alerts, repeatedly creating coins that died at $4K to $15K.

Real-money dev autobuy trades had an 11% win rate and an average return of -17%. Across the 10 people trading real money this week, two were in profit; dev autobuy was the main drag.

We removed those seven developers and 18 dead wallets and blocked them from returning. The developer list now refills automatically from developers whose coins crossed $100K and who retained little supply at launch. Their token-creation events are streamed into the bot.

The removal addresses the identified source of the flood. Its effect on trading results will be measured in the following reports.

What users can now inspect and control

My Wallets lets users add up to 10 wallets of their own. Tier selection controls which tracked wallets they copy.

Positions now includes Report 24h and a spreadsheet export, so users can review their own trades alongside the scanner’s alert history.

A watchdog checks the data feed, services and alert flow every 15 minutes. If a monitored service hangs, it restarts it and notifies the administrator.

The review gives us three separate records to keep improving: how consistently each wallet finds runners, how buying activity develops into a cluster, and what the autobuy system actually returns.

The next report, in a week, will add more live trades and show what changed after the developer-list cleanup.

https://t.me/kotte_writes