聊天记录
共 18 条，显示第 1-18 条
It might work. But have a simplistic baseline, you will realise it is much better.
Let's take a simple example:
Y is the stock price. x is just time passed. Assume Y=sin(x)+e at all times, e is the noise. You know this for sure. Then nothing can be better than sin(x) function. Fitting a neural net will lead to no improvement at all, with overfitting as a bonus.
In a market causal relationships may be few, with noise you will almost always overfit to the noise, which cannot be controlled
It might work. But have a simplistic baseline, you will realise it is much better.
Let's take a simple example:
Y is the stock price. x is just time passed. Assume Y=sin(x)+e at all times, e is the noise. You know this for sure. Then nothing can be better than sin(x) function. Fitting a neural net will lead to no improvement at all, with overfitting as a bonus.
In a market causal relationships may be few, with noise you will almost always overfit to the noise, which cannot be controlled
Don't know what, just curious
Don't know what, just curious
Very suboptimal.
ML is used for approximating functions. Let's say you knew stock price is sin(x), where x is something. Then using ML to fit this just introduces complexity without any upside. The problem is determinism, the probability of outcome is not what is the focus usually.
Very suboptimal.
ML is used for approximating functions. Let's say you knew stock price is sin(x), where x is something. Then using ML to fit this just introduces complexity without any upside. The problem is determinism, the probability of outcome is not what is the focus usually.
You should teach your model to draw candlestick patterns. It will have maximal sharpe trust
You should teach your model to draw candlestick patterns. It will have maximal sharpe trust
Just pack it up lil bro
Just pack it up lil bro
Drawing candlestick patterns
Drawing candlestick patterns
At least quants can justify the economic benefit, here it's just nothing, might as well play CS2 instead of tokenmaxxing😭
At least quants can justify the economic benefit, here it's just nothing, might as well play CS2 instead of tokenmaxxing😭
What even is the point of this? One bot steals off other, there is minimal retail flow😭
What even is the point of this? One bot steals off other, there is minimal retail flow😭
Yeah, started at 1500 UTC
Yeah, started at 1500 UTC
Chainlink is misbehaving right? It's throwing 429s
Chainlink is misbehaving right? It's throwing 429s
Yeah
Yeah
Rest rtt
Rest rtt
London vm to polymarket latency. I am running on dublin and i wanted to see how much of a difference colo makes
London vm to polymarket latency. I am running on dublin and i wanted to see how much of a difference colo makes
This is from london to polymarket right?
This is from london to polymarket right?
Like general p50/p90 on whatever data
Like general p50/p90 on whatever data
what about REST RTT because ws is kinda misrepresentative for cancel
what about REST RTT because ws is kinda misrepresentative for cancel
How much is the difference between dublin and colo?
How much is the difference between dublin and colo?
No. EOA address. (Wanted to fetch taker tier not maker rebates)
No. EOA address. (Wanted to fetch taker tier not maker rebates)
Polymarket