Quant Beckman @quantbeckman

Quantitative Researcher & Dev. | Financial Data Scientist | Machine Learning Engineer | Mathematical Research | Algorithmic Trading Systems quantbeckman.com Fresh takes on Quant Research Joined July 2019

Tweets

4K
Followers

12K
Following

0
Likes

844

Quant Beckman @quantbeckman

2 hours ago

$VERU 144% It's a hairy candle, it looks like dynamite... so check in more detail here

0 0 1 408 0

View Details

Quant Beckman @quantbeckman

7 hours ago

quantbeckman.com/p/quant-lectur…

0 0 2 283 0

View Details

Quant Beckman @quantbeckman

7 hours ago

📘[QUANT LECTURE] Quality standards of a strong hypothesis📘 👇👇👇

3 2 18 655 13

View Details

Quant Beckman @quantbeckman

14 hours ago

🟥For this one and more papers with code: quantbeckman.com

0 2 3 468 2

View Details

Quant Beckman @quantbeckman

14 hours ago

Risk-off strategies often look good because they avoid a few large drawdowns (2008, 2020, 2022)

3 7 63 4K 55

View Details

Quant Beckman @quantbeckman

23 hours ago

Trading has 2 complementary legs and for some reason a lot of people focus only on one of them

1 2 12 511 8

View Details

Quant Beckman @quantbeckman

a day ago

Vol-of-vol can move before spot volatility fully reprices. When VVIX jumps faster than VIX while front VIX futures lag, the options market may be pricing a convexity shock before the futures curve catches up. Buy defined-risk VIX call spreads to express that convexity repricing without unlimited premium exposure. Exit when VVIX mean-reverts while VIX remains elevated or when the futures catch-up has already occurred.

1 5 35 2K 46

View Details

Quant Beckman @quantbeckman

2 days ago

🟥For this one and more papers with code: quantbeckman.com

0 0 4 623 3

View Details

Quant Beckman @quantbeckman

2 days ago

The graph construction is depends on Pearson correlations estimated from 125-day windows, and more than 5 parameters. Small changes in these design choices reorder centrality rankings and therefore change the selected stocks. So the results are pretty unstable.

3 2 26 1K 21

View Details

Quant Beckman @quantbeckman

2 days ago

One of the most used phrases in quantitative finance: "correlation ≠ causation"

2 1 12 571 5

View Details

Quant Beckman @quantbeckman

2 days ago

Equity-index options often price weekend uncertainty through implied variance, while realized Friday-close to Monday-open movement may be smaller in quiet regimes. The trade sells a small delta-hedged index straddle near Friday close when implied weekend variance is rich. The position is bought back Monday open if the realized gap stays inside a preset band. The edge is weekend theta capture, but the risk is jump exposure while markets are closed.

2 2 22 1K 23

View Details

Quant Beckman @quantbeckman

3 days ago

🟥For this one and more papers with code: quantbeckman.com

0 1 2 833 2

View Details

Quant Beckman @quantbeckman

3 days ago

The authors show that performance is highly sensitive to noise, and bottleneck size. Once Gaussian noise rises beyond around 5%, or the bottleneck becomes too large, the Information Ratio declines.

3 2 23 2K 24

View Details

Quant Beckman @quantbeckman

3 days ago

These are basically the pitfalls I look for every time I read a paper, and in this same order.

1 5 40 2K 32

View Details

Quant Beckman @quantbeckman

3 days ago

@algotrading_ios Hey man! I should develop this idea more but this are mere notes. If I have to choose, the simplest is: median Hajahahahah

0 0 1 14 0

View Details

Quant Beckman @quantbeckman

5 days ago

Some spreads revert fast, some revert slowly, some jump, and some trend after breaking. Embed each spread into a latent space using features or sequence encoders that capture its behavior over time. Cluster the latent points into archetypes, then assign trading rules suited to each cluster.

3 6 43 2K 50

View Details

Quant Beckman @quantbeckman

3 days ago

Some factors can unwind together when liquidity disappears, especially in small caps. Build a portfolio objective that rewards expected edge but penalizes exposure to known crowded factors.

2 8 34 2K 37

View Details

Quant Beckman @quantbeckman

4 days ago

🟥For this one and more papers with code: quantbeckman.com

0 1 4 507 5

View Details

Quant Beckman @quantbeckman

4 days ago

The supervised stage trains the DDPG actor to imitate Markowitz allocations. The problem is that Markowitz weights are extremely sensitive to expected returns and covariance estimates. Atiny estimation error can produce a large change in weights

4 6 42 2K 39

View Details

Quant Beckman @quantbeckman

4 days ago

Research --> Model --> Execution

0 5 29 1K 20

View Details

Quant Beckman @quantbeckman

4 days ago

Each state represents a market environment where mean-reversion sleeves may behave differently. Instead of running MR continuously, estimate which latent state is active today. Only allocate to MR sleeves in states where historical survival and drawdown are acceptable.