Re: [心得] 用FB Prophet預測美股未來行情
https://link.springer.com/chapter/10.1007/978-981-16-3067-5_12
這東西已經很多論文幫你回測過了
基本上是沒什麼用
https://www.backtrader.com/blog/2019-08-22-practical-backtesting-replication/practical-replication/
I've reproduced 130+ research papers about "predicting the stock market",
coded them from scratch and recorded the results. Here's what I've learnt
Because it has been deleted here a quick summary:
The strategies do not work
If the authors claim a given strategy stopped working due to alpha decay, the
tests were run against past data and it still did not work
Bottomline: It's all overfitting, p-hacking or a tiny alpha which doesn't
need to decay because commissions do already destroy the alpha.
就算是學術論文也充滿陷阱
我建議看論文時
沒有sharpe ratio的都直接跳過
不考慮交易成本的也直接跳過
只用個股OHLVC做預測的也直接跳過
你的交易成本夠低嗎?
你的系統速度夠快嗎?
你的資料品質夠高夠細膩嗎?
建議以上幾點都做到最好 再來玩量化交易
個人淺見
--
※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 36.227.155.174 (臺灣)
※ 文章網址: https://www.ptt.cc/bbs/Stock/M.1638033557.A.1DD.html
推
11/28 01:35,
2年前
, 1F
11/28 01:35, 1F
噓
11/28 01:49,
2年前
, 2F
11/28 01:49, 2F
推
11/28 01:56,
2年前
, 3F
11/28 01:56, 3F
推
11/28 02:03,
2年前
, 4F
11/28 02:03, 4F
→
11/28 02:04,
2年前
, 5F
11/28 02:04, 5F
→
11/28 02:04,
2年前
, 6F
11/28 02:04, 6F
→
11/28 02:04,
2年前
, 7F
11/28 02:04, 7F
→
11/28 02:05,
2年前
, 8F
11/28 02:05, 8F
→
11/28 02:05,
2年前
, 9F
11/28 02:05, 9F
推
11/28 02:24,
2年前
, 10F
11/28 02:24, 10F
推
11/28 02:26,
2年前
, 11F
11/28 02:26, 11F
推
11/28 08:57,
2年前
, 12F
11/28 08:57, 12F
討論串 (同標題文章)
完整討論串 (本文為第 2 之 2 篇):