AI &Crypto

Traditional to 100% remote analyst or investor job.

[AI &Crypto @willbeingbill]

Token兼具货币和资产的属性 #token

[AI &Crypto @willbeingbill]

量价分析，成交量在前，价格在后。

[AI &Crypto @willbeingbill]

K线要配合位置、交易量和基本面来综合分析 #K线

[AI &Crypto @willbeingbill]

成交量与多空分歧大小相关，分歧越大，成交量越高。

[AI &Crypto @willbeingbill]

连续下跌，一次比一次猛，是很危险的抄底机会

[AI &Crypto @willbeingbill]

MACD 柱状图：价格新高，但柱状图没有出现新高，连续2次背离，行情可能结束/开始。（附属指标）

[AI &Crypto @willbeingbill]

EMA 短周期5/10/12/20/26 长周期50/100/200

价格在EMA之上且EMA上升，市场上涨；
价格在EMA之下且EMA下降，市场下跌。

短周期EMA上穿长周期EMA，金叉，看涨；
短周期EMA下穿长周期EMA,死叉，看跌。

交易是一个概率游戏。

[AI &Crypto @willbeingbill]

市场80%时间都是震荡行情

[raghav @rargulati]

Martin is absolutely right to question this stuff. The problem with decentralized AI protocols are:

1. There is no cost/time effective way to do useful online-training on a highly distributed architecture of commodity hardware. This would require a big breakthrough that I’m not aware of yet. It’s why FANG spends more money than all the liquidity in crypto to acquire expensive hardware, network it, maintain data centers, etc.

2. Inference on commodity hardware sounds like a great use case, but the hardware and software side is evolving so rapidly that commodity decentralized underperforms for most critical use cases. See latest openai latency and growth in Groq.

3. On its face, doing inference off of properly routed requests to a gpu cluster closely co-located to the request sounds like a good idea, and using a decentralized cryptocurrency to crush the cost of capital to compete against aws by incentivizing hobbyists to participate sounds smart. But given so many providers, and fractured liquidity in the GPU spot market, no one has aggregated serious supply to offer to anyone running a real business.

4. The software routing algos need to be really, really good. Commodity hardware across consumer operators is very spotty operationally. Forget breaking out of networks and congestion control, if someone decided to play a game or use anything that leverages webgl, you might have an operator that blips out. Unpredictable supply side adds operational headaches and lack of certainty to demand side requesters.

Tough problems that will take a very very long time to solve. The bids are all memes.

[AI &Crypto @willbeingbill]

Crypto还属于高风险资产，美国国债利率高时，Crypto较难拥有较大汇报。 #无风险利率

[AI &Crypto @willbeingbill]

Crypto类比股票的PE，可以用Price/EPS计算得到

[AI &Crypto @willbeingbill]

Crypto类比股票的EPS（每股收益）可以用fee/issued token计算。#EPS

[AI &Crypto @willbeingbill]

去中心化不可能三角：性能、去中心化和安全，不可三者都要，但可以自由组合其中的两角。#去中心化

[AI &Crypto @willbeingbill]

比特币减半后关机币价计算：矿机算力110Thash/s,功耗3.3kw,电费0.45元/度，全网实际算力500Ehash/s,区块奖励3.125/区块，出块时间10min，每天挖出BTC450枚，美元兑人民币汇率7.25，按1天周期计算，则关机币价=3.3*0.45*24/[110/（500*1000000）*450*7.25]=$49655 #关机币价

[AI &Crypto @willbeingbill]

生态发展和币不一定有强关系，生态币在DeFi是资产，NFT和GameFi可能只是交易费，比如Flow。#Flow

[AI &Crypto @willbeingbill]

Toke释放时间表

[AI &Crypto @willbeingbill]

解析https://t.co/1QSt2flEOY. 摩尔定律的终结和新机器学习应用程序计算需求的飙升。因此，应用程序需求和单节点性能之间的差距迅速扩大，让我们别无选择，只能分布式这些应用程序。

[AI &Crypto @willbeingbill]

从运营数据上看https://t.co/hh9qI9qQjx的营收，GPU/CPU 以及Twitter关注量都超过了后两者，有理由期待https://t.co/hh9qI9qQjx的FDV将超过后两者。初始流动量约为2亿枚，按Market Cap 计算，io的单价约为$ 35/2=$17.5 ;按FDV算，io的单价约$42.26/8=$5.2。仅供参考！

[AI &Crypto @willbeingbill]

RWA的土壤在于美债的高额收益率。#RWA