fml (@fmlonchain) header
fml (@fmlonchain) profile picture

Beating the markets with algorithms | prev. @streamdotfun

Joined March 2024
Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)Photo by fml (@fmlonchain)
1/ Not promoting or shaming @JoinProxima, just an interesting bit of data They pitch themselves as the best way to launch your token But launch through them and almost 100% of the time you will get rinsed by snipers Pulled the last 11 days of on-chain data to show it 🧵
1
1
4
26
3/ Same launch, different tool I took launches with an identical setup (same dev buy %, same bundle %) and split them into Proxima vs non-Proxima In every setup, the Proxima launch gets 3–16x more sniper SOL Example: bundled launch, dev 1–3%, bundle 30%+ Proxima: 3.53 SOL sniped per launch Everyone else: 0.22 SOL The coin looks the same on-chain. The only difference is Proxima, and that's what the snipers are keying on
1
2
6
4/ And yes, I've separated devs own side wallets from the real on-chain snipers in this data. It isn't inflated by devs sniping their own launches It's the biggest snipers doing it The top bots by SOL try 10–30% of Proxima launches, but only 0.4–4% of everything else The #1 sniper tries 1 in 4 Proxima launches and 1 in 50 of the rest So yeah, do with this info as you wish
2
4
Replying to @inno_sol
You should have been here when we went from 260 to 8 dollars. The vibes were 10x worse within broader CT. Solana was supposed to be dead, cooked, over for good. Solana builders kept building and hit that 2024 ath 2 years later. TLDR Those same teams, and newer teams have more conviction than ever. Ignore the noise.
3
4
56
2,118
1/ Been analysing some on chain block-0 snipers and their behaviours Despite market volumes being low, there are operators putting up to 1k+ SOL a day into create slot sniping Taking an extended look at one of the biggest today dKMjeGJ4Mk9S3hxdfTThvZ7exks2xMwWqSM5NvN2CTK
1
1
6
115
4/ Size is one of its key params It buys in fixed tiers: 2.1, 3.6 and 8 SOL 8 SOL tier: +130 SOL on 102 buys, 63% win rate 2.1 SOL tier: +37 SOL on 380 buys, 48% win rate Then it scalps. It sells in about 3 pieces, the first one under a second after buying, and it's fully out within 30 min. Its overall win rate is only ~50%. The edge comes from sizing big when its score is high, not from picking winners.
1
2
17
5/ So how does it decide? It's not deterministic If it were, you could write the rules down. But the best rule I could fit to its buys covers only 20% of them, and 80% of the launches that rule flags still get skipped What it looks like instead: every launch gets a score Round dev buy → +points Mesh-funded dev → +points Dev funded seconds ago → big −points Clear the bar → buy Higher score → bigger size (2.1 → 3.6 → 8 SOL) That's why two near-identical launches can get different answers. Do you think you can reverse-engineer its full execution mechanism? Lmk
2
11
fml retweeted
i guess we'll never know
everybody wants to know what i'd do if i didn't throw the best event at breakpoint
40
16
7
154
21,976
fml retweeted
crypto has this obsession with “shipping for the sake of shipping” the simplest products are often the most thoughtful.
100
23
5
390
19,738
who do i ping to get approved for 200ms? @triton_one @jito @anza_xyz @Helius @mert
1
3
110
fml retweeted
Bring back independent thinking 👍🏻
130
131
32
1,033
51,294
fml retweeted
social trading must die we have done a full 180 as an industry we used to prioritize information, edge, and the best tooling. now we prioritize little information, simplicity, and what kol owns a coin. return to actual trading.
43
36
11
456
25,870
The @Pumpfun caller rewards pay you based on what your tailers make, so the people who end up earning are the ones with the biggest following, and from what I'm seeing in the data that has very little to do with who actually calls well A lot of the best callers I've backtested are in chats of like 20 people, and with a following their size that model pays them close to nothing
Been running calls from a bunch of TG and Discord group chats through my backtester, mostly to see if any of them outperform, and there's a lot more consistent callers in small chats than I expected, some sitting at 20%+ average per trade over a few hundred calls Most of them are trading it with sub 1 sol ports tho
5
129
Been running calls from a bunch of TG and Discord group chats through my backtester, mostly to see if any of them outperform, and there's a lot more consistent callers in small chats than I expected, some sitting at 20%+ average per trade over a few hundred calls Most of them are trading it with sub 1 sol ports tho
1
173