your time in my space
128 posts


@IBadirli21592 这种叙事一天已经看到好几个了。一会claude 一会kimi。模型视图都一样的😂。 就是营销号而已
中文

他曾在 Jane Street 带了一个量化团队,整整 11 年。去年 4 月,公司把他裁掉了——一个季度的业绩没达标,年薪 41.2 万美元的工作,一周之内就没了。
昨晚,他用一个免费下载的中国开源 AI,在 Polymarket 上一天赚了 917,893 美元。
Kimi K3,2.8 万亿参数,百万 token 记忆。他只租了一个 GPU 集群,跑了一个 agent loop。
链上钱包在这儿:
@0x75973c667cec880353450be3a6f17c5da67b4421-1784278069335?via=Laowu" target="_blank" rel="nofollow noopener">polymarket.com/@0x75973c667ce…
说起来,他在 Jane Street 那十一年,学会的不是怎么用更贵的模型,而是怎么让交易系统更快、更直接。Moonshot 这周刚把 Kimi K3 的开源权重放出来,他一看就知道,这就是他等了大半年的东西——一个能长时间自主运行、还能自己改代码的 agent。
他从春天就开始搭 bot,缺的只是一款能把“整个世界塞进脑袋”的模型。K3 的百万 token 记忆,一次吃掉了他的全部交易日志、一整年的 5 分钟 BTC K 线、所有历史回测数据——什么都不用裁剪。而且它快得离谱:Kimi Delta Attention 让百万 token 推理速度提升了 6.3 倍,循环周期从原来的 1 分钟压缩到几秒。
它交易,观察结果,自己改规则,继续跑。第一晚就完成了 917 个循环。
这故事最讽刺的地方在于,那些对冲基金花了数亿美元搭起来的私有模型,还在等风控委员会审批的时候,一个被裁掉的 quant,用一台租来的 GPU 和一个开源模型,已经跑完了 917 轮迭代。他没有在智商上碾压 Jane Street——他只是在速度上,彻彻底底地把他们甩开了。
K3 上线才一天,这已经是第一个真实暴富故事了。而且,这只是个开始。
中文

@mustafap0ly @MiSTkyGo Fee charges is killing PnL bro. Please cost down fee rates.!!
English

Polymarket has introduced withdrawal fees (0.07%). I’m not a fan of the change, but to be fair, it’s not an outrageous amount. They also claim you can still withdraw for free on Polygon (I’ll need to verify that).The real issue is how they handled it. They advertised free withdrawals for a long time, then quietly changed the policy without any announcement or warning. This is the 218,423,085th time they’ve pulled something like this. At this point, it doesn’t even feel like it’s about the money , it just feels like pure incompetence. It genuinely makes me sad. Polymarket used to be excellent, largely thanks to its huge selection of markets. But between the poor communication, bad management, and the constant bugs and errors popping up every week, it feels like they’re starting to lose their way.
English

@Johnboy13579246 @Mooowagiihhh 第二个进球应该回溯梅西踩脚犯规 进球应该取消。 不过英格兰是真的踢的差
中文

@Mooowagiihhh Hes a dirty cheating cnut like the rest of them. It’s a shame for him really because he could’ve been the best in the world but he’s just got by with officials and cheating. Sad really


English

@ada0211dada 写的非常专业了。 把预测市场基建底层逻辑和机器人脚本缺陷分析的非常透彻了。 想要在poly赚钱,首先要学会输。
中文

卧槽
我这是球神附体啊
英格兰进一个后
英格兰上了6个后卫
我押了1:1平和双方进球
阿根廷进球后
我止盈1:1平
又押了阿根廷晋级
阿根廷又 进一个 2:1 锁定胜局
动态押注
全部高赔率命中
在 @bagel_win 押的
bagel.win/?ref=BTC
邀请码 BTC


UNICORN⚡️🦄@UnicornBitcoin
英格兰 1:0 阿根廷 第55分钟 戈登抢点包超破门 英格兰先拔头筹 暂时领先
中文

Hyperliquid's HIP-4 is already dead.
Their Word Cup champion market made a pitiful 0.69% of Polymarket's volume
World Cup Winner (Polymarket) -> $4.26B volume
World Cup Winner (Kalshi) -> $1.2B volume
World Cup Winner (Hyperliquid) -> $29M volume ...
This is not what we expected from the "Polymarket Killer".
And this looks even worse when you consider that @polymarket and @kalshi both list thousands of markets while HIP-4 lists only 10 (ten).
Yet they can't even scratch 1% marketshare on the few markets that they have
This looks extremely bearish for their platform.
They may already have missed their critical take off opportunity with the World Cup.
They absolutely need to fix these 3 flaws if they want to survive :
1 - Prediction Market UI looks like trash (not retail friendly)
2 - Only validators can propose new markets
3 - No coherent liquidity incentive program
if they don't do it quickly it will be @hyperliquidx first failed product since their launch

English

我真他妈的是个疯子
押了1万刀法国赢
最后还赚了1400刀
在法国丢1球后直接止损
亏4500刀
余额押西班牙
当西班牙再进一球
我直接又冲了23000刀
押西班牙赢
最后赚了 1400刀
中途没功夫发推
全部都在 @bagel_win 有记录
bagel.win/?ref=BTC
盈利榜上那个就是我



中文




















