cheri
344 posts
Sory Kawan, gua ngilang lagi😭
bukan di telan bumi, tapi otak maksa gua terus berfikir
bot gua udah gua porting ai agent
lu bisa liat sendiri di log nya openrouter
gua udah cape mikirin market, biar agent aja yang overide, web3 cuman naratif, gua cuman mining candle aja
panjang umung perjuangan🥹
airplanestar 🏴☠️@airplanestar_
progress bot trading gue hari ini 🦀 terminal udah estetik ala open claw feature yang udah jalan: > bot type (daytrade, scalping, dll) > trading style (custom) > market profile (auto) > localhost backend > dashboard web lengkap (termasuk bot control) > realtime data yang paling penting? winrate 100% cumannya.. entry sehari cuma 2-3 kali 😂 nanti pas publish, mending installer atau format bot doang?
Indonesia
@putraxor @_auracabal Yang penting sudah ikhtiar dengan riset dari bot yang dibuat bang @_auracabal .
Tidak asal2an nebak. 😀
Indonesia
Note, gua mulai uji coba bot polymarket start dari $7 (scroll aja postingan gua). Sekarang portofolio $35 berarti udah x5 ya? uji coba bot engga berhenti sampai disini aja, akan gua terusin sampai tanggal 22 April (Tepat sebulan dari pertamakali gua mulai uji coba bot dengan $7 aja). Apa hasil akhirnya akan bagus? sejauh ini udah 5x atau profit 400% sih.
*disclaimer,
-Gua engga jual bot, tapi gua akan dengan senang hati kirim bot nya ke kalian bagi yang udah support gua dengan donasi besar (dalam jumlah tertentu) sebagai tanda berterima kasih nya gua.
-Gua cuma orang miskin/ orang susah (gua serius)
-Gua bukan lulusan IT atau programmer.
-Gua bukan penasihat keuangan.
-Gua mau buktiin apakah orang luar yang pada ngebuild bot polymarket itu hanya marketing aja atau memang plot twist nya bisa di implementasikan.
-Gua tidak sama sekali mempromosikan / mereferalkan polymarket.
-Polymarket itu judi, ilegal (dibeberapa negara) dan highrisk.
Indonesia
Still finding ideal exits but hey caught some x on $picante !
Mining helium@mininghelium1
OK you all wanted a repo, here you go 🌙 github.com/fciaf420/moonb… Use at your own discretion. And since my time is valuable, I take a 0.5% fee on Ultra swaps that goes back to supporting the build. Time is money. Reality check: 0.5% on a 7x is nothing. You keep the 6.965x. We both eat.
English
semalem kita build agent live di @MeteoraIDN buat @colosseum hackathon dari @SuperteamINDO yang campus edition.
kita mulai dari brainstorming, bikin plan, sampe eksekusi hingga dry run.
ngabisin 2 limit codex dan 1 claude dan hasilnya sudah cukup memuaskan, 1 shot plan dan 1 shot eksekusi.
temen-temen yang mau belajar bisa join @MeteoraIDN dan minta getok admin untuk role Garuda Builder.
Indonesia
There are runners hourly from @scg_alpha , hey anon when are you gonna join the party and start catching some gains?
github.com/fciaf420/moonb…
English
Hey 👋 Here's a friendly rundown of what's landed recently:
🧪 Backtesting
• New /backtest Telegram command — run sweeps right from chat, with a live "adopt" flow to apply winning settings
• Adopt now shows a side-by-side diff + confirmation step before anything changes
• Expanded trail range to include 0.65 / 0.75 / 0.85 for finer tuning
🤖 LLM Advisor
• Added partial_exit action so we can bank gains on runners instead of all-or-nothing
• Split "stuck_loser" verdict into 4 meaningful buckets for smarter reads
• Bumped HTTP timeout 20s → 45s + added a protective backtest strategy
• Auto-enables track-record injection after 20 closed trades
• Memory + anti-jeet prompt + shadow logger upgrades
📣 Alerts & UX
• Milestone alerts: Telegram pings at PnL thresholds with an inline sell button
• Direction-aware copy for LLM trail changes (cleaner notifications)
🔌 Infra
• OKX OnchainOS credential onboarding
• Reduced OKX CLI burst failures
📚 Docs
• README updates covering milestone alerts, backtester flags/output, and proper SCG Alpha credit
Net effect: smarter LLM decisions, safer runner management, better backtest workflow, and fewer OKX hiccups. 🚀
Mining helium@mininghelium1
OK you all wanted a repo, here you go 🌙 github.com/fciaf420/moonb… Use at your own discretion. And since my time is valuable, I take a 0.5% fee on Ultra swaps that goes back to supporting the build. Time is money. Reality check: 0.5% on a 7x is nothing. You keep the 6.965x. We both eat.
English
@noentryzz curl -sSL raw.githubusercontent.com/okx/onchainos-… | sh
or just
npm run install:onchainos
English
Rill enak bgt wok, gacor banget anjer
Alhamdulillah besok makan enak
Noctis@sequencetraders
Enak yh trading pakai bot
Indonesia
Two days ago my bot did very minimal LP, there were a lot of runners and big volumes on many scam coins. A lot of people earned on Unc and Pepe while my bot rejected doing LP on them with its 23 levels of technical filters. It does hurt seeing that I am missing out on the action.
