Oh there's also chat room..
這邊有聊天室: https://t.me/+kciefBpHhTk2MTM1

群組人數超過150人就來發如何提前知道策略失效以及分層過濾的方法～

Btw 下週一二晚上有常態量化講座！有興趣的可以dm我一下！ 目前有3人想聽～人太少就不開了

[ BCAGENT TG 群友限定文章 ] :

以信號數量作為參考，市場中不同策略屬性也會有不同的數據差異，以相同長度做為資料源的 fetching period, 在不加上濾網以及經過優化的純策略下:

順勢 ( 動量 ) 策略的信號數量會比反轉逆勢( Reversal ) 策略開單數來的高，原因是市場中不穩定的時間相對來的高，因此順勢策略通常會被來回磨耗；
再者真的動量出方向，反轉逆勢策略通常進入超買/賣區，或是95%信心區間外，在經過大趨勢時會鈍化很嚴重 ( 如果完全沒有做 Normolization 或是Standardization )，因此order數也不會提高。

這樣一來會發現: 順勢策略的勝率偏低，然後逆勢的單數也偏低，要如何去做調整?

順勢 ( 動量 ) : 勝率低的狀況下，試著減少不必要的磨耗，也就是 order amount, 採取減法系統策略。
逆勢 ( 反轉 ) : 勝率要求高的狀況，並且由於常規逆勢的策略損益比較低，應該要試著提升order amount.
(ps: 順勢降低 order amount 時需要避免underfitting； 逆勢(反轉) 再增加單數應該以同質性為主)

光這兩點做到，就可以幫你優化策略一大截了。

方法都給你了，還不快點試試看，還是不會的話 dm 我 "快速優化" ><

以前有跟我上過課的就知道濾網跟分層觸發區域可以做，晚點學習群內分享。
-
大家也可以幫我分享
IG: https://www.instagram.com/bca_daniel_/

I came to Maybank today at noon to chat with a Vietnamese market analyst friend.
If you are interested in going overseas, I will work with him on the transfer of funds. T+2.5 can earn 5% of total assets in one day, which is enough...🧎‍♂️ 🙇‍♂️🙇🏻

今天中午來maybank 跟越南市場分析師朋友聊天，如果有興趣海外的，之後也會跟他一起配合在傳金的部分，T+2.5 能夠現貨一天總資產 5%收益也夠了..🧎‍♂️🙇‍♂️🙇🏻

I mentioned the article on dynamic adjustment of position size before. There is also a paper that does a good job. I will put it below:

https://bcadaniel.medium.com/linear-regression-used-in-quant-portfolio-bcagent-backtesting-machine-tutorial-a55ad4b74551

Related Papers - Hsieh, Chung-Han. “Generalization of affine feedback stock trading results to include stop-loss orders.” Automatica 136 (2022): 110051:

Use different MDD sizes to adjust the feedback size parameter: gamma (including MDD ratio)
https://arxiv.org/pdf/2004.12848v1
And: On Data-Driven Drawdown Control with Restart Mechanism in Trading
https://pdf.sciencedirectassets.com/313346/1-s2.0-S2405896323X00093/1-s2.0-S2405896323005700/main.pdf?X-Amz-Security-Token=IQoJb3JpZ2luX2VjECMaCXVzLWVhc3QtMSJGMEQCIBUhTGQX LePsO%2F3XCf9WSlKuIWPOdM4LJaNF%2ByKF1aU6AiAvxvC2sewAGcXCZNvZNaFfe9B4IdeG2 OpUTuuqihrA4CqzBQgrEAUaDDA1OTAwMzU0Njg2NSIMdZqb4Xhv0Rp4mAcXKpAFcn2jF6%2FS6 TSLyRsIweOclvjuvzYBp%2FO7Q81TCK7dRpj9fezTuVGBbs5afOqk72A% 2FiH%2FnGnJZ1%2FhFZL7RBbieUyxR8gtaLPVA98W%2B8Jza15D0snqhjIRV15kAaW8fZ7%2BbWCxPZgT9oyPGDr%2FHPOX9Scqiuv7Azbg0D%2BE5lrkGo85GEpmI%2FPhq1dgdKsrRhSl gIbMV7QntMQIzvPvfMkBSZzzE%2BJGPZCnyGIH6j2tGybvMBF2vlsAaCf72zHpbEXCZNa8MoPiqrdExCtEi4Ex2HCGO0b8lKAxi9JDt%2BC7C2nOWEVfwc%2FZKyDAFCnLVd5O%2FYe2fgCzapChIQFX FO4hPcNdG5gS1cr0BLcWpd3l4tDnOZZAxEX6IxNbJyh4ZzbuEl73lVUb1n0uyAn8M%2B9 zDFmElUiBhsnBwxdxzFJepiVWf%2BMIg5Oqr2FS2W3wgRCL9oGnxDI8R1K57l9A5iAKDEH 2c6K2r6KKkDz7r2jUxx5ncSIRrHPHRNWV1nRWFbJ4wrTgXwsQ2Xr8UUBYiQ7sR1xjtKdq 5jP5TEkZqefl22I39hXVsmm%2B0HfCfZfS7BPGxcrL7aRvfAnzHMZdXYWkrEzlUOqKB2y M0VJ6iVneOQBuWxcv5tWSWt67xHw0UZxDdUSPzv15zCUXLOX8CigvlLd%2Ffcccl3UnZktQbFZ 6Y%2BCkVOl7%2B2iKrqwiDb3GGedP7qdYOmgPhhH9gt%2BhBLxeEjogeupIwAi461BspmndIFKP mErfQcILyVwpphxUFTKCXhHrc6hnhFpPMz6WHjVQ%2FZQggikiQ%2FkPq%2BIhuWWrjehGf0lro cVK4AZH%2F0AV%2FQm7AGGC4EKd%2FjSu7ghWJR6unZWFT68NSm8Ewjhnec2U5hoUwkZCHtgY6sg E% 2FmVVs79YeysFeVXJhrSFZTVUXdR4Djb5BKK%2Ful5PRgT4fVuLrQ7SZmZaBU%2Bv1J958orJI9ZYxpDe2IngNpf8ri9tjqjqZ26wDd0cGZPC9CYuyxQqSNbWn3bJBJZ5D0JD6jZqwa Dg0MJYOGwJ10LbDLeudC4HKRJj4Zua6%2BQtb5bSblVqUeQikaLzFlHl27%2BQRE94%2B2zNHNbzFTHfR4%2Fl8oiJIH8CfMHCdrUtDeHfI%2B%2Bg1&X-Amz-Algorithm=AWS4-HMAC-SHA256 &X-Amz-Date=20240818T110550Z&X-Amz-SignedHeaders=host&X-Amz-Expires=300&X-Amz-Credential= Asiaq3phCVTYQ55DRZXDC%2F20240818%2fus-East-2FAWS4_Request & X-AMZ-SIGNATUR 1885384A33B121EA0 & Hash = 20066620930FDF133788E3044444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444490F61C's. 0F6AF2C76D8DFD086A07176Afe76C76C2C61 & PII = S2405896323005700 & Tid = SPDF-86302ECE-6597-4A00-A5C4-871863333ce & Sid = 936FAFA6230A666 94E391BC5C5C285A2731F01cegxRQA & Type = Client & Tsoh = D3D3LNNJAWVUY2VKAXJLY3QUY29T & UA = 14125B065754560357 & & cc = tw

As shared on IG, I use linear regression to measure the offset, and directly deal with the performance part.

It is similar to the MDD ratio, but I use a fixed constant in the example (such as 80% of the original leverage adjustment), of course it can Add layering for more detailed dynamic leverage management, that is, region area
-
在IG分享的，我用linear regression去衡量偏移量與否，就績效部分直接處理，跟MDD比率有異曲同工，但是我在範例中用的是固定常數(例如80%原槓桿調整)，當然可以加入分層做更細緻動態槓桿管理，也就是regime area

btw, 點開 channel有討論區，記得加入哈，可以講話分享經驗，不過不要人身攻擊
https://t.me/+kciefBpHhTk2MTM1

Basically, in quantification, the concept of "investment group" is used to make strategies into bucks, so the following issues will be focused on:

1. What are the timing and conditions for strategy replacement?
2. Strategic risk control benchmarks
3. The proportion of funds allocated to different strategies
4. Response to investment team failure
5. Funds Rebalanced
6. Correlation test/ensure that the scatter points have a certain vector distance difference
7. Coping with strategically vulnerable scenarios
8. Wait...

Mathematical model calculations, multiple reference papers, multiple exchanges and discussions on engineering statistics, etc.

If you just feel that you have the most powerful high-yield strategy in the universe and the risk part is barely covered, maybe you still have a long way to go.
-
基本上在量化當中會用"投組"的概念去把策略做bucks，所以後面著重的問題是:
1. 策略替換時機與條件?
2. 策略風控基準
3. 資金分配不同策略比例
4. 投組失效應對
5. 資金 Rebalanced
6. 相關性檢驗/確保散點有一定的向量距離差
7. 策略弱勢場景應對
8. 等等...

總之市場上各種東西都有，給大家一些背景知識點即可。
對我而言，還有好長一段路要走，每天繼續加油，繼續研究。

BCAGENT TG group friends give the first preview!

In the process of making a new post, this is the performance (backtest) parameter I designed myself.
I will share it with you this week, as well as the design logic. In the strategy, it is not a reference parameter on the market (such as SR, ROI...) and can only be used for reference.

sometimes you need to help yourself understand your own strategy better, so you can design your own parameters.

This week’s explanation: IDR (XX ratio)
-
BCAGENT TG 群友搶先預告！
新貼文制作中，這個是我自己設計的績效（回測）參數，本週會分享給大家，以及設計邏輯，在策略當中並非市面上的參考參數（例如SR, ROI..)只能參考，有些時候需要幫助自己更了解自己的策略，就可以自己設計自己的參數。
本週講解： IDR (XX比率）

Maybank VN account opened successfully

Maybank VN開戶成功
炒股囉🥹 沒 這邊外匯管制很嚴 一不小心不用領了

Don’t miss your DOGS token FREE!

Claim it -> transfer to Binance or OKX -> waiting the buy and sell (I saw some already trade on bitget)

不要錯過DOGS! 有TG 都可以免費領，領完轉去Okx或是幣安，要上幣安了！

https://t.me/dogshouse_bot/join?startapp=DrpbJEkJTHKBmJdJwtRAvA

OKX signal for quant trading is really good 👍