Quant Beckman

3.8K posts

Quant Beckman banner
Quant Beckman

Quant Beckman

@quantbeckman

Quantitative Researcher | Algorithmic Trading Newsletter: https://t.co/dMLUqhqn8z Platform: https://t.co/2WMqDkiRbt

Katılım Temmuz 2019
0 Takip Edilen12.4K Takipçiler
Sabitlenmiş Tweet
Quant Beckman
Quant Beckman@quantbeckman·
Hey guys! I’ve just released a refined, compact edition of one of my newsletter series on the foundations of quantitative trading. It is a theoretical introduction designed for aspiring quants who are beginning to explore the field. The focus is on first principles, method, and the conceptual framework behind quantitative trading. Link in the comments. I hope you enjoy it.
Quant Beckman tweet media
English
11
47
506
26.5K
Quant Beckman
Quant Beckman@quantbeckman·
I saw a tweet today saying that people rarely talk about research on X. Honestly, they talk about research all the time. What they rarely talk about is what isn’t actually research. For example, plugging an AI into a couple of platforms and getting it to generate a great-looking backtest isn’t research. It’s overfitting. Then reality hits, and the systems being sold perform nothing like those backtests. I’m not saying there’s anything wrong with doing it, but let’s not call it research, you need something more to call it like that.
Quant Beckman tweet media
Quant Beckman@quantbeckman

You know, that situation where you're talking to a buddy and just can't see past the end of your nose. Has that ever happened to you? Getting into an absurd argument like that?

English
1
1
14
1.1K
DeflatedTrader
DeflatedTrader@deflatedtrader·
@quantbeckman The subtle part: the overfitting cost is paid even by the features you reject. Every candidate you tried and discarded was a test, and enough tests guarantee something looks great by luck. The backtest only shows the winner, never the hundred auditions behind it
English
1
0
0
16
Quant Beckman
Quant Beckman@quantbeckman·
📄[WITH CODE] Feature selection: Wrapper-based feature selection methods📄
Quant Beckman tweet mediaQuant Beckman tweet media
English
3
1
4
831
Quant Beckman
Quant Beckman@quantbeckman·
70.4% return and 3.15 Sharpe ratio? With only 3 stocks, 1 test year and no fees!? 😟
Quant Beckman tweet mediaQuant Beckman tweet media
English
2
7
128
10.1K
Quant Beckman
Quant Beckman@quantbeckman·
Oh, did I ask for that!? I doubt I asked for anything 😁 You are right that total option net supply is zero. That does not mean dealers are net zero. With participant-tagged buy/sell and opening/closing data, dealer inventory can be measured much more directly. By the way, remember to stop following a fraudster. Nothing could be more absurd than that. I’ll help you with it later, don’t worry.
English
1
0
0
19
gotthetwitts
gotthetwitts@nogoodkwant·
@quantbeckman You don’t have net positions. You understand that market net is 0? What you got is a party's _modeled_ gex ... and that model rests on unverifiable positions assumptions Reminder: asking your readers to believe you got alpha proof-less is the hallmark of grift.
English
1
0
0
10
Quant Beckman
Quant Beckman@quantbeckman·
Hey everyone, it's vacation time! So I thought, why not build an agent that makes decisions based on GEX and VEX data?
English
5
1
33
3.3K
Quant Beckman
Quant Beckman@quantbeckman·
Full trading pipeline in one line: Alpha discovery → nested validation → risk management → optimal execution
Quant Beckman tweet media
English
2
10
65
2.6K
Quant Beckman
Quant Beckman@quantbeckman·
@qibbler 100% glorified useless academia Pretty soon in my second brain 🤭
English
1
0
1
61
Sam Weil
Sam Weil@qibbler·
these researchers writing these papers, if they don’t trade their own book and just put together fancy stuff without real OOS testing/execution is it just glorified academia? Curious on your opinion. From your professional trading experience, where do you recommend looking at papers. Any specific database at all?
English
1
0
0
68
Quant Beckman
Quant Beckman@quantbeckman·
@nogoodkwant Yo don't infer it, you have the net positions and how imbalanced they are
English
1
0
0
35
gotthetwitts
gotthetwitts@nogoodkwant·
@quantbeckman so you're trying to tell pple you're inferring dealer positioning anywhere close to correctly?...
English
1
0
0
19
SpreadGreg
SpreadGreg@optiongreg·
@quantbeckman Good idea, but do the agents get good market data? As far as I know, the main point is the data. With CBOE, if you pay more they give it to you marked so you know exactly if it is MM or not!
English
1
0
0
93