But I know that I don't have to be in every trade. The filters that prevented my bot from earning may well be the same filters that saves me from multiple rugs in the future. If I earn today and lose more tomorrow, that is actually bad. As long as my bot still prints green, no matter how small it is, is a win.
@met_lparmy @MeteoraIDN @MeteoraAG
English
Invite Only Access!
My APR looks pretty solid on Zidee Wallet! 👀
Zirodelta@zirodelta
You want deposit and forget, remember? Zidee is the answer. Let that yield coming in EVERY SINGLE HOUR. Invite-only.
English
major cooks today
republicans > 34x
marvin > 12x
conviction > 9.13x
its free-> scgalpha.com/vault
English
My agent will always be getting better at market analysis.
Hylexa.@4thinfected
My agent never stops printing even when I sleep 😋
English
Just shipped 3 major upgrades to the auto-buy bot:
1. Jupiter Ultra Integration
Swapped from Jupiter v2 to Ultra API. Automatic slippage via real-time estimation (RTSE) replaces the old fixed 25% ceiling. Test trade round-tripped at -0.76% vs potentially losing 19%+ on fast dumps. Every buy and sell now routes through Ultra's optimized execution.
2. Moonbag Strategy
Backtested 14,896 parameter combos across 95 trending tokens. Found that keeping a 20% moonbag after the main trailing stop fires captures second-leg pumps that all-or-nothing exits miss. Results: 5.8x more total PnL vs the old strategy. On the Republicans token alone, moonbag added +140% more profit.
How it works: trail fires → sell 80% to lock profit → 20% moonbag rides with its own trail (60% drawdown) and 2hr timeout. Two sells per trade max.
3. Data-Driven Backtester
Added hybrid strategy grid search — tests scale-out, moonbag %, moonbag trail, and moonbag timeout alongside the existing ARM/TRAIL/STOP params. Real exit logic simulated against live kline data so every setting decision is backed by numbers, not vibes.
Shane Simpson@shanesimpson513
boom
English
Wow just caught a runner!
Shane Simpson@shanesimpson513
ur ngmi trading on tg calls trade on data @scg_alpha
English
3.5 weeks in with the autonomous @MeteoraAG DLMM agent. ~$150 profit this month so far. $280/3.1? SOL profit total Max 4.5 SOL deployed at any one time.
For context: I built an AI that LPs on Meteora 24/7 using its own wallet. It scans the top pools, scores the best LPers on consistency, and uses their strategies to decide when and where to enter for future positions. Full breakdown in my pinned video.
Since the last update, 4 things shipped that actually changed how it performs:
1. Position frequency went from ~15/day to ~61/day. It now cycles capital 4x faster, mostly short-hold entries in active pools. Faster cycles = more fee opportunities on the same balance as I wanted it more aggressive. Mainly just opens and closes when out of range but still noticeably more positions.
2. SOL-only hard filter. It used to occasionally take token-sided entries. Now it only deposits SOL, earns fees without directional token risk. This alone cut the variance (big fluctuations).
3. Pool familiarity. If the agent has profited in a pool before, it gets a confidence bonus on re-entry. It builds memory instead of treating every pool as brand new. Not sure if this is the best but if Claude likes it I’ll go with it
815 positions. 156 pools. 71.5% win rate. $971 in fees. +$290 net PnL. Most days green.
The biggest lesson so far: the improvements that mattered most weren't about finding better strats for all coins. They were about filtering out bad entries before they happen so entering the right coins.
Next up? It might be time to give it more SOL + more autonomy over positions, it is still currently checked by me everyday (of course). Slowly removing the guard rails. Also think I now need lpagent upgraded tier for the top wallet data now.
Still early. Still iterating.
Using @track_lp_ with @lpagent_io @Helius @AnthropicAI
Razzaer@razzaer
Update on the autonomous @meteoraag DLMM agent 2 weeks in now. Scoring is on “v4” for the wallets. Biggest change was adding revenue consistency as the primary signal, how many days a wallet is actually profitable matters more than win rate. The agent uses this alongside hold time, fee yield, and deposit patterns to decide which wallets are worth following. I struggled to get this working I thought it was previously but after checking some of the wallets used as a reference for a pool or position I saw they were high risk high reward (hence the big losses). It caught something I wouldn't have. One wallet it was tracking had a great score but averaged 11-hour holds. It worked out on its own that this was locking up capital while missing faster opportunities elsewhere, and recommended dropping it for wallets that cycles positions every 2 hours due to our low balance so for now this is a newish rule. It also flagged that it wanted more sol as it kept missing entries because the balance couldn't keep up. Attached is the LP Agent view of the agent wallet. Still small scale but the process is working, and the intelligence behind it is the same system @track_lp_ on the intelligence tab. Next up: either I tell it to be more aggressive (we are still just 4 -> 5.2 sol after round trip but could be rly aggressive I think) and better exits either later or earlier than it does right now but they are not always ideal.
